AN IDEA

This idea, which first appeared in this Web site in late April 2006, was later expanded. We state the idea in phases because it is complicated. There is a long introduction to the idea.

1. Introduction: Ipedmis^SM (Investigation, Patrol, Emergency Dispatch Management Information System)
   1. Ipedmis^SM
      The introduction discusses GC (geographical coordinates) records, 911 emergency dispatch, police patrol, and related matters. is an
   2. GC (geographical coordinates) records
      1. Plain

         Sometimes, a police detective investigates a crime. He knows an address that is important in the case. He wants to know every incident of his police department (for example, every address) both within one thousand feet of the address and within a fifty-day period he he specifies. He wants to know if a related event was done there in the specified period of time. XXXXXXX XXXXXX (within 1000 feet of where the corpse was found), a 911 (emergency) call came from there, a traffic ticket was given there, an arrest was there, etc. This is not always an easy thing to find out. One thousand feet away could be on a different street, in a different 5-digit zip code, in a different city, in a different county, maybe even a different state or country. He doesn't want to look at records of every incident of his department, just incidents within 1000 feet (or 200 feet, or 1 mile, or any other distance) of where the corpse was found.

         There are many GC (geographical coordinate) systems. The system which is most famous and widely used is latlon. Latlon is our term, not a standard term, for the latitude and longitude system. The UTM (Universal Transverse Mercator) system is based on latlon. Also, there are GC systems that have been invented for many countries; for example, for America. The American system is abbreviated SPS and SPCS (State Plane Coordinate System). SPS generally is more accurate than latlon for counties, cities, and other small places. We are under the impression that NATO and the Warsaw Pact developed their own GC systems, perhaps because those systems are better than latlon for some military work. A GC-user should use the GC systems best suited to his purposes.

         A police department could routinely calculate the GC for many incidents. For example, there are computer programs that automatically convert street addresses to GC (for example, giving the latlon for each address). There are computer programs that convert 9-digit zip codes to GC. There are small, convenient, GPS devices that one can hold in one's hand that show GC (usually as latlon) and altitude.

         We once read about a rural police department in America that uses GPS devices. If rural police in that department find a corpse deep in the woods, writing just the address in the police report isn't always helpful. The address could be a farmhouse a mile away. The officer writes in the police report the GC where the corpse was found. The rural officer uses a GPS device to tell him the GC where he's standing.

         The conversion of a non-GC system for describing where a place is (for example: a street address, or a county real estate tax assessor's parcel number) to a GC system (for example, 12.12345 latitude, 23.12345 longitude) is called geocoding. Converting in the opposite direction (for example, from latlon to street address) is called reverse geocoding.

         When 911 telephone calls come into a 911 dispatch center, the dispatchers usually know the addresses that the calls come from. The phone company has that information for each call and supplies the information to the 911 dispatcher. From the street address (especially if the phone company supplies a 9-digit zip code as part of the address), a computer can calculate the GC that the caller is at while he makes the call. A 911 call involves at least two addresses: the address where the call comes from, and the address where the police are requested to go. Usually, they are the same.

         Finding the GC for some kinds of incident is much faster and cheaper than for others.

         If the police department has a computer which stores the GC and date for many incidents, then the detective (remember the detective in the homicide example that begins this discussion) can easily find all incidents within a certain distance (of where the corpse was found) within a certain range of time (for example: the 200 days ending the day that the corpse was found). It is easy for a properly programmed computer, if given an address or GC (for example, the latlon of where a corpse was found), to find every incident within 1000 feet (for example) between any two dates. Off-the-shelf database software makes this easy to do. The computer gives the detective a list of incidents that meet his specifications. The list can be sorted different ways according to the detective's wish: by closeness in time, closeness in distance, category of incident (for example, places of arrest first), etc.

      2. Fancy

         In the example above, there is a circle. The center is where the corpse was found and the radius is 1000 feet. The GC search does not have to be a circle or only one circle. Immense variety is possible.

         Imagine a suspect who is believed to walk from his home to a burglary site, do a jewelry burglary, leave the premises with the jewelry loot, walk to a fence, then sell the jewelry to the fence. The suspect is believed to walk twice: from his home to the jewelry burglary site, then to the fence. Probably each walk is under 4000 feet. The detective knows where the suspect lives and where the suspect sells to the fence. The detective tells the computer about two circles. The first circle is centered where the suspect lives. The second circle is centered where the suspect sells to the fence. Each circle has a radius of 4000 feet. The detective wants a list of police incidents (for example, jewelry burglaries) that are in both circles (not incidents that are in one circle only). In other words, the incidents must be both within 4000 feet of the suspect's home and within 4000 feet of the place where the suspect sells to the fence. (To use set jargon, the detective wants the intersection of two sets.) This list of incidents would be easy for a computer to supply.

         The search through computer records can be more complicated. If the detective thinks that the suspect doesn't burglarize within 500 feet of his home, the suspect can tell the computer about a third circle (with a center at the suspect's home, and a radius of 500 feet). The detective can have that third circle excluded. The computer would then search for all incidents in both of the first two circles (the two circles' intersection, not their union) EXCEPT incidents in the third circle.

         Computers are not limited to circles. Computers are good at geometry. Above are a few, simple examples of using geometry and set theory to search geocoded police records. This Web page is not suitable for a thorough discussion of this important subject.

      3. So much for the GC phase of this discussion.

   3. Finding patrol cars

      Most police departments have cars which patrol. It is easy to know where all patrol cars are at all times. Furthermore, if one specifies a place (1234 Liberty Street, for example), one should be able to immediately know the patrol car nearest to that place.

      There are two systems: location broadcasting and triangulation.

      In the location broadcasting system, each patrol car has a location finding device which often calculates where the patrol car is (using a GC system such as latlon). The device also calculates altitude of the patrol car although we don't know how useful that would be. The location finding device gets information from broadcasts by an American satellite system called GPS. As a result, the location finding devices are called GPS devices. Today (19 February 2007), there is only one location finding satellite system that is fully functional (the American one, GPS, which means Global Positioning System), and during or soon aftre 2010 there will be one satellite complementing GPS (Japan's Michibiki, which will much improve GPS accuracy and reduce GPS blind spots). Galileo, a Europe-controlled satellite system, is now gradually being developed. We guess that, when Galileo eventually is fully functional, cars that receive America's GPS broadcasts will also receive Europe's Galileo broadcasts. We guess that a small computer in the location finding device will consider both systems' signals and automatically use the system that is more accurate at the time. GPS broadcasts are free to the user. Galileo eventually will also provide broadcasts free to the user. The accuracy of the free services, America's and Europe's, may be good enough for the idea discussed on this page. The Galileo satellite system will, in addition to providing free broadcasts, provide an encrypted CS (Commercial Service) broadcast for a fee. CS will be much more accurate than the free broadcasts provided by GPS and Galileo. Furthermore, a perfectionistic police department could also use ground stations to broadcast information to cars' location finding devices to allow them to calculate their GC. If a location finding device simultaneously used broadcasts by CS and ground stations, the device could calculate its GC to within less than four inches. Galileo will have special broadcasts for police and similar government agencies. We don't know if those broadcasts will go to America or have information useful to Americana police departments. When fully functional, Galileo may differ from what is now planned. Galileo is developing behind schedule. The U.S. government tried to stop Europe from developing Galileo. We don't know if the U.S. government will try to discourage Americans from using Galileo.

      In any event, a local police department can easily have each patrol car frequently broadcast its GC (calculating GC by using information broadcast to the car by GPS, Galileo, ground stations, or some combination of the three). The police department can receive the broadcasts and record the GC information in a computer. Thus, a police department can know where every patrol car was very recently (as recently as the most recent broadcast, and broadcasts can be frequent). So much for the location broadcasting system.

      The other system (for knowing where all patrol cars are) is triangulation. Each patrol car frequently broadcasts a signal identifying itself but not its location. The police department has towers which receive the broadcasts. The towers are connected to a computer which uses triangulation to calculate where each patrol car is.

      In summary, there are two systems for allowing a police department to automatically know where its patrol vehicles are. In both systems, each patrol car has a device which frequently broadcasts a message. In the location broadcasting system, each patrol car's device figures out where it is (using GPS, for example), then broadcasts its location to the police department. In the triangulation system, each patrol car's device broadcasts an identifying signal to the police department, and the department then learns where the car is by figuring out where the signal came from.

      For many years, it has been common for management of some organizations to automatically know where vehicles are. For example, some trucking companies know where all of their trucks are while the trucks are on the road. For another example, the San Jose city (Santa Clara county, California) police department knows where its patrol cars are, we were told.
      

   4. faster 911 response
      1. How

         One way to improve 911 dispatching is to get a patrol car to the emergency destination faster. Because of everything written above, it is easy for a 911 computer to automatically know which patrol car is closest to the emergency address. In many cases, when a 911 call comes in, the 911 dispatcher should be able to automatically see (on his video monitor) the address and latlon that the call is coming from. Usually, this is the same place that the patrol car should go. The computer can easily, automatically find the patrol car nearest there, the patrol car second nearest there, the third nearest, etc. Each patrol car can have a mobile phone with a unique telephone number. The mobile phone system can be encrypted. The dispatcher can press two computer keyboard buttons (for example, control key and F1 function key) to automatically be connected by phone to the nearest patrol car. Then, the dispatcher can describe over the phone the 911 case and ask the officer if he wants it. If he wants it, he goes, and the dispatcher tells the caller that an officer is going there already. If he doesn't want that case, the dispatcher presses control and F2 to automagically be connected to the second nearest patrol car. If the dispatcher wants, he can connect the caller with the officer so that they can talk directly with each other by phone. The officer could talk to the caller as he drives to the 911 emergency destination ("I'm a police officer. I'm driving to you right now. I'll be there soon.").

         In some cases, the 911 computer system may not automatically provide the dispatcher with the destination address. Then, the dispatcher may have to type it into the computer's geocode program to find the GC (for example, latlon) of the destination. This is easy. The software for this is user-friendly. Regardless of how the destination GC is discovered (automatically or by a dispatcher using geocode software), the computer will automatically discover the nearest patrol cars.

         By offering the 911 case to the nearest patrol car, response time can be reduced. Furthermore, the dispatcher can see the destination and patrol car on a map on his video monitor. The dispatcher can, if he wants to, watch the patrol car (shown by a green dot, for example, on a map on the video monitor) approach the destination (shown by a red dot, for example). The dispatcher, who can watch everything on a map on the video monitor, can give driving advice to the officer (for example, "Destination 1234 Liberty Street is north of the zoo, near a corner, several blocks east of County Hospital. Maybe you should take Jones Street south to Liberty, then turn right."). The dispatcher can watch the patrol car dot (the dot on his map which represents the patrol car) approach the destination dot. If the patrol car has an accident and therefore suddenly stops moving, the patrol car dot on the dispatcher's map will suddenly stop moving. The dispatcher can call the patrol car to ask what happened and if another patrol car should be dispatched instead.

      2. Management information

         If a patrol supervisor in a police building wants to, he can look at a map (on a computer's video monitor) which shows all patrol cars of his department. He can see the same map that the dispatcher sees on the dispatcher's video monitor. Imagine that the supervisor is in charge of patrol in precinct 12 and he instructs the map to display only precinct 12. He sees no patrol cars east of Liberty Street, which displeases him. He watches for a while and it stays that way, so he calls one of the patrol cars and asks the driver, for the rest of the driver's shift, to stay east of Liberty unless he's handling an emergency. The supervisor then sees (on the map on his computer video monitor) one patrol car dot move east of Liberty and then start cruising east of Liberty. The patrol supervisor notices that often there is no car on Liberty. The supervisor thinks that there should be more patrol on Liberty Street. He calls another patrol car west of Liberty and tells the driver that, for the rest of the driver's shift, he should (unless he's responding to an emergency) patrol within one block of Liberty and often on Liberty. The patrol supervisor sees (on the map on his video monitor) a patrol car dot start patrolling within one block of Liberty, often on Liberty. This map idea (namely, displaying all patrol cars' positions on a map on a video monitor) can help patrol supervisors.

         At any one time, a patrol car either is handling a 911 dispatch (in which case the car is shown by a white dot on the video monitor map) or the car isn't (in which case it is shown by a green dot). If every patrol car dot on a precinct's map is white, then every patrol car in that precinct at that time is handling a 911 dispatch. The computer system can display a table showing, for the entire city (and for each precinct), the percent of patrol cars which are white (handling a 911 dispatch) and green (cruising; in other words, patrolling). Many kinds of similar statistics can easily, automatically be displayed. If a precinct often has a low number of green dots (compared to that precinct's crime or population, for example), that precinct may need more patrol cars assigned to it.
         

      3. Batch dispatch

         By the way, there is no need for the dispatcher to call only one car at a time. For example, imagine that an officer is the one with an emergency. The dispatcher can automatically call ("Officer down at 1234 Liberty Street, north of the zoo.") to the nearest 10 cars, 50 cars, every car within 3 miles, etc.

      4. Help finding the street number
         1. Help for police

            Consider a patrol car that is on Liberty Street but cannot find street number 1234. Maybe the number cannot be seen by an officer in a car in the street. The officer would like to know what the house (or other building) looks like; for example: how many floors, what kind of roof (for example: shake, shingle, tile), whether the house is on level land or a slope, what the exterior of the house is made of (for example: aluminum siding, shingles, stucco, brick), and on-premises parking (for example: garage, carport). Apartment buildings and office buildings often have a building name on a sign in front (for hypothetical example, "Regal Hall" and "Commerce Tower"). The officer would like to know the name of the building. It should be easy for him to quickly get the information. He should be able to dial a telephone number of the HIS (house information service) computer of the 911 dispatch office. The HIS computer should know that the call is coming from his car and will know his destination (because the dispatcher will have typed information, connecting that patrol car to that destination, into the 911 dispatch computer). An HIS voice synthesizer should answer the phone by telling the officer about the house; for example, "This is HIS info about 1234 Liberty Street. 2 floors, brick exterior, sloping land, carport, tile roof, detached [in other words, not connected to another building], one family residence." In many jurisdictions, he HIS computer should be able to get the house information from the local real estate assessor's computer. Most assessors' computer systems have that information and much more about every building (warehouse, office buildings, factory building, one-family house, etc.) in their jurisdiction. The HIS computer would provide information only to help the officer (who is in a car on a street) recognize the house. For example, the HIS computer would not provide information, for a one-family house, about whether there is a finished basement or a swimming pool. To use computer jargon, the HIS computer would only say fields (for example, a roof field) useful to a police officer trying to recognize the right building.

            In addition to obvious information about what buildings look like, there are clues. For example, some assessor records will say if a house has an English basement. English basements usually are in houses on a slope, not in houses on level land. Thus, an English basement is a clue that the house is on land that slopes (although the slope is not always visible from the street). This Web page is not the place for an extended discussion of all the clues (about how to recognize a house while sitting in a moving patrol car) that are in assessor records.

            Assessor computers are not the only source of information about how to recognize a house. When a seller wants to sell real estate, he may go to a real estate broker. The broker prepares a detailed description of the house which is entered into a computer system maintained by a local association of brokers (often called a listing service). Then, a buyer goes to a broker and tells the broker what he wants. The buyer's broker looks through the computer system's files for real estate matching the buyer's interest. Much real estate (one-family houses, warehouses, office buildings, etc.) is listed in the ACS (real estate broker association's computer service). Some information is the same that the assessor has. Sometimes, the ACS has information that the assessor does not have, that an officer could use to recognize the house (for example: "huge, impressive lawn", "quaint, picturesque cottage", "circular driveway leads up to the house", "ultramodern ranch house"). The HIS computer should not have all ACS fields, only those fields that are useful to a police officer trying to find a house while he sits in a patrol car in the street. For example, the HIS computer should not tell an officer, and therefore does not need to know, what elementary school children in that house are eligible to attend, or whether there is a tennis court in the back yard. ACS records only have what's listed for sale, not all real estate. ACS archives describe what was listed for sale, which may still be the same. For example, it a ranch house at 1234 Liberty Street was listed for sale five years ago, it's probably still there. The older the information gets, the more likely it is to be wrong. If a ranch house was there 30 years ago, maybe it's still there, maybe not. If a one-year-old assessor record shows an apartment building there, that probably should override a 5-year-old ACS record showing a one-family, ranch house. We don't know if a brokers' association would provide records to the police even if the police were willing to pay.

            There may be other sources of computerized information about what houses and other buildings look like from the street; for example, building departments (which inspect buildings and process applications for building permits) and fire departments.

            In summary, the HIS database would be made by occasionally copying useful fields from other databases (donor databases), especially assessor databases. When there would be conflicts between information from two donor databases describing the same address, software would automatically resolve the conflict, probably by automatically erasing the old information. It is unlikely that doing any of this would be difficult.

         2. Help for firefighters

            The discussion above is written from the perspective of a police officer. Fire departments have, in addition to the same need to find an address quickly, a need for extra information. A fire truck crew on the way to a building might want to know about flammability, for example. Is gas connected? (Not all buildings receive gas from a local, public utility.) How many fireplaces are there? Is there a back door? Is there a side door? Is it possible to enter the building from outside going directly into a basement? How many rooms are there? A 911 police officer probably does not care about those things unless maybe there is a hostage problem (or criminals who refuse to come out or let anyone in). Thus, we guess that a fire department HIS would supply much more information than a police HIS would. This does not mean that a jurisdiction would need two HIS systems. One system could recognize whether a fire truck or police car is calling, and give appropriate information automatically. Another possibility might be two phone numbers to call: one number for police info only (in other words, how to recognize the building from a patrol car in the street), the other number for fire info (police info plus additional info useful to firefighters such as building flammability).

         3. Custom input

            By the way, after an officer is through with a 911 matter, he might call the HIS computer system to leave a 60-second recorded message that would become part of the HIS record for that address. In the future, when an officer (going to that address in response to a 911 dispatch) called HIS to get info, he would hear after (or before) the voice synthesizer saying the usual info, the recording made by the previous officer at that address in response to a 911 dispatch. For example, an officer might say (to help future officers going to that destination), "It's an old-looking, yellow house with a white, wooden fence in front.". Recording a message would be voluntary. In many cases, there would be no need to record because it would be easy to find the address without calling HIS (or because the voice synthesizer would describe the house so well it would be easy to find). If it had been hard for an officer to find an address, he could (if he wanted to) record a message that would help the next officer who called HIS about that address.

      
      
   5. most appropriate officer dispatched to 911 emergency

      Sometimes, the 911 system is busy. There is a shortage of patrol cars to respond to incoming 911 calls. Then, one dispatches the nearest patrol car that wants the case, or maybe no patrol car. For those busy times, this phase (most appropriate officer dispatched to 911 emergency call) should not be used.

      Sometimes, the 911 situation is quiet. There are patrol cars cruising but there isn't much for them to do. Then, when a 911 call comes in, it may be possible to dispatch two cars. One car is the closest. When is it appropriate to dispatch a second car (the auxiliary car), and how does the dispatcher choose the right car to dispatch as an auxiliary?

      Imagine that the 911 caller knows Spanish well but barely knows English. It would be nice to dispatch a Spanish-speaking officer, in addition to the closest officer, if a Spanish-speaking officer is nearby. Imagine that the destination address (1234 Liberty Street) was recently visited by an officer who is now on patrol nearby. Maybe he could be helpful because he has experience dealing with people at that address. Maybe a woman has been raped, and it would be appropriate to have a female officer interview her. The dispatcher may think that it would be useful to have a second officer respond to the emergency (not just any officer, but one whose background and traits make him appropriate for that particular 911 destination).

      The 911 dispatch computer system has a record of every officer who patrols: badge number, languages spoken, sex. The computer system, if it is given a destination address, quickly search its records for the badge number of every officer who's been connected to that address (for example, who went to that address a month earlier responding to a 911 call).

      At about the time an officer, early in his shift, gets into a patrol car, someone goes to a computer terminal in a police building and enters, into a computer form, the patrol car number (for example, 1234), and the badge numbers of all officers in that car (for example, badges numbers 1234 and 2345). The computer already knows about each officer (for example, that 1234 knows Spanish, or is a female, and many addresses that officer 1234 has been at). When the car is returned at the end of the shift, log out information for those officers is entered into the computer terminal.

      Imagine that a 911 call comes in, and the dispatcher thinks that a Spanish-speaking officer should be sent. He can make his computer automatically dial the nearest car (nearest to the destination) and also the nearest car (in that precinct) with a Spanish-speaking officer in it. Thus, two cars are dialed by the dispatcher's computer: nearest car (to get fast response) and nearest car (in that precinct) which has a Spanish-speaking officer (to get an appropriate response). The nearest car might have a Spanish-speaking officer in it, and the computer would realize that. Then, only one car's phone would be dialed. Maybe there is no Spanish-speaking officer on patrol in that precinct when a 911 call comes in for a destination in that precinct. Then, only one car (the nearest) would be dialed. When appropriate, two cars (nearest and Spanish-speaking) would be dialed by the computer. Thus, the dispatcher can easily dispatch an appropriate officer in addition to the nearest officer.

      Spanish-speaking is just one example of a trait that makes an officer especially useful in handling a situation.

      When any 911 call comes in, the 911 computer system would quickly know the destination address. The 911 system can search many police department records (because they have been copied into the 911 computer system) to find officers, on patrol when the 911 call comes in, connected to that address. Then, the 911 system can automatically display, on the dispatcher's video monitor, something like this:

      Cars Nearest to Destination
      1. Car 1234, 0.4 miles northeast of destination
      2. Car 3235, 0.5 miles south, Spanish, female
      3. Car 1436, 1.1 miles west, same address 3Feb2006 for 911 response
      4. Car 1298, 1.3 miles southwest
      
      In the example above, the computer is configured to display by default the nearest 4 cars to the destination, and car 1436 (third in the list) has an officer in it who was there in February 2006 responding to a 911 call. Ordinarily, the dispatcher would ask only the nearest car (1234, first in the list) to go. Imagine that a Spanish-speaking officer would be useful. Notice that a Spanish-speaking officer is in the second nearest car (3235), and that car is almost as close as the nearest car. The dispatcher might decide to dispatch the second nearest car instead of the nearest car. If it's a quiet night, so that two cars can be dispatched to the destination, the dispatcher might also dispatch the third-listed car (in addition to car 2), because the third-listed car has an officer in it who was there in February 2006.

      This system can result in appropriate officers showing up, officers who are well suited to the 911 call they respond to.

      Above, there is a description of how the computer discovers which officers are in each patrol car (badge numbers and vehicle numbers are entered into computer terminals at the beginning of each shift, in each precinct building and other building from which patrol cars operate). There are many other ways that this information could be entered. For example, while the patrol car cruises, an officer in the car could make a phone call, from the car, to the 911 computer to enter the information by pressing keys on the car's mobile phone. ("Welcome to patrol vehicle officer registration. Please press your badge number and then the star key. You entered 1234. If that's your badge number, press star. Otherwise, re-enter the number and press star." etc.) The 911 computer knows the telephone number that the call is coming from, so there's no need to enter the vehicle number.

      Similarly, officers can log out by making a phone call from their car, not just by entering log out information into a computer terminal in a police building. To log out is, in effect, to tell the 911 computer that the officer who logs out is no longer in the car.
      

   6. By the way
      1. 911 history and records
         1. General

            As far as we know: for any phone number, the phone company knows the customer's name and the service address (the street address where the phone is). As far as we know: when a 911 call comes in, those three facts (customer name, service address, phone number) should be saved (for a few years or several years, we guess) in the 911 computer system. When a 911 call comes in, the 911 computer system can realize (by comparing information about that call to information about previous 911 calls) that a previous 911 call came (1) from that customer's phone, or (2) from that telephone number, or (3) from that service address. Some customer-related facts can change. Sometimes, a telephone company's customer moves but keeps his telephone number. Sometimes, a customer changes his number but stays at the same address. We suppose that sometimes a customer changes his name but keeps his telephone number and address.

            Consider a call for help made to a 911 dispatch office, when a previous call came four months earlier. There are many possibilities for the dispatcher to give useful information to a police officer about to go to that address.

            The simplest is when all facts (customer's name, phone number, and address) are the same. Then, the dispatcher can say, for example, "We got a 911 domestic dispute call from the same address in February. Seems like the same people." If some (but not all) facts are the same, the dispatcher needs to pay attention to which facts are the same and which have changed.

            For example, consider a 911 phone call from a service address about a drunken neighbor who's trying to pick a fight. Later, the caller moves. A new tenant moves in. The same service address now has a new customer and a new telephone number. A 911 call comes from the same service address about a drunken neighbor who's trying to pick a fight. Maybe the dispatcher should mention to the police officer that earlier there was a 911 call from the same address about what might be the same problem.

         2. Medical: 911 medical records and information, EMIS
            1. Introduction

               Some people call 911 to ask for an ambulance or to report a medical emergency.

            2. Records about the possible cause of a medical emergency

               Below we provide a hypothetical example of 911 medical information. Then we discuss that example and related examples.

               Harry and George are watching TV in Harry's apartment. As they watch, they eat pretzels and drink soda. Harry goes to the kitchen to get some more pretzels. He doesn't come back. George goes into the kitchen, sees Harry unconscious on the floor, talks to Harry and nudges him a little, sees that Harry does not respond, goes to Harry's phone, calls 911 to ask for an ambulance. The 911 dispatcher automatically sees on his video monitor: the caller's name, phone company service address (street address where the phone is), telephone number, and "male 1988 diabetes". The word "diabetes" is useful to the EMT and ERU staff. George has no idea what happened to Harry but the 911 dispatcher has a good guess. How can the 911 dispatcher's video monitor automatically display "male 1988 diabetes" when a call comes from Harry's phone?

               We will now briefly sketch a system to automatically provide 911 medical dispatchers with a short message (for example, "male 1988 diabetes" or "female 1971 heart nitroglycerin").

               A telephone customer calls the telephone company and asks for its EMIS (emergency medical information service). He agrees to pay several dollars a year. In the postal mail, he gets a PIN (personal identification number) for his telephone number, and written instructions. There also are instructions in the phone company's Web site. He calls the EMIS telephone number from his telephone number, then enters his PIN. His PIN works only in telephone calls from the telephone number for which it was issued. The EMIS computer system verifies that the PIN matches the phone number used to call EMIS. He then is guided by a voice tree of instructions. He identifies himself as male or female. For example, "If the patient is female, press 1. If the patient is male, press 2.". The customer enters the year of birth. The customer types, using the telephone keypad, the name of the disease that he wants the 911 dispatcher to know about. In a phone company long list of just about every disease and medical problem. Such lists already exist. The list has scientific names (for example, myocardial infarction) and vernacular names (for example, heart attack). When what he types is matched by an item on the list, the computer tells him, and asks him if he wants that added. The list also has the names of body parts. If the customer wants to, he can type the word "heart", for example. That word, even without the name of a disease, might be useful to EMTs.

               The list also has medicines by scientific name, brand name, and vernacular name. Such drug lists already exist or can easily be made; for example, in America, from sources such as USP and PDR. Even without knowing the name of the patient's disease, an EMT is helped by knowing that the emergency may be related to the patient's taking insulin or nitroglycerin. Entering drug information into EMIS can be done by sitting by a phone, then typing into the telephone keypad the name of a drug by copying from the label. For example, if a patient has a vial of tablets, the vial probably has a label with the name of the drug.

               Every day at 1:00 a.m., the EMIS computer system checks to see if any information in each patient's file has changed since the previous day's check. If it has, the patient is notified. For example, he gets a copy by postal mail of the medical information that the 911 dispatcher will automatically be shown. There might be a fee for this. For example, if he gets notice by postal mail, there might be a fee charged by the phone company for the mailing.

               Some customers might have a long list. For example, a customer might have "female 1964 insulin hypertension narcolepsy infection antibiotic penicillin asthma bronchitis". For some telephone numbers, there might be two patients. For example, a phone company customer has a child. The EMIS account for that customer shows two patients, patient 1 and patient 2. Patient 1 has "female 1975 stroke". Patient 2 has "male 2003 asthma bronchitis Abcdol [a hypothetical drug's brand name] allergy".

               A customer could request the phone company send him information about his EMIS account by email, SMS, or fax instead of, or in addition to, postal mail. We guess that postal mail notices are the most private and the least likely to be abused by Internet crackers. EMIS is not an Internet system. Except to the limited extent requested by specific customers, there is no need to connect it to the Internet.

               Above, we introduced EMIS with an example of guest George calling 911 about host Harry. In the example, guest George does not know the cause of the medical problem, a situation host Harry foresaw. EMIS is useful even if the caller knows everything. For example, a man's wife could know he has diabetes but doesn't know the English word for it. EMIS could be useful there. Furthermore, many 911 dispatchers could more conveniently and quickly look at a video monitor and spot "male 1977 hemophilia" than get the information by talking with a 911 caller. The information would be on a video monitor even before the phone conversation began.

               Above is a sketch of one way in which EMIS might work. There are alternatives.

            3. 911 medical records related to treatment for a possible, medical emergency

               Some phone company customers may want emergency (EMT and ERU) staff to know about a medical problem even if it is unlikely to cause a call to 911. For example, a patient may want emergency workers to know that he is allergic to antibiotics. He doubts that his allergy will cause a call to 911 but he wants emergency staff to know about the allergy. For another example, he takes a drug daily. He doubts that it, or the condition for which he takes it, will cause a call to 911 but, because the drug might interact with drugs given to him for the emergency, he wants the emergency workers to know about it. He should be able to enter this information into the EMIS system.

            4. Routing calls to an ambulance dispatcher

               In some 911 systems, the same dispatcher does not dispatch police cars and ambulances. Some people may think that, if they call 911 in the future, the call is more likely to be for a medical problem than for a criminal problem. To save time, those callers may hope that the first 911 dispatcher who answers the phone can dispatch ambulances. If 911 management wants, that's easy to do. For example, the word "ambulance" could be available for EMIS users to choose. If they choose that word and then dial 911, they will first be connected to a dispatcher who can dispatch ambulances, if such a dispatcher is available when the 911 telephone call comes in. Such a user's information might be, for example, "female 1974 ambulance hypotension cardiac". Such a system would be based on EMIS users' forecasts.

               By the way, this routing can be done even without EMIS. Consider a 911 system without EMIS. Someone calls 911 to ask for an ambulance. An ambulance is dispatched. The 911 computer system saves a record of that incoming call and dispatch. Eight months later, another telephone request for an ambulance comes in, and again an ambulance is dispatched. The second request is from the same source (same service address, phone customer, and telephone number) as the first request. The 911 computer system saves a record of that incoming call and dispatch. Four months later, a third 911 call comes in from the same source as the first two calls. The 911 computer system automatically directs that call to a dispatcher with authority to dispatch ambulances. A smart routing system like that would be easy to install. It, too, would save time. Such a system would be based on history. The 911 computer system would observe how 911 dispatchers responded to a source's most recent two calls (ambulance dispatch or not), and would route that source's calls accordingly.

            5. Mandatory confirmation

               At least once every six months, an EMIS user would have to call EMIS, log in to his account, listen to a voice synthesizer say the medical information that the 911 dispatcher sees on his video monitor, and confirm that the information is still right. If he doesn't, he gets a reminder notice. If he still doesn't confirm, the information is no longer displayed to 911 and the user gets a notice telling him so. If he doesn't confirm even after getting the notice, his EMIS account is not renewed.

      2. One call ends it all.

         When the officer finishes responding to the 911 call, he should be able to dial the 911 system's computer, and type in his vehicle number. There is no need to type in the vehicle number if he's calling from the vehicle. He should reach a computer-operated answering system, not a human. The system should know the address that that vehicle was dispatched to. The system should tell him the address and he should confirm (by pressing star key, for example) that the system knows the address he is reporting about. Entering the information over the phone would be easy. There is almost no information to enter. If he doesn't tell the system that he has finished responding, a 911 dispatcher should call him to ask him if he's finished responding, should another vehicle be dispatched instead of his, should another vehicle be dispatched to help him, etc. Earlier, when he told a dispatcher that he would go to the 911 destination, the dispatcher entered information into the 911 computer system connecting that patrol car to that destination. Suddenly, the vehicle no longer appears as a green dot on the map on the dispatch video monitor. The dot color changes (to white, for example). A white dot is a patrol car devoted to one 911 destination. A white dot is not cruising. When someone with an emergency calls 911, the automatically displayed list of the four nearest patrol cars is limited to green dot vehicles (cruising vehicles). White dots appear on the dispatcher's map but they don't get listed because they are not available for possible 911 dispatch (because they are then on a 911 task). When the officer calls the 911 computer to report that he's through with his 911 assignment, the color of his car (on the dispatcher's map) instantly changes from white to green because his car is again cruising, available for 911 dispatches. When the 911 system is busy, a high proportion of the dots will be white. The officer's call to the 911 system (to report that he's through) should be from a secure phone (maybe the one in his car). If a mischievous stranger calls from his home phone, the 911 computer system should automatically reject the call.

      3. Saves gas

         An incidental effect of this proposal is that less gasoline will be used. If the nearest vehicle is the one that goes to the 911 emergency, less gasoline will be burned.

      4. Reliable

         In the paragraphs above, there is a description of a system which would help in some investigations, and would makes 911 response faster and more appropriate. Everything described above can easily be done using reliable, existing technology. Nothing described above requires new, untested technology.

      5. May not work everywhere

         Notice that much of the above dispatching system requires knowing where patrol cars are. It is not always possible to know that. For example, if a patrol car is on a narrow street with high buildings on both sides, that car's location may not be known by the system described above, we think. Therefore, that car might be the nearest car to a 911 destination yet not be dispatched. Instead, the second nearest car might be dispatched. This may be a petty problem. However, if much of a city has narrow streets with high buildings on both sides, that city may not benefit much from the dispatch system proposed in detail above, because such a city may be unable to automatically find many of its patrol cars.

         Repeaters may be able to solve this problem.
         

      6. FD (fast dispatch)

         Sometimes, a 911 dispatch office is slow. There may be many 911 calls coming in compared to the number of dispatchers working. The 911 phone may ring many minutes before a dispatcher answers the phone. Even then, a 911 dispatcher may put a caller on hold for a long time. Response time can be slow because of delay in the 911 dispatch office. Even if there are plenty of dispatchers on duty, the dispatcher's work is inherently slow. The 911 caller has to tell information to the dispatcher. This takes time even for an ideal caller. Furthermore, some callers are not ideal. They may be crying or hysterical. They may know little English or have speech defects. After the dispatcher finishes interviewing the caller, the dispatcher has to call a patrol car to ask if the officer in it wants to go to to the destination. Thus, there are two causes of dispatch delay: shortage of dispatchers, and inherently slow dispatches. There is a question of what can be done to speed work in a 911 police dispatch office.

         Consider a police officer who has responded to a 911 call at an address (for example, 1234 Liberty Street). His work on that 911 job is finished. He now calls the 911 computer system to report that he has finished that 911 dispatch job. Imagine that he guesses that there will be serious 911 trouble at that address soon. He guesses that there may be a serious 911 call from that address that day or the next day. If there is such a call, he thinks that the department should immediately respond. Right after he informs the 911 computer that he is finished with that 911 job, he can hang up. If he doesn't hang up, the system asks him, "How many days will this address have fast dispatch? Press a number that means the number of days." If he hangs up or presses 0 and star, that means no fast dispatch. If he presses 1 and star, that means fast dispatch for the remainder of that day for that address. In our example, he presses 2 and star, which means he creates fast dispatch for the remainder of that day and for the next day (total of two days). The computer then says to him, "If you want to, leave a one-minute message for the next officer. Wait for a beep." He then hears a beep. He can then leave a message not more than one minute long. He records a brief message with a few facts that the next officer there would want to know. "Talk to the wife, not the husband. There's a neighbor who wants a fight."

         About ten that night, another 911 call comes from that address. As soon as the incoming call reaches the 911 computer (even before the phone is answered), the 911 computer knows the address that the call is coming from, and checks whether the address should get fast dispatch. If the 911 computer does not recognize that the call is coming from that address, the call will get ordinary dispatch (not fast dispatch).

         Imagine that the 911 computer gives fast dispatch to the incoming call The computer automatically identifies the four nearest patrol cars. The computer now must decide which car to ask to go to that address. Usually, it chooses the nearest patrol car.

         The 911 computer can be programmed to choose a qualified car even if it isn't the nearest. For example, if there is a car now on patrol which is almost as close as the nearest car (to the destination address), and if that car has an officer who has already been at that address, the computer can choose that car to call. This is not hard to program.

         Anyway, the 911 computer now calls the patrol car it has chosen. All of this (deciding to treat the incoming call as fast dispatch, deciding which patrol car to call, dialing that car) takes several seconds, maybe just a few seconds. A voice synthesizer, not a human, talks. "This is fast dispatch for 1234 Liberty Street." The computer then plays a one-minute recorded message if there is one (a message recorded by the officer who put the address on fast dispatch). The computer then says, "Press 1 if you will go to 1234 Liberty Street, press 2 if you won't." If he presses 2, the computer hangs up, then calls its second choice patrol car. The computer never calls more than 2 cars for fast dispatch. If both cars refuse, the incoming 911 call is handled as an ordinary 911 dispatch.

         If a patrol car's officer presses 1, meaning that the car will go to 1234 Liberty Street, the computer then asks, "Press 1 if you need directions." If the officer presses 1, the computer (which knows cross streets and landmarks) says, "1234 Liberty Street is seven tenths of a mile northeast of you, crosses Jones Road [which means that the nearest cross street is Jones Road], and is west of Smith Park. Press 1 to make me say this over and over until you hang up." If the officer presses 1, the 911 computer will say the directions over and over until the officer hangs up.

         The officer then drives to 1234 Liberty Street and handles the 911 matter. No human dispatcher is necessary if a patrol car agrees to go there. An enormous amount of time-consuming talk has been skipped. If there's a shortage of human dispatchers, there is no need to wait until a human dispatcher is available.

         Remember the person who dialed 911 to ask for police at 12324 Liberty Street. He doesn't know that a car is on the way. As soon as he reaches the 911 computer, he is automatically put on hold. A voice occasionally tells him that the police department knows he's on the phone because of an emergency, and the department will help him as soon as possible. Sixty seconds after a patrol car presses 1 (meaning that the car will go to 1234 Liberty Street), the 911 computer tells the caller, "A police car is going to 1234 Liberty Street. If that's why you called 911, you can hang up. A police car is now going to 1234 Liberty Street." If the caller doesn't hang up, he is connected to ordinary dispatch (which has human dispatchers). After the human dispatcher in ordinary dispatch talks to the caller, the dispatcher can, if he considers it appropriate to do so, call the patrol car to suggest not going to the destination.

         Any officer (whose work involves 1234 Liberty Street) could set up fast dispatch for that address, not just an officer who had just finished 911 work there. For example, consider a person who goes to the movies, comes home, and sees that her place was burglarized. A police officer comes, talks to her and looks around, and guesses that the criminal broke in hoping to kill or rape her, then pretended that he was a burglar to avoid alarming her. This was not a successful burglary, it was an unsuccessful attempt at murder or rape, the officer guesses. The officer could put that address on fast dispatch for the next ten days. If a 911 call comes from that address in the next ten days, it will fly through the 911 system, skipping human dispatch (to save much time).

         Remember the one-minute message that an officer may voluntarily leave in the 911 computer when he creates fast dispatch for an address, for example, "Talk to the wife, not the husband. There's a neighbor who wants a fight." That message should automatically be deleted when fast dispatch ends.

         Should there be a maximum amount of time that an address can be on fast dispatch? Maybe any officer with a work connection to an address (for example, he went there in response to a 911 call, or as a detective) could put an address on fast dispatch for up to 15 days. Maybe only a high executive (the boss of a precinct, for example) could put an address on fast dispatch for longer than 15 days.

         Remember that an officer, if he creates fast dispatch for an address, can leave a message to be played for patrol cars being asked if they want to go to that address. Actually, recordings should be for all 911 dispatches, not just fast dispatches. When an officer finishes a 911 job, he calls the 911 computer to inform the computer that he's through. He should have the opportunity to voluntarily leave a one-minute message about that address. When someone calls 911 from or about an address (remember that some 911 calls are about a different address; for example, the house across the street), the dispatcher should be able to hear the message that was recorded by the last 911 officer at that address who wanted to leave a message. The message would be played two ways. In fast dispatches, the message would automatically be played by the 911 computer to a patrol car whose officer has to decide if he wants to go to the address. In ordinary dispatches, the message would be played by the 911 computer to the dispatcher if the dispatcher wanted to hear the message. When a 911 call is handled as an ordinary dispatch (not a fast dispatch), the dispatcher's video monitor would show information about recordings about that address. For example:
         

         recordings about address
         1. 2006 January 1, 12:17 a.m., badge 1234, 51 seconds, 1234 Liberty Street, 911
         2. 2006 March 3, 10:43 p.m., badge 5432, 28 seconds, 1234 Liberty Street, 911

         The dispatcher, could listen to those messages if he wanted to. If the dispatcher listened, he might handle the 911 call better, and he might pass on useful information to an officer in a patrol car.

         Fast dispatch can save much time. Sometimes, a 911 caller can wait 10 minutes for a dispatcher to pick up the phone. Sometimes, after the dispatcher picks up the phone (in other words, answers the call), he puts the caller on hold for 10 minutes. Fast dispatch skips all that. Fast dispatch can save much time compared to ordinary dispatch.

         Why can't everyone get fast dispatch? Why not abolish ordinary dispatch? Some calls to 911 shouldn't go to 911. For example, someone might dial 911 although he meant 411. Some calls to 911 aren't serious enough to result in a dispatch, especially if there is a shortage of patrol cars at that moment. Some calls to 911 aren't for the police. For example, they are for the fire department. A 911 dispatcher talks with 911 callers and screens incoming calls. Fast dispatch saves much time because the caller doesn't have to wait for a dispatcher to become available, and doesn't have to talk with a dispatcher. Fast dispatch is only for situations in which a police officer has been to the destination, talked to the people who would call 911 from that destination, and concluded both that there is a temporary need for exceptionally fast 911 police service for an address and that it is extremely unlikely that 911 calls from that address would deserve to be be screened out during the temporary period that fast dispatch is in effect for that address.

         There is no limit to how many addresses fast dispatch could apply to.

         By the way, FD (fast dispatch) might speed ordinary dispatch because FD would reduce the number of 911 calls handled by ordinary dispatch. By reducing the number of calls handled by ordinary dispatch, the backlog of ordinary dispatch might be reduced. People (whose 911 calls are handled by ordinary dispatch) might have their 911 calls answered sooner, and might be on hold for briefer periods, because fewer calls would be handled by ordinary dispatch.

         Solely to illustrate fast dispatch, we want to give an example of how it might work. A call to 911 is made from an FD address. The 911 computer recognizes the fast dispatch source of the call, finds the nearest patrol car, then calls that car. By coincidence, the car is right in front of 1234 Liberty Street. The driver presses 1 (agreeing to go to 1234 Liberty Street). Remember that the person who creates FD for an address may record a one-minute message. In this hypothetical example, there is no message. Right after the officer presses 1, he parks the patrol car, goes to the door of 1234 Liberty Street, and rings the doorbell. Someone inside opens the door, then sees one or two police officers there. No human at 911 has talked to the caller. From when the caller dialed 911 until when the doorbell rang, about 90 seconds passed. The caller has not yet heard a human voice or said anything to anyone. The caller may be astounded.

   7. Crime and crime rate by block

      As far as we know, a few police departments in America can easily get a computer report of all incidents (for example, arrests) at any, specific address. Those departments can easily see, for example, which ten addresses had the most incidents in the previous year or month. This kind of information helps police departments reduce crime. However, we guess that many of those high-crime addresses have a big apartment building. In general, the more people who live at an address, the more crime there will be at that address. For example, we guess that, all other things being equal, there's more crime in a big apartment building than in a one-family house or a small apartment building. Although it's valuable to know which addresses have the most crime, it would also be valuable to know which addresses have the most crime compared to the number of people living there. If an address has many incidents, that's bad. If the address has many incidents compared to the number of people who live there, that might be worse. If an apartment building with 200 residents has as many incidents as a building with 300 residents, it may be more dangerous to live in the first building. Police departments should know which places have the highest crime rate (crime divided by population).

      We don;t know how to easily get the population of each address in a city or county. The Census Bureau knows. Maybe the Census Bureau would, if asked, gave a city's police department the population of each address in the city, at least for addresses with a large population.

      If the Census Bureau does not tell the population of each address in a city, maybe the Census Bureau would tell the population of each block (or, at least, the population of high-population blocks). As far as we know, the Census Bureau has numbered each block in America and knows the population of each block.

      It would be useful for a police department to know, for example, the ten blocks in the city and the ten blocks in each precinct with the highest crime rate (crime divided by population).

      The amount of crime is necessary to calculate the crime rate. The amount of crime is the numerator. How can we easily find out how much crime there is on each block in a city? If we know how much crime there is at each address, all we need to know is which addresses are on each block. We guess that the Census Bureau has software that connects every address in America to the block it's on. Maybe the Census Bureau will share this software.

      Local real estate tax assessors estimate the value of real estate. They number every parcel of real estate in their locality. For example, the Foobar county real estate tax assessor has assigned an identification number (a parcel number) to every parcel of real estate in Foobar county. He has software which can translate between street address and parcel number (for example, 1234 Liberty Street is parcel 123-456-789). We don't know how parcel numbers are created. Different assessors may have different systems for deciding which number each parcel gets.. We guess that some assessors number each block, then number each parcel. In some counties, a parcel number includes the number of the block that the parcel is on, we guess. We guess that, if an assessor looks at two parcel numbers, he doesn't need to find them on a map to know if they are on the same block. If our guess is right, one could easily figure out how much crime there is on each block without Census Bureau cooperation (if the police department knows how much crime there is at each address) because the assessor has software to convert each address to parcel number, and it's easy to aggregate parcel numbers into the blocks they are on. This would allow a police department to know the amount of crime on each block the way it now knows the amount of crime at each address. For example, a police department could know the ten highest-crime blocks in its city and in each precinct.

      A block could be the highest crime block in a city even if it did not have the highest-crime address in that city. It is important to know which blocks have the most crime and the highest crime rate (crime divided by population).

      
      

   8. Patrol statistics
      1. Patrol information

         A police department can know where all patrol cars are at all times, as described far above on this page in the suggestions for improved 911 dispatching. This system involves each patrol car often broadcasting a message to a police department's computer system. By the way, to avoid draining a patrol car's battery, it would be a good idea just to have cars broadcast location information when their engines are running.

         There now exist computer data files which, for an address, specify that address's latlon. For each address in the data file, there is a latitude and longitude. 1200 Liberty Street has a latlon. So does 1300 Liberty Street. If one wants the latlon for 1250 Liberty, the computer automatically estimates based on the latlons for 1200 and 1300. When a Web surfer types an address into a Web map site, a computer finds the address's latlon (or estimates it), then plots that latlon on the map that the Web surfer sees.

         Above on this page, we suggest that police departments know the GCs (for example, latlons) of patrolling vehicles to improve 911 dispatch.

         Consider a police department in which all patrol cars, while their engines are running, frequently broadcast a signal which allows their location to be known. A computer saves that information. For each broadcast, the computer records which vehicle made the broadcast, the time of the broadcast, and the vehicle's GC at the time of the broadcast.

         The department would now have much information about patrol in that city. For example, a patrol supervisor could, sitting at a computer terminal, type in an address, month, and distance, and see a total. For example, he could type in "1234 Liberty Street", "August 2005", and "200 feet", then see as his answer, "247 times in August 2005, a patrol vehicle was within 200 feet of 1234 Liberty Street. To see an itemized list of all 247 times, click here." Consider a person who calls his local police station to request more that patrol cars drive past his address. An officer takes the call. While the caller is on the phone, the officer asks the computer (in other words, types into a Web form): 1234 Liberty Street, 14 September 2004, 75 feet. The computer answers (in other words, displays on the video monitor, "9 times on 14 September 2004, a patrol vehicle was within 75 feet of 1234 Liberty Street." The officer tells the caller, "9 times yesterday, 14 September 2004, a patrol vehicle was within 75 feet of 1234 Liberty Street."

         The officer then asks the computer about that day (the day he's talking to the caller by pone). The computer answers 4, which means that the total so fat that day is 4. "The officer tells the caller, "4 times so far today, a patrol car was within 75 feet of your address, 1234 Liberty Street." The officer then clicks an icon on the monitor, sees a list of the four times, and reads them to the caller: "1:16 a.m., 5:50 a.m., 7:24 a.m., and 8:43 a.m."

         This information is useful to both people (the civilian caller and the officer talking to him) in figuring out if the address gets enough patrol.

         The patrol information computer program examples above give only a total and a time period (in the examples, a total for a month or a total for day). The officer should be able to enter a range of time (for example, from 3 January 2003 to 17 April 2003). The officer should be able to get an average, not a total, For example, he might want the daily average for the previous 90 days, or the hourly average for the last 48 hours. He might want the daily average for the last 5 Saturdays. This kind of information is easy for a computer to calculate and display. This kind of information helps him decide if the address gets the right amount of patrol.

         For any place, it is easy to know what sunrise and sunset will be every day, years in advance. We suppose that a good rule of thumb might be that it's dark outside from half an hour after sunset until half an hour before sunrise. In a computer, there could be a data file which lists when it will be dark outside for many years to come. For some addresses, one might want a different frequency of patrol for dark and light hours (for example, for 1:00 a.m., when it's dark outside, and 1:00 p.m., when it's light). For example, one might want more patrol when it's dark outside than when it's light outside. An officer might have the computer to display, for the previous 30 days, the average number of times per day (for dark hours only) that a vehicle has been within 75 feet (or any other distance) of 1234 Liberty Street. He might discover that there's too little patrol at night, for example.

         Dummy addresses are an important concept. Consider a city with a beach. Sometimes, patrol cars drive on the beach. An officer wants to know how often patrol cars are within a certain point on the beach. The point has no address. The officer goes to the beach. He walks to the point he's interested in. GPS devices are highly portable. Many are about as big as a CD player. GPS devices are easy to use. Turn it on, press only a few buttons, wait a little, then the GPS device displays latitude, longitude, and altitude. GPS devices are truly easy to use. The officer sees latlon on the GPS device, then writes down the two numbers (lat and lon). He has a piece of paper with three notes: where he's standing (for example, "Liberty Beach approximately midway between the boardwalk and the water, southeast of where 14th Street deadends at the boardwalk"), latlon he copied from the GPS (for example, "latitude 41.1234567, longitude -75.1234567"), and a dummy address that he arbitrarily creates (for example, "1000 Liberty Beach Plaza"). There is no such address. He could have created a different, dummy address; for example, "987 Foobar Lane". Anyway, he repeats this process for ten more places on the beach. Each place gets a unique dummy address. He then enters the information for all eleven places into the department's patrol report computer system. The dummy address is stored with real addresses. Now, he can ask the computer how many times yesterday a patrol car was within 300 feet (for example) of 1000 Liberty Beach Plaza; in other words, the spot on the beach where he stood. He can now learn how much patrol the beach gets.

         Dummy addressees can be used for anyplace; for example, the middle of a bridge, a spot in a park, a spot in a river. It's easy for the officer to know when patrol vehicles are within a specified distance of that spot.
         

      2. Itemization

         The officer can get information about how many marked cars, and separate information about how many unmarked cars, were within a specified distance of an address. For example, he could discover that, on 5 March 2008, there were 5 marked vehicles and 1 unmarked vehicle within 75 feet of 1234 Liberty Street between 3:00 p.m. and 7:00 p.m. He might decide that more unmarked vehicles should go past an address but that the number of marked vehicles patrolling is is fine.

         The system could list every address in the city in rank order of how many times a patrol car was within a certain distance of that address. At one end of the list would be the most patrolled address, which we guess might be a police station or a police garage. At the other end of the list would be the least patrolled address. It would be interesting to know if that address was under-patrolled. An address might get the right amount of patrol even if gets less patrol than any other address. One could make similar lists for each precinct. One could find out how much patrol each of the hundred highest-crime addresses in the city gets.

         Consider a car which goes within 75 feet of an address at 1:00 p.m. and at 2:00 p.m. That counts twice. The system counts how many times a car is there, not how many cars are there. However, if a patrol supervisor wanted to know how many unique cars were there (within 75 feet of 1234 Liberty on a specified date, for example), he could. If he wanted a count of cars, the system would count each car only once regardless of how many times it was there.

         Far above on this page, we suggest a way to dispatch cars to 911 destinations. A city could adopt that 911 dispatch system, this patrol report system, or both. Each works well without the other. There is a little synergy that results from adopting both systems. For example, if the patrol report computer system can read data in the 911 dispatch computer system, then the patrol reports would ignore patrol cars going to a 911 dispatch (cars shown by a green dot on the 911 dispatch maps mentioned above in the discussion of the 911 system). Consider a car going to a 911 destination. The car passes 1234 Liberty Street. That car is not on patrol at that moment although normally it is a patrol car. The time that that car (going to a 911 destination) is near 1234 Liberty should not be counted by the patrol report system because the car is not on patrol as it passes 1234 Liberty. This (the patrol report system ignoring a patrol car which the 911 computer system knows is going to a destination) is easy to do. The result of the synergy is that the patrol reports are more accurate.
         

      3. Weakness

         Consider a vehicle that periodically broadcasts its message. It might broadcast a message when it is not within 75 feet of 1234 Liberty Street (the car might be at 1180 Liberty Street, more than 75 feet from 1234 Liberty Street), go within 75 feet of 1234 Liberty Street, go more than 75 feet from 1234 Liberty, then broadcast when it's in front of 1300 Liberty Street (more than 75 feet from 1234 Liberty). The department's computer would think that the car was not within 75 feet because the computer never received a broadcast made while the car was within 75 feet of 1234 Liberty.

         The less often the car broadcasts, the worse the problem. For example, if a car broadcasts once every three hours, a huge number of addresses will be missed. Therefore, the car should automatically broadcast often. The more often there are broadcasts, the fewer addresses will be missed. If broadcasts are frequent enough, extremely few addresses will be missed by the computer system even for patrol cars moving 60 miles per hour (which is much faster than the speed of most cars on patrol).

         By the way, if cars patrol in a neighborhood with many, tall buildings on narrow streets, many broadcasts may fail to reach the department. This would result in the department's computer system thinking that addresses there are patrolled far less often than they really are patrolled. This is a weakness of the system. No matter how frequently a patrol car broadcasts, unreceived broadcasts will be ignored by the department's computer system.

         Repeaters may be able to solve this problem.
         

      4. Speed

         The slower a patrol car moves, the more likely it is to be detected by the system. To use an extreme example, consider a patrol car going 1 foot every 5 minutes as it drives past 1234 Liberty. The department's computer system will know that it was within 75 Feet of 1234 Liberty Street. To use another extreme example, if a patrol car goes 120 miles per hour past 1234 Liberty Street, the system may not know that the car was within 75 feet of that address (unless the car broadcasts its location extremely frequently). The faster the car goes, the more likely its proximity to 1234 Liberty is to be missed by the department's computer system. For example, a patrol car going 85 miles per hour past 1234 Liberty Street is much more likely to be missed by the department's system than one going 30 miles per hour.

         Patrol is usually at low or moderate speed. Police can't patrol well at high speed. To patrol a neighborhood well, a patrol car should travel slowly enough so that officers inside the car can notice, for example, people on sidewalks, people crossing the street, and people in vacant lots. A car going 100 miles per hour is not really on patrol. Many of the few cars missed by the patrol report computer system probably would not really be patrolling, they would just be quickly driving somewhere.

         If the car broadcasts its message often enough, the department's patrol report computer system will know all addresses passed despite the car's high speed.
         

      5. Inference

         In addition to counting how many cars were within 75 feet, the computer should infer. The computer's display of patrol information should include the count (explained above) and the inference (explained below).

         Consider a request for information about how many times a patrol car was within 75 feet of 1234 Liberty Street yesterday. The computer can display 14 as the count. This is the number of cars which broadcst within 75 feet. However, the computer can infer how many other cars were within 75 feet although they didn't broadcst while they were within 75 feet.

         The computer can see how many cars were within 75 feet of 100 street numbers of each side of 1234 Liberty (in other words, from 1134 Liberty to 1233 Liberty, and from 1235 Liberty to 1335 Liberty). For example: at one time yesterday, a patrol car was within 75 Feet of 1146 Liberty, and, soon afterwards, that car was within 75 feet of 1332 Liberty. The computer infers that the car was within 75 feet of 1234 Liberty between the two times The inference system is for perfectionists. The basic system would work well but might occasionally miss an address passed by a patrol car. The inference system would permit the computer system to infer how many times, in addition to those counted by the basic system, patrol cars were within, for example, 75 feet of 1234 Liberty Street. By the way, even the inference system wouldn't infer all cars. For example, if a quickly moving patrol car turns left or right every time it comes to an intersection (never driving straight through an intersection), it may sometimes be missed by the inference system. This is because, in essence, the inference system has the computer automatically see if a patrol car was on the street the management officer is interested in (in the example above, Liberty) at a nearby street number above and below (the street number the officer wants to know about; in the example below, 1234) at about the same time. Therefore, if a patrol vehicle changes streets whenever possible, it may not be detected by the inference system.

         The normal way to change streets is to turn right or left. It's dangerous to turn from one street to another at a high speed. If a patrol car is changing streets whenever possible (which is not a common way to patrol, by the way), it's probably not going quickly. The more slowly it goes, the more likely the basic system is to detect it (in other words, the more likely it is to be counted).

         In conclusion, the basic system should work excellently and the inference system should make it even better.

      6. C+I

         If one adds inferences to the count, one can get even closer to the true number than if one uses the count alone. We use the abbreviation C+I to mean count plus inference. The computer can be programmed, whenever it reports on patrol, to supply the count and the C+I. For example, a computer monitor might display, "On 25 June 2008, the number of times a patrol patrol was within 75 feet of 1234 Liberty Street was: count 17, C+I 18.". In other words the system detected 17 vehicles and inferred the presence of an 18th.
         

      7. Details

         The computer system should allow a patrol supervisor to use any length unit he feels comfortable with: feet, yards, meters, furlongs, kilometers, miles. He should be able, for example, to find out how many times a patrol car was within a furlong of an address.

         An officer should be able to find out about an entire street or a range of addresses on a street; for example, how many times a patrol car was within 50 yards of any address from 1100 Liberty to 1800 Liberty.

         Above in this discussion of patrol reports, we basically discuss asking about circles: a point (defined by an address) and a radius (distance from the point). An officer should be able to learn about the patrol of an area which is not a circle. For example, if he gives three addresses, those addresses define a triangle. He should be able to ask how many times a patrol car was within that triangle. Asking about non-circular areas is useful. For example, he might be interested in patrol in a neighborhood that has the shape of a quadrilateral.
         

      8. Conclusion

         The proposed system generates two numbers: count (how many times vehicles were within a certain distance of a certain address) and C+I (in which the computer system infers a supplemental number and adds it to the count). Right now, patrol supervisors don't have a clue how much patrolling is done at various places. Consider a department which wants to decrease crime at an address in its jurisdiction. Sometimes, increasing patrol is a good idea. Some kinds of crime go down if there's more patrol. Sometimes, there is already so much patrol of an address that an increase in patrol there is unlikely to help. With the system suggested above, patrol management can easily learn how much patrolling there is at any address or in any area at any time. This could lead to more productive deployment of patrol vehicles.

   9. Patrol officer running a make on an address 2206dec14
      1. Introduction

         Running a make on an address is getting crime-related information about an address. Now and then, a patrol officer may want to do this.

      2. How to run a make

         There are several ways to do this. Here, we give an example or two. The officer has a mobile phone in his car. He dials a police department telephone number. He's asked to enter his badge number, which he does (by pressing buttons on the phone or by speech recognized by a voice recognition system). Maybe, as a security measure, he's asked to enter a password. We guess that usually he's near the address (for example: in front of the address, diagonally across the street, half a block away). He's probably not running a make on an address two miles away. Recall that the computer knows the GC of where he is and therefore knows the closest address or can estimate it. The computer asks him (based on its opinion of the address closest to where he is), "Do you want to run a make on 1234 Liberty Street, Lake city, Mountain county? Y for yes, n for no." If he types y, he hears the menu of different kinds of information (described below). If he types n, he's asked if he's within 500 feet of the address, yes or no. He enters y. (If he doesn't know if he's within 500 feet, or if he knows he's not within 500 feet, he enters n for "no".) He's asked to enter the street number. He enters it. He's asked to spell the first word of the street's name. For example, in Liberty Street, the first word is "Liberty". He's asked to spell the second word, and so forth. The computer then says, "Do you want 1234 Liberty Street"?

         After the computer knows the address, there are questions about what part of the address. For example, the officer might want to run a make on the entire address or just on apartment 15, suite 3B, or the rear entrance.

      3. Short cut

         Often, the officer won't have to enter the full address. For example, there might be only one "1234 Liber" in the city. Then, as soon as he enters the "r", the computer will ask him if he wants 1234 Liberty Street.

         Consider a police department which automatically knows where all vehicles are while their engines are running. Recall that the computer asks the officer if he's within 500 feet of the address. Usually, he'll type y (meaning yes) as his response. That narrows down the possibilities. Maybe there's only one address with street number 1234, or only one address starting "1234 L" within 500 feet of his patrol car. Often, if he's within 500 feet of the address he wants to run a make on, the computer will know what he wants after he enters just a small part of the address.

      4. Voice recognition

         The officer should be able to say his responses instead of pressing buttons on a phone. Voice recognition software does not recognize everyone's speech. For example, maybe some officers have an accent or speech defect, or just an unusual pronunciation. They should get personalized voice recognition. They would go into a police building. There, they would speak into a phone just like the one in a patrol car. They would say all ten digits (zero to ten), all 26 letters, and some words and phrases (for example: yes, no, say that again). When an officer later entered his badge number (as part of running a make on an address), the computer would immediately check if he had personalized recognition. If he did, the computer hearing him say "3", for example, would compare that sound to the recordings he had made (for example, to his recording of "3"), not to the computer's knowledge of how most people speak. Most officers won't need personalized voice recognition. The system won't understand some officers even with personalized voice recognition. To enter information, they will need to push buttons on the phone (for example, the button that has the number 1)..
         

      5. What information is supplied to the patrol officer running a make on an address

         The computer can supply a few kinds of information: information from the police department's own records, information that the police department got from other government agencies (for example: probation, parole, alcoholic beverage control, real estate assessor), information which results from research that the computer does while the officer waits.

         The computer might say, "Press 1 for 911 (info about 911 requests from that address and 911 dispatches to that address), 2 for arrest (arrests at that address or of people who live at that address), 3 for crime and incident (crime and incidents at that address, as shown by reports of crime and incident at that address), 4 for criminal record of owner, 5 for parole and probation (people living at that address), 6 for real estate (for example, name of landlord or owner and, if the building's an apartment building, number of apartments), 7 for ABC (info about businesses there which have a license to sell alcohol; for example, bars, liquor stores, some restaurants), 8 for other info, 9 for everything (on this menu), 0 (zero) to repeat this menu."

         Some of the information on the menu requires records discussed above on this page. For example, if a police department does keep records, by address, of where arrests are made, it is possible for anyone in that department (not just a patrol officer trying to run a make on an address) to find out about arrests at that address. Consider a department which has all of the records discussed above on this page. The patrol officer presses 2 because he wants arrest information. A voice synthesizer tells him, "3 arrests in the five years ending today. possession of controlled substance, Monday 3 March 2003, concealed weapon Friday 17 January 2005, battery Saturday 30 April 2005. Press 1 for more information about these arrests, 2 to return to the main menu, 3 to repeat this message." If he presses 1, he gets more information about the addresses; for example: the time of day that the arrest occurred; name, age (which the computer usually figures out because it usually knows the arrestee's date of birth), and description of the person arrested.

         What if there's a flood of information in a category? For example, when someone calls 911, it's easy to know the caller's telephone number, which means one can automatically find out from the phone company that number's site address (address where the phone is installed). A police department can have records, by street address from which a 911 call came, of all 911 calls. Consider an officer who presses 1 because he wants 911 information. He hears, "273 calls from this address in the five years ending today, 211 dispatches to this address in the last five years. Press 1 for detailed information about the last five dispatches to this address, press 2 to hear recordings of 911 calls from this address (the department may save 911 telephone call recordings for a few days), press 3 to hear all this again."

         This discussion of a telephone tree is an off-the-cuff sketch. The purpose is to briefly explain our idea of approximately how the tree would operate.

         Consider an officer who presses 4 for criminal record of owner. County assessors generally keep records by parcel number, and have software which automatically converts most, not all, street addresses into a parcel number. The police department probably can get a copy of the assessor records that the police need to run a make on an address. If the conversion software doesn't work the way the department likes, the department can modify the conversion software in the department's computers. For each parcel number, assessor records show the owner (for example, "Eric W. Smith and Jane J. Smith"). Consider a police department which periodically (maybe once a year) gets a copy of the assessor's records, or at least enough of the records to know the owners' names. If a patrol officer enters a street address he's curious about, the police computer can tell the parcel number (using the address-to-parcel-number conversion software that the department got from the assessor); and, from the record of that parcel number, the police can get the parcel owner's name as it appears in the most recent copy of assessor records that the police department has. By the way, the average parcel of real estate is not sold often. If one knows who owned a parcel a year ago, one probably knows who owns it today. Getting back to our example: the police computer now knows "Eric W. Smith and Jane J. Smith" and "1234 Liberty Street". Assessor records may show that the parcel is a homestead; in other words, the owner claims to live there. In other words, the owner of the address claims to live there if the property is a homestead. The police department computer now needs to get the criminal records of owner-inhabitants Eric and Jane. The department's own records may contain some of Eric's and Jane's criminal records. For example, the department's own records may show that it arrested Eric W. Smith at that address on 2 February 2004. If the department keeps records by name of arrestee, the records might show that Eric W. Smith, of that address, was arrested. If the department keeps records by address of arrest, the records might show that he was arrested at that address. The department's computer can also, we guess, automatically get criminal records of Eric and Jane from other law enforcement agencies; for possible example, the state police. In many states, all arrests must be reported to a state agency. Maybe that state agency's computer system has a record for a ticket for DUI given on 31 October 2002 to Jane J. Smith of 1234 Liberty Street although the ticket was given somewhere else in the state. What we've sketched above is simple research that a police department's computer can automatically do. The research will pick up some of the criminal record of the owner-residents. The search may not seem simple but it is. Slightly complicated is to have the police computer look through assessor records to find Eric and Jane's previous address, the address at which they lived before they lived at 1234 Liberty Street. Many assessors' records show when property is acquired. The department's computer could sometimes find a previous address, then infer beginning and ending dates of residence at that address, then search for criminal records of Eric and Jane at their previous address while they lived there. The assessor records sometimes have other clues of when residence begins and ends, clues that a police computer could automatically search for. Assessors have different amounts and kinds of information in their records, and cross-index records differently. Real estate tax collectors are different from real estate assessors. Those tax collectors send bills for each parcel to the address that the owner wants the bills sent to. The tax collector has lists of who owns each parcel. Many tax collectors probably have archives showing previous owners of each parcel and the dates of when ownership changed (or, at least, the date that the collector decided that ownership had changed, which probably was soon after the ownership change). For some assessors' records, it would not be possible to find an owner's previous addresses if the police department uses a search algorithm that's only slightly complicated. However, even the current address can generate a criminal record as shown in the example above (Eric's arrest and Jane's ticket). The owner-inhabitant's criminal record (supplied to the patrol officer running a make on an address) can be useful to him even if it's incomplete. Many parcels are owned by corporations but we suppose that corporations' criminal records are not useful to patrol officers. What about property owned by a human who does not live there? Consider 1234 Liberty Street, a non-homestead property owned by John Smith, who wants the real estate tax bill mailed to him at 9876 Jones Street. The computer could tell the officer that the property's owner is John Smith, and that he doesn't seem to live there. The computer could then tell the patrol officer about arrests and tickets of owner John Smith of 9876 Jones Street.

         We guess that a police department could probably get periodically, from parole and probation agencies, computer media (CDs, for example) with information about people on parole and probation: name, address, date of birth, and much more. That information could be entered into the department's computer system. If the patrol officer (when he runs a make on an address) types 5 for parole and probation, he should hear about people on parole or probation who live at that address; for example: name, age, description, the crime for which the person is on probation or parole. This information would be provided in brief and detailed versions. The officer would hear a brief version with the only the most important facts. Then he would hear a menu (1 for hear the information again, 2 for a detailed version, 3 to go back to the main menu).

         The department probably can get information about buildings from local housing, building, and rent control agencies; for example, the owner, the landlord, whether the building is residential (as distinguished from an office building, for example), how many rental units, condo units, coop units there are. By the way, occasionally the landlord is not the owner. For example, in New York's Empire State Building, we are under the impression that tenants pay rent to the landlord but that a different corporation owns the land under the building and maybe the building too. When the landlord is not the owner, we guess that it is unusual that either is a human. Anyway, information about a building might include how many units there are (for example, how many apartments) and the manager's name.

         The patrol officer could press 7 for information about businesses there which have a license to sell alcohol; for example, bars, liquor stores, some restaurants). For example, if there's a bar on the ground floor of 1234 Liberty Street, pressing 7 would tell the officer about it. Maybe the record would include the bar's disciplinary record (for example, the bar got in trouble five months ago for selling alcohol to minors), which occasionally would be useful.

         The patrol officer could press 8 for other information. If he pressed 8, he would recordings that other officers had made about that address. An officer could make recordings intended to help other officers curious about that address. A few, possible examples are here.

         1. "There have been 3 push-in robberies here since March 2009. When the robber's in the building, he has an accomplice sitting in a getaway car nearby."
         2. "This used to be a crack house. People still come here once in a while looking for crack."
         3. "A teenaged boy was shot on the sidewalk here in a drive-by shooting about 1:15 a.m. on Sunday, 3 August 2012."
         4. "This is one of the highest-crime addresses in the precinct."

         Someone would occasionally have to listen to all the recordings to see which should be deleted. There are two people who could delete: an officer whose job it would be to delete, and the officer who recorded the message. Periodically (maybe every six months), every officer who had recorded a message would be sent a memo listing his recordings and asking him to listen to them occasionally to delete any that should be deleted (because they are no longer true or useful, for example). Maybe, after a recording is made but before it can be heard by officers running a make on an address, the recording should be screened in advance by a supervisor (who would approve the message, reject it, or replace it by a different message that he would record). There are a few ways to screen in advance. For example, after a message is recorded, it could automatically be sent to a computer directory (folder). Whenever a supervisor wanted, he could listen to the messages. He could approve or reject messages, or he could record his own message to be used instead of one in the directory.
         

      6. How the department gets information from other government agencies

         Above, we suggest a system which is convenient to the source: that the source occasionally supply the information on computer media such as CDs. This is convenient to the agency which supplies the information. There is at least one other way to provide the information. The source department has a special computer which just has information that the receiving department wants. The receiving department's computer system is connected to that special computer and can copy from that special computer at will. This allows the receiving department to have recent information and eliminates the nuisance of periodically getting computer media. However, maintaining a special server might be a nuisance to the source department; for example, an assessor's office.

      7. Public relations problems and legal problems
      8. Where am I?

         Consider a patrol car driver who doesn't know where he is. Maybe he's in a neighborhood he doesn't know, on a dark street, in heavy rain (so he trouble seeing through the windshield), in a thick fog, and he can't read the street numbers in front of houses (or there aren't any street numbers there to read). He stops driving. He's in front of 1234 Liberty Street but doesn't know that. He starts to run a make on an address. The computer knows where he is. After he enters his badge number, one of the first questions the computer will ask him is, "Do you want to run a make on 1234 Liberty Street, Lake city, Mountain county? Y for yes, n for no." Now he knows where he is. If he doesn't want to run a make, he just hangs up.

      9. City

         Notice that, when an officer runs a make on an address, the computer asks him a question which includes the city (in the example above, Lake city). There are at least two, convenient sources of information about what city an address is in: post office and assessor. Assessor records always show which city the address legally is in. If the address is one inch outside the city (in unincorporated territory, for example), assessor records will not show that address in the city. The post office cares about efficiently delivering mail. If an address in Los Angeles city is one inch from Beverly Hills city, the post office might have the Beverly Hills post office deliver mail to it. The post office would tell people who live there that Beverly Hills, not any other place, is the city in their address. People who live there will have the Beverly Hills zip code even though they are not really in Beverly Hills. We guess that a law enforcement agency would identify the city based on the legal definition (which the assessor has but the post office doesn't). This distinction applies to city only, not to county. Postal records and assessor records never disagree about the county, as far as we know. By the way, the Census Bureau's computer records (like assessors' records) show which city an address legally is in.

   10. Help me get to a destination

       Consider a patrol officer who wants to drive to an address but doesn't know how to get there. From his car, he could dial a police telephone number, then choose a menu item that will guide him there (for example, "Press or say 6 for help getting to an address."). Then, he enters the address.

       The computer records his present GC. The computer tells him to drive straight ahead. After he drives at lest 150 feet, the computer records a second GC. From those two GCs, the computer knows the direction that the officer is facing; for example, 91 degrees (which is slightly south of east). The computer calculates the GC of the destination address. The computer knows the direction of the destination from him, and its distance. The computer talks to him as if he's in the middle of the face of a clock, looking at the 12. If the destination is ahead of him, the computer says that it's at 12 o'clock. If the computer is directly behind him, the computer says its at 6 o'clock. If the destination is to his right, the computer says it's at 3 o'clock. In the example, we're using, the computer says, "1234 Liberty Street is 3.4 miles away at eight o'clock." Six o'clock is behind him and 9 o'clock is to his left. The destination is behind him to his left. He starts driving.

       The computer knows his distance from the destination when he started driving, 3.4 miles. Once every minute, the computer tells him his distance to the destination, the name of the street he's on or closest to, and, if he drive approximately straight for the ten seconds preceding the announcement, the direction of the destination; for example, "1234 Liberty Street is 3.1 miles away, River Road, 11 o'clock.". He now knows that he's getting closer and that he should drive a little to his left. When he's within a mile of the destination, the computer tells him the distance in feet. When he's within 50 feet of the destination, the computer tells him so.

       In the example above, the computer talks to him every minute.

   11. APC: automatic patrol change
       1. Introduction

          Consider a patrol supervisor who knows that patrol cars drive within 50 feet of 1234 Liberty Street about 6 times a day. He would like a patrol car to drive past the address about 12 times a day. He logs into his department's APC program. He enters the address (1234 Liberty Street) and a distance (for example, 500 feet). He records a message (for example, "Low priority. Patrol past 1234 Liberty Street."). The department's computer system knows where all patrol cars are while their engines are running. His car has a speaker phone. When the system discovers that a patrol car has come within 500 feet of 1234 Liberty Street, the system automatically calls that car and plays the message. He then patrols past 1234 Liberty Street. This is called SR (stimulus response). The stimulus is that a patrol car came within 500 feet of 1234 Liberty. A second ago, the car was not within 500 feet but now it is. That's the stimulus. The response is that a patrol car patrols past 1234 Liberty.

          A patrol supervisor can check often to see how much patrol there is within 50 feet of 1234 Liberty Street. If there's still not enough patrol, he can extend the distance (for example, to 700 feet). More cars will come within 700 feet than within 500 feet. He can have different distances for different times, by the way. For example, he could have 700 feet for when it's dark out but only 500 feet for other times. He could have one distance between 8 a.m. and 5 p.m. on business days but a different distance at all other times and days. He increases the distance to get more cars to patrol past 1234 Liberty and reduces the distance to get fewer.

          The recorded message starts with the words "Low Priority." What does that mean? When an officer driving patrol hears the message, he needs to know how important it is to patrol past 1234 Liberty Street. Is it a life-and-death matter. If he's in the middle of giving someone a ticket for not stopping at a stop sign, should the officer abandon giving the ticket to instead patrol past 1234 Liberty? Should the officer turn on his siren and lights, and rush to 1234 Liberty? Low priority means that he should do the response (patrol past 1234 Liberty Street) if the only thing he's doing is driving patrol. If he's watching something interesting across the street that he wants to continue watching, he should continue watching. If he's following a car that he wants to follow, he should follow. If he's going to lunch, he should go to lunch. If all he's doing is driving patrol, he should do the response (patrol past 1234 Liberty Street). Low priority also means that the officer should patrol over there, not rush. If the patrol officer, while patrolling toward 1234 Liberty, notices a car he wants to give a ticket to, he should give the ticket. Let's say that, when the officer has finished giving the ticket, he no longer remembers the address he's supposed to go to. Fine, he can resume patrol and forget all about 1234 Liberty. Low priority really is low priority. A nice feature of low priority is that it's easy to understand. The department could create a high priority response. The department could create many priority levels.

          Sometimes, several patrol cars drive near each other. For example, consider a restaurant that several patrol cars meet at for lunch. When lunch ends, several patrol cars drive out of the restaurant parking lot at about the same time. If 1234 Liberty Street is near the restaurant, four patrol cars may come within 500 feet of the address at about the same time, causing all four to hear the message, causing all four to patrol past 1234 Liberty Street at about the same time. Maybe this is fine with the patrol supervisor. What if the patrol supervisor doesn't want four cars patrolling past the address approximately simultaneously? He can set a delay. If he sets a 10 minute delay, the system will broadcast a message only once in 10 minutes. If four patrol cars are driving near each other, the first (to come within 500 feet of 1234 Liberty) will hear the message. Then, the computer system will wait ten minutes without broadcasting (10 minutes of silence). The first car (to come within 500 feet of 1234 Liberty after the ten-minute delay) will hear the message. Then again there will be ten minutes with no message broadcast. The supervisor may set any delay he wants (for example: none, 1 minute, 20 hours). By the way, delay can be used to reduce how many cars hear the broadcast. Consider what happens if, after SR is set up for 1234 Liberty, 40 cars daily patrol past 1234 Liberty. The supervisor could reduce the distance (from 500 feet to 250 feet, for example), increase the delay (from 10 minutes to 30 minutes, for example), or both. The likely result of all three techniques is that fewer than 40 cars would patrol past 1234 Liberty.

          By the way, the response does not have to be to patrol past an address. The response may be any order the supervisor wants to give. Here are a few examples.

          1. "Low priority. Patrol around the block that 1234 Liberty Street is on. 1234 Liberty."
          2. "Low priority. Patrol around the block across the street from 1234 Liberty Street."
          3. "Low priority. Park your car on the sidewalk in front of the main entrance of 1234 Liberty Street. Wait one minute. Then resume patrol. I repeat. Park your car on the sidewalk in front of the main entrance of 1234 Liberty Street. Wait one minute. Then resume patrol. " < LI>"Low priority. Walk into the lobby of 1234 Liberty Street. Look around for 30 seconds. Then go back to your car and resume patrol. I repeat. Walk into the lobby of 1234 Liberty Street. Look around for 30 seconds. Then go back to your car and resume patrol. 1234 Liberty."
          4. "Low priority. Ask the security staff at 1234 Liberty Street to let you onto the roof. Stand on the roof looking across the street for half an hour. Then go back to your car and resume patrol. I repeat. Ask the security staff at 1234 Liberty Street to let you onto the roof. Stand on the roof looking across the street for half an hour. Then go back to your car and resume patrol."
          5. "Low priority. Park your car across the street from 1234 Liberty Street. Watch the main entrance of 1234 Liberty Street for five minutes, then resume patrol.. I repeat (and repetition)."
          6. "Low priority. Patrol from 1200 Liberty Street to 9600 Liberty Street: 1200 Liberty to 9600 Liberty."
          7. "Do not drive past 1234 Liberty Street. Do not drive past 1234 Liberty Street." This message could be broadcast without delay. Every car coming within 500 feet of 1234 Liberty would hear the message. This is a negative response, a stay-away order. maybe the stay-away order could be sent to all marked vehicles. Maybe only unmarked vehicles could be allowed to drive past that address.

          A supervisor could want several orders sent as a response. He would have to tell the computer in which sequence the orders should be sent. For example, the first car within 500 feet gets message 1, the second car gets message 2, the third car gets message 1, the fourth car gets message 3, the fifth car gets message 1, the sixth car gets message 4, then start over. That would be a multi-order response.

       2. Teaching the computer about the outside world

          Consider a response to an aspect of the outside world which cannot be predicted with precision. For example, if it's hot and sunny with no rain, a patrol supervisor might want much patrol at the beach. We emphasize that there are other aspects of the outside world (the world outside the computer) which affect how the supervisor wants patrol done, aspects other than weather. Responding to great beach weather is just one example. In any event, how does the computer know the weather?

          A simple way to know the current weather is to look at the Web. For example, the current weather (as we write this sentence) for zip code 12345 is at TODAY'S WEATHER for Schenectady, NY (12345). As we write, the page says:
          

          35 degrees F Feels Like 24 degrees F [in other words, cold]
          "UV Index: 1 Low" ["1 low" means not sunny]
          "Humidity: 64%" [not raining; the humidity is 100% during rain]

          
          It's cold, not sunny, and not raining. There's probably not much outdoor swim ming or sunbathing in zip code 12345 right now. A computer script or program can automatically go to that Web page often, copy the key facts ("feels like" temperature, UV code, humidity), then send those facts to the APC (automatic patrol change) computer (the computer which notices stimuli and then sends orders to appropriate patrol cars to do responses). When the computer sees that "feels like" temperature and UV index are low, or that humidity is 100%, the computer does not deploy patrol cars as if there are large crowds of outdoor swimmers and sunbathers. The patrol supervisor might set the stimuli to all of the following: "feels like" temperature at least 85 degrees, UV index at least medium, humidity below 100. In each locality, a patrol supervisor would learn what settings mean that many people are swimming or sunbathing outdoors.

          The weather Web page we're looking at says that it was updated at "12:00 p.m. ET" (noon, eastern time). Other weather Web pages may have been updated more recently than the one we're looking at. Maybe there are RSS feeds or other feeds that were updated more recently than the Web page we're looking at. The department's computer system can automatically know the weather outside although there may be a lag of a few hours. In other words, the computer might know what the current weather is or what the weather was an hour or so ago.

          Maybe there automatically might be different patrol of a city on a hot summer night and on a cool summer night. The APC computer should automatically know the temperature outside.

          Heavy traffic or light traffic may influence where patrol cars should be. For some cities and counties, there may be Web pages which supply current information about local traffic conditions on various highways and streets. A local traffic department's computer system might have current traffic information that could frequently, automatically be transferred to a local police department's computer system.

          There are many Web pages with a variety of current conditions which patrol supervisors might want automatic responses to. A police department's computer can know about current conditions in the outside world although, for some conditions, there may be a lag of an hour or so.

          Thus, deployment of patrol cars can automatically be changed in response to weather, traffic, and other events and conditions.

       3. Too much patrol

          Liberty Park needs patrol when it's dark. Two cars should always patrol in the park when it's dark. Three is on the high side of okay. Four is too many. A patrol supervisor defines a rectangle which includes the park. The computer system continuously knows how many patrol cars (other than those going to a 911 destination) are in the rectangle. If there are at least three cars in the rectangle when it's dark outside, any patrol car (other than one going to a 911 destination) that enters the rectangle hears a message "There already are enough cars patrolling the park."

          Liberty city's waterfront area needs patrol. A patrol supervisor specifies a rectangle that includes the waterfront area. Two cars should always patrol the waterfront when it's dark. Three is on the high side of okay. Four is too many. One is too few. One night when it's dark out, three cars are patrolling the park (on the high side of okay) but only one car is patrolling the waterfront area (too few). The computer knows that the park is on the high side of okay but the waterfront area is underpatrolled. A fourth patrol car enters the park rectangle defined by a patrol supervisor. As soon as the car enters, its officers hear a message, "There already are enough cars patrolling the park. The waterfront area needs a patrol car." We just provided a simple example of the APC computer automatically informing a patrol car about a place which needs a patrol car. This concept, of the computer system automatically sending messages which tend to shift patrol cars to high-need areas, could be complicated. A city or county could have 50 arrears which are continuously monitored to try to get optimum distribution of patrol cars. The area near county hospital might have an extra car and the area near a housing project might need a car. City hall might have an extra patrol car and a tourist area might need a car.

          What if more than one area needs another patrol car but only one extra car is available? Which area gets the car? One possibility might be to send the the extra car to the nearest area that needs a car, so as little time as possible is wasted while the patrol car is in transit. Another possibility is to assign priorities to different areas (for example, the housing project area has priority over the tourist area).

          Another way to assign extra cars might depend on the background of the officers in the car. The computer could know about the officers in each car. For example, a police department could take into account work an officer has done for the department. To give a specific example, if an officer in the extra car used to work for the department in the housing project, the department might decide to send the car to the housing project area. Maybe one of the officers in the car lives in the tourist area. Maybe he therefore knows that area better than the other areas. Maybe, if the car patrols the tourist area, he'll know where it should go and what to look for. Maybe the car should be sent to the tourist area because, with him in it, the car will patrol more effectively there than anywhere else. If he lived in the housing project until three years ago (in other words, recently), maybe he knows that area well. If he graduated from high school five years ago (in other words, recently), and if his high school was across the street from the housing project, maybe he knows that area well and therefore maybe the car he's in should patrol the housing area. This approach is based on the idea that patrol will be most effective in an area that the officers know.

          A police department might assign cars randomly. If two areas need a car, the car is assigned randomly to one of the areas.

          Thus there are at least four ways to decide where to send an extra patrol car when more than one area needs the car: closest receiving area gets the car, receiving area highest in priority gets the car, the car is sent to the area best known by the officers in the car, the car is assigned randomly.

          The different systems could be combined. For example, a patrol car could be sent to where the officers in it are likely to be most effective. If they are likely to be equally effective in all places that need another patrol car, then the car could be sent to the nearest place. In the unlikely event that two places are equally near, the computer could randomly choose a place to send the car.
          

       4. Varying value

          The APC system should work best when patrol officers have time to obey low priority orders. The APC system should work worst when officers are busy; for example, on Friday and Saturday nights.
          

       5. SR variety

          The response can be long after the stimulus and far away. For example, at 12:17 a.m. on 2 June 2009, a patrol car comes within 500 feet of 1234 Liberty Street, which is in the southeast of a county. As a result, at 12:17 a.m. on 2 December 2009 (6 months later), the patrol car nearest Liberty Bridge (which is in the northwest of the county, several miles away from 1234 Liberty Street) is ordered to patrol across the bridge.

          The response can involve the simultaneous action of more than one patrol car. For example, a patrol car comes within 500 feet of 1234 Liberty Street. Simultaneously, the APC computer orders that car (car 1) to patrol around the block 1234 Liberty Street is on, orders the car nearest 1234 Liberty Street (but not car 1) to patrol around the block across the street from 1234 Liberty Street (that car is car 2), and orders the car second nearest to 1234 Liberty Street (but not car 1 or car 2) to briefly park on the sidewalk in front of 1234 Liberty Street. Remember that these are low priority orders. If all three patrol cars do as ordered, three cars will approximately simultaneously respond to the stimulus (which was car 1 coming within 500 feet of 1234 Liberty Street).

          Responses can influence each other. There is a parking lot which is always empty in the middle of the night. Car 1 comes within 500 feet of the parking lot in the middle of t he night. It is ordered to go the lot and wait in it for an hour. If another patrol car (car 2) enters the lot during that hour, car 1 should follow that car (car 2) for two hours. Then, the car nearest the lot (but not car 1) is told that another patrol car soon may follow it for a while, and to drive in a way that lets that other patrol car follow. Car 2 then is ordered to patrol into the parking lot, drive in a circle, then resume normal patrol. If both cars do as ordered, car 1 will follow car 2 for 2 hours. The APC computer could have ten patrol cars waiting in the lot, with the result that car 2 leads an eleven-car convoy for hours.

          A response can include an officer providing information to another officer. For example, car 1 goes within 1000 feet of Liberty Bridge. The car is ordered to drive back and forth across the bridge for 3 hours unless the car is told before then to do something different. When car 1 stops (either because the 3 hours are up, or because it was told to do something different and then it did do something different), car 1 should resume normal patrol. Also, car 1 is told that, if he sees another patrol car driving back and forth across the bridge , he should tell the other car that it should stop. Half an hour passes. Then, a patrol car (patrol car 2) goes within 1000 feet of the bridge. Car 2 gets the same order car 1 got (drive back and forth across the bridge, etc.). Every half hour, the APC computer looks for another patrol car to relieve the car driving back and forth across the bridge.

          The examples above, some of which are unlikely, illustrate complicated SRs. Almost always, SR would be simple: the response is at about the same time and place as the stimulus, and involves the same car and no other car. However, a patrol supervisor can create complicated, multi-car, long-term SRs.

   12. Batch processing 911 work 2007aug26
       1. Introduction

          Elsewhere in this introduction, we suggest how to decide which police officer to send to a 911 emergency destination. When there's one 911 destination to be serviced and more than one officer available, it's appropriate to think about which officer to send. Sometimes (for example, on a Friday or Saturday night), the opposite problem exists: deciding which destination to send an officer to. When many 911 calls come in, the dispatcher has to decide which 911 destination to send an officer to first. Consider one officer available for dispatch and six, possible 911 destinations. All six destinations are equally urgent. The officer is equally qualified to handle all of them. For example, he hasn't previously been at one of the destinations, which might make him especially suitable to handle it again. In that case (six equally urgent destinations with the officer equally qualified for all six), a dispatcher might ask the officer to go to the nearest destination. We want to suggest asking him to go to a destination which is not necessarily the the nearest.

       2. explanation

          Consider 911 emergency destination A, 20 blocks west of the officer; destination B, 30 blocks west; C, 40 blocks west; and destinations D, E, and F, all on the same block, 21 blocks east of the officer. Destinations D, E, and F are on three different streets but on the same block.

          If one asks how the officer can do his next 911 assignment as soon as possible, the answer is to ask him to go to destination A. It's the closest so it may require the least time to get there. If one asks how the officer can do his next three 911 assignments as soon as possible, the answer is to ask him to go to destination D because D, E, and F are on the same block. The total driving for the next three assignments probably will be least if he does D, E, and F. If one wants him to finish his next three 911 destinations as soon as possible, he should be asked to go to destination D, because he can do E soon afterward D, and F soon after E. Therefore, he should go to destination D even though destination A is closer.

          When the dispatcher starts talking to the officer, the 911 computer system already knows where the officer is. The computer automatically identifies the five 911 destinations nearest the officer, the five destinations nearest each of those five (a total of 25 pairs of destinations), and the five destinations nearest each of those 25 (a total of 125 triads of destinations). The computer calculates the total distance for each of the 125 triads. The computer knows which triad is shortest, second shortest, and so on. On the dispatcher's video monitor, the computer displays the shortest triad for that officer (shortest total driving distance), the nearest 911 destination (because he may not want to do a triad, a possibility discussed below), and nearby 911 destinations for which he is especially suited. If there is nothing in the last category (nearby 911 destinations for which he is especially suited), the dispatcher asks him to go to the first destination of the triad displayed on theTh. video monitor (triad with shortest total driving distance), explaining that it's the first destination of a triad. The officer knows that the destination is part of a triad of destinations when he decides whether he'll go. When he finishes that first first destination, he notifies the 911 computer system with a phone call (described elsewhere in this introduction), then waits in his patrol car. He does not patrol far away. If he patrols, he patrols near the destination address. The time saving caused by batch processing results from his next 911 assignment being near the one he just did.

          Consider an officer who plans, after he finishes the next 911 assignment he expects to get, to have lunch or end his shift. It's Saturday night and he knows that his department does batch processing on Saturday night and that he's one of the officers who gets batch work. The dispatcher calls him to tell him about a 911 assignment. He has two alternatives:

          1. He immediately tells the dispatcher that the next 911 assignment should not be part of a batch.
          2. He's asked to go to a 911 destination and does. After he finishes that 911 assignment, he calls the 911 computer system to notify it that he finished, then he drives away to have lunch or end his shift. As soon as he drives away, the 911 computer system will realize that he's not at the first destination. The remaining destinations will automatically become available to other officers.

          Let us define "trip time" as the time between start (when an officer says to a dispatcher that he will do a 911 assignment that the dispatcher suggested) and end (when the officer's patrol car arrives at he 911 destination). By having dispatchers plan three destinations at a time instead of just one, average trip time should drop when batch processing is done (for example, Friday and Saturday nights). Planning ahead saves time. We guess that the total number of 911 assignments done would rise a little because less time would be consumed by going to 911 destinations.

       3. Conclusion

          We think that batch processing of 911 work would substantially reduce a police department's average trip time. In the explanation above, we refer to triads. Why plan three trips at a time, not two or four, for example? The best number of destinations per batch would be discovered through experimentation. By measuring the effect on trip time, one would discover the best number of destinations per batch.

          Having a computer automatically spot groups of nearby 911 destinations, then giving those destinations to an officer to save officers' time going from destination to destination, is a fundamentally sound idea, we think.

       
       
   13. 2008oct5
       1. Introduction

          Consider an officer in a patrol car. He wants to go to 1234 Liberty Street to respond to a 911 call but he has no idea where that address is. The dispatcher provides some information; for example: names of the nearest cross streets, and the fact that the destination address is two blocks north of County Hospital. This means nothing to the officer, a new officer who has never lived or worked in Liberty City. Even people who drive much in that city have trouble finding Liberty Street. There are a few ways to handle this. For example, the dispatcher could talk the officer all the way to the destination.

          The ideal city has a user-friendly address system. In the ideal city, every address is easy to find without a map. One way to do this is to build all the streets in a city systematically in a rectangular grid. Every street can be consecutively numbered, and every building can be consecutively numbered. This system exists in some Utah towns. Every address is easy to find. There is no need for a street map to drive to an address. Before one starts driving in those towns, one can visualize where the destination street is and even where the destination building number is.

          Before one starts driving in New York City's Manhattan, one can visualize where most destination streets are and where some building numbers are. Someone in Manhattan can visualize where 3rd Avenue is although he might not know where 500 3rd Avenue is. The Utah and Manhattan systems both require a systematic grid of consecutively numbered streets.

          A problem arises if streets are laid out willy-nilly. Furthermore, even if streets constitute a rectangular grid, a problem arises if their names are unsystematic. Liberty Street may be a prestigious street. Buyers may have paid a premium for a Liberty Street address. It may not be politically possible to tell everyone on Liberty Street that they now are on 47th Avenue. Even if the street isn't prestigious, there may be other objections to a new name. For example, someone may have much money invested in a restaurant named Liberty Street Deli. He may want the street to stay Liberty Street. Some people are simply attached to the name of their street and therefore don't want it to change.

          Even if streets are not systematically numbered, their names may provide useful information. For example, in Manhattan, numbered streets named Street generally run east-west and numbered streets named Avenue generally run north-south. Fifth Street runs east-west and Fifth Avenue runs north-south. One might name streets to tell many other traits, not just direction. However, in many places, there would be resistance to name change.

          Meanwhile the burglary in progress at 1234 Liberty Street continues and the officer doesn't have a clue in which direction to drive. He shouldn't have to get out a street map, reconstruct the city's streets so that they are a grid, or rename the streets (for example, so that they are consecutively numbered).

       2. Intersections, E.N. Numbers

          On a map, draw a four-sided polygon with two north-south lines and two east-west lines. The north-south lines should be lines of longitude. The east-west lines should be lines of latitude.

          The polygon is the smallest one which includes all of Liberty City. The two east-west lines are parallel. The two north-south lines are not quite parallel. Those two lines, if extended far enough, will converge at the North Pole and South Pole. The polygon is a trapezoid which is almost a rectangle. In the northern hemisphere, the north, east-west line is slightly shorter than the south, east-west line.

          Every intersection in the city has an east number and a north number. For example, an intersection might be East 123, North 456. This is written E123 N456. If an intersection is at the east edge of the city, it is at E999 (highly east). If it's at the west edge, it's E001 (hardly east). An intersection at the northern edge is at N999 (highly north). An intersection at the southern edge is at N001 (hardly north). Every intersection in the city can be described this way. If an intersection were in the exact center of the city's trapezoid, it would be E500 N500. An intersection at the extreme southeast is E999 N001 (highly east, hardly north). The example we gave earlier in this discussion (E123 N456) is on the west side of town (the E number has to be above 500 to be on the east side). That intersection is, on a north-south dimension, slightly south of center. (N500 is center, N numbers below 500 are in the southern half, this intersection is almost 500 so it's almost in the center north-south speaking.)

          East-west lines are equidistant. For example, the distance from N345 to N346 is equal to the distance from N700 to N701. North-south lines are equidistant. For example, the distance from E672 to E673 is equal to the distance from E205 to E206. The 999 east-west lines (N001 to N999) intersect the 999 north-south lines (E001 to E999). For example, E123 intersects N456. That's a linear intersection because two lines intersect. There are almost a million, linear intersections (999 times 999). A street intersection is an intersection of two streets. No city has a million street intersections, we think. Few cities have over 50,000 street intersections. Each street intersection is described by the linear intersection it's closest to. For example, if a street intersection is closer to linear intersection E123 N456 than to any other linear intersection, the street intersection is designated E123 N456. Because there are far more linear intersections than street intersections, most linear intersections will have no, corresponding street intersection.

          Linear intersections are defined by latitude and longitude. To find which linear intersection a street intersection is closest to, one needs to know the latitude and longitude (lat-lon) of that street intersection. This is already known for many intersections. If it is not known for some street intersections, someone goes to the intersection with a GPS device to find the lat-lon there.

          At every street intersection in the city, there are signs with the E and N numbers. For example, a street intersection might have signs saying:

          E123 N456

          If a driver knows the E and N numbers of a street intersection, he can easily drive there.

          The police dispatcher, using his computer terminal, finds and then tells the officer the E and N numbers of the Liberty Street intersection closest to 1234 Liberty Street. For example, the dispatcher says "1234 Liberty Street. Do you want the E.N.?". If the officer says that he does (which he might if he doesn't know where 1234 Liberty Street is), the dispatcher says "E628 N792". It's easy for the officer to find that intersection.

          It may not be necessary to have E.N. signs in every intersection. For example, consider a city in which it's easy to find addresses except in the southwest of the city. The city government might post E.N. signs in intersections in the southwest of the city only.

          In the system described above, the traveler goes from intersection to intersection, not from street to street, to get to his destination intersection. The intersections are numbered using a pair of short decimal-free numbers; for example, E123 N456.

          Police can read. However, an illiterate person could use this system to go to any intersection in the city if he knows the numbers from 001 to 999.

          The E.N. numbers on street signs do not need leading zeroes. For example, on a street sign, E7 N345 would be easier for many people than E007 N345.

       3. Lion

          We call this intersection system the E.N. system or Lion (local intersections orthogonally numbered).

       4. Comments
          1. Other Users

             The system above is designed for police and similar, emergency use. Other people could use it. For example, consider someone at an address, 1234 Liberty Street, which he knows is difficult to find. He calls a taxi company by phone and asks the dispatcher for a taxi to pick him up at 1234 Liberty Street. If he knows that taxi drivers have trouble finding his address, he might mention to the dispatcher that the nearest Liberty Street intersection is E123 N456. Anyone can find the two E.N. numbers at the intersection nearest his home by going to that intersection and reading the E.N. sign, then tell the E.N. numbers to those people who would appreciate knowing (such as a taxi company's dispatcher dispatching a taxi to the address). If a business were difficult to find, the business could put its E.N. in its advertising and on its stationery.

             Consider the following text written on a piece of paper:

             E773 N456
             1234
             E774 N454
             

             There are two intersections there. On the street connecting them, there is a building with building number 1234. Someone in a different part of the city can go to that building without even knowing the name of the street it's on. First, he finds one street intersection. Second, he finds the other street intersection. Third, he finds building number 1234 between the two intersections. He can get to 1234 Liberty Street without knowing that it's on Liberty Street. An illiterate person can apply for a job or go to a doctor even if he arrived in Liberty City the day before. There are two types of illiterate people. Some people are illiterate in every language. They cannot read. Some people are illiterate in the language used in Liberty City's street signs. They can expertly read Arabic, Chinese, Farsi, Greek, Hebrew, Japanese, Korean, Thai, and Urdu but that doesn't help when trying to read Liberty City street signs or a street map of Liberty City. Thus, this system can be used by illiterate people as an address system if they know the numbers 001 through 999.
          2. East and North

             In the system described above, each street intersection has two numbers, the east and north numbers. On signs, they don't need to be labeled east and north (or E and N). For example, there could be a * number and a % number (a star number and a percent number). The system works just as well regardless of whether the traveler knows that one number tells how far east the intersection is and the other number tells how far north. A traveler (for example, a police officer) could be asked to go to intersection *123 %456 (pronounced star 123, percent 456). He would, at intersections, see signs with a star number and a percent number. The traveler merely has to match two pairs of numbers: the pair he was given (in other words, the destination pair) and the pair on the nearest sign. He doesn't need to know what they mean. It's not bad to tell travelers that the two numbers mean east and north. However, east and north labels are meaningless to a driver who got to the city the day before because he does not know where east and north are. East and north would sometimes be useful clues to someone who knows the city well. For example, E972 is at the far east of the city, which is a useful clue for someone who knows where the far east of the city is.

             In summary, it is not necessary to tell drivers which number describes east and which describes north; however, that information sometimes would be useful to drivers who know the city well. That's why we suggest E123 N456 rather than *123 %456, for example.

          3. Three Digits

             In the examples above, the E and N numbers each are three digits; for example, E123 and N456. One merely needs enough digits to find intersections. Consider a city that has about 40 intersections from east to west. It would need only two digits for its E number; for example, E12. With two digits, the highest E number would be E99. The city would have 99, equidistant, north-south lines: E01 to E99. That's sufficient for a city with about 40 intersections from east to west. A street intersection there might be, for example, E72 N582.

          4. Close Intersections

             Consider two intersections which are extremely close to each other. Liberty Street and River Road intersection (intersection 1) is about ten feet from Prosperity Avenue and Valley Lane (intersection 2). Both intersections are E123 N456. The officer wants 1234 Liberty Street. The nearest Liberty Street intersection is intersection 1, so he was told E123 N456. When he gets there, he sees two intersections. Then he may have to look at both to find Liberty Street. Often, the first intersection he gets to will be the one he wants. If not, he should look at the other intersection. What if he gets there in the middle of the night when it's raining heavily, only notices intersection 2, sees that it is E123 N456, but can't find Liberty Street? He should look around for another intersection nearby.

             We guess that this problem (two or more intersections with the same E.N. number) will happen rarely. Also, we guess that, when the problem does happen, it will be easy for officers to handle. However, the problem can be prevented.

             If the destination address is 1234 Liberty Street, the computer system shouldn't direct the officer to go to E123 N456. When possible, the computer system should avoid an E.N. number with more than one intersection there. If possible, the computer system should provide a Liberty Street intersection which does not share an E.N. number with a different intersection. Maybe there's a nearby Liberty Street intersection at E128 N453. That intersection is the sole intersection which has that E.N. number. The computer system should direct the officer to E128 N453. When he gets there, he'll get on Liberty Street. He'll drive on Liberty Street until he gets to 1234 Liberty Street. The computer system should be programmed to avoid providing E.N. numbers with more than one intersection when possible. The dispatcher enters "1234 Liberty St", then a Liberty Street E.N. number near 1234 Liberty Street appears on the dispatcher's video monitor. The system tries to provide an E.N. number which has one intersection only.

             The software is written to, whenever possible, send every officer to the nearest street intersection with a unique E.N. number. That way, when the officer finds an intersection with tstreet intersection he wants. In this approach, the driver does not seek a street intersection which has the same E.N. number as a different street intersection. The problem won't occur often and is easy for officers to handle, we guess. Nevertheless, the problem can can be prevented.

             There are other ways to handle this problem.

             One way is to paint a red line on the E.N. signs in problem intersections (assuming that there are any problem intersections). As soon as the officer sees red on the sign, he knows that there's at least one other intersection nearby with the same E.N. number. If the intersection he's at doesn't have Liberty Street, he knows that there's another intersection nearby which does.

             One way to prevent the problem is to have more digits in E.N. numbers; for example, four digits for each number. If an intersection number is E1234 N4567, for example, then there will be fewer intersections with the same E.N. number. If there are four digits in the E number and four in the N number, there are almost one hundred million linear intersections in the city's trapezoid (9999 times 9999). There will be few, linear intersections with more than one street intersection. One should be reluctant to use this four-digit system. Our intersection-numbering system should be as user-friendly as possible. Many people don't like long numbers. We suggest handling this problem without four-digit E.N. numbers.

          5. traffic congestion

             The E.N. system is designed to help emergency responders quickly get to a destination such as 1234 Liberty Street even if they don't know where it is. We think that this system would help other people, too.

             As you drive somewhere, traffic may move slowly. This is partly because, two cars ahead of you, an ignorant driver is looking for Liberty Street. Every time he approaches an intersection, you and others behind him are delayed because he drives slowly, carefully looking for a "LIBERTY ST" sign. He thinks Liberty Street is somewhere near. He may even stop his car in the street to ask a pedestrian where Liberty Street is. You and the others behind the slow driver would get to your destinations faster if Liberty City became easier to navigate; for example, by posting E.N. signs at street intersections (and by systematically renaming streets to the extent that that's feasible).

             Traffic congestion is a big problem in many places. The normal solution is to increase transportation capacity by building more roads, bridges, tunnels, subway lines, and similar, structures, and by buying more buses. This can be slow, not just expensive. At any given time, many cars on the road are there because the driver is taking a long cut (not a short cut): an unnecessarily long route. There are many drivers who discover a route and then keep using it merely because it works. It's not the fastest route but those drivers, because they have trouble driving through Liberty City even with a map, use the route they know. The driver expertly knows his route but it's a route which keeps him on the road an unnecessarily long time. We guess that, at any time, a few percent, perhaps even several percent, of the vehicles on the raod are there because: some drivers struggle to find their destinations, other drivers fluently use an unnecessarily long route. To reduce traffic congestion, one can either reduce the number of vehicles on the road by making addresses easier to find (for example, with the E.N. system), or one can increase transportation capacity (which is expensive to do).

             Thus, traffic congestion partly is caused by a pair of related phenomena: some drivers struggle to find their destinations, other drivers fluently use an unnecessarily long route. These two phenomena are more likely to afflict cities with many newcomers and many difficult-to-find streets. Newcomers may have two problems: ignorance of where streets are, inability to read simple terms such as "LIBERTY ST".

             The E.N. system is presented here to help emergency responders quickly find destinations such as 1234 Liberty Street. We think that the system would also reduce traffic congestion. The E.N. system like most of the other ideas on this page, would work in many countries, not just America.
             

          6. CompStat
             2010feb23
             1. General

                This page is not about CompStat. If you're considering ideas on this page, we suggest that you also consider CompStat. In essence, CompStat is database software which provides simple statistics about local crime-related events. The statistics are available by place (for example: by precinct or by street address), by kind of event (for example, arrest), and by kind of crime (for example, homicide). CompStat does not make maps and therefore is not mapping software although mapping software (for example, GIS software) can make maps based on CompStat data, as far as we know. CompStat generates basic statistics. For example, one can find a sum (for example, the total number of complaints in a precinct in a week) and a percentage (the sum for one week divided by the comparable sum for another week). As far as we know: if you were an excellent pupil in primary school but left school the day before you would have entered high school, you know all the arithmetic required to understand CompStat statistics. It is a waste of everyone's time to provide statistics which the intended audience does not understand. Many expertly run, non-police organizations have far more sophisticated, useful statistics. In a police context, CompStat and similar software is considered advanced (because Compstat usually is better than nothing, which is what Compstat replaced).

                Above we briefly described the essence of CompStat. CompStat reports also include basic, demographic information for the places involved. It is common, in police departments which use CompStat (or some similar program), to regularly discuss CompStat statistics and closely related information. Crime-reduction techniques often rely at least partly on CompStat statistics, in departments which use CompStat.

                CompStat is not a philosophy. Philosophers grapple with profound questions such as what is causation, knowledge, virtue, and life. CompStat is relatively superficial. At a CompStat discussion, a police executive might discuss homicide statistics but he wouldn't discuss what life is.

                Definitions of CompStat vary. It and comparable software is used in different form in different places. Furthermore, CompStat and similar software evolved. Originally, it was database software which provided simple statistics about local crime-related events. (In our opinion, that is still CompStat's essence.) Furthermore, some people use "CompStat" narrowly to refer merely to the reports, especially to the statistics in the reports, but other people use the term broadly to also include the full set of police responses to the report.

                We will now give an crudely oversimplified example example of one way that Compstat might be used. A Compstat report might list the highest-crime addresses in a city one month. The police department might try to drive down crime at those addresses. Crime statistics for a subsequent month might show much less crime at those addresses. This reduction would be considered a success attributable to Compstat. Compstat requires true statistics. If the statistics provided to Compstat are wrong for any reason, Compstat will not provide a benefit (for example, Compstat will not know which addresses and precincts really are highest in crime, and whether crime there went up or down). To give another example: crime really may go up, but incorrect crime statistics provided to Compstat may make Compstat report that crime went down. Police departments, regardless of whether they use Compstat, sometimes hide crime in their jurisdictions (sometimes by hiding entire crimes, sometimes by downgrading a crime to make it seem less serious than it really is).

             2. Compstat in New York City
                1. Quality of Statistics

                   According to our understanding of Leonard Levitt's 18 September 2006 article, "CCRB: Dead Board Walking", in the nypdconfidential.com Web site, statistics describing crime during the Bloomberg administration may be based on NYPD lies. The Mayor's Commission to Combat Police Corruption is a city government agency. The Commission's chairman was former federal prosecutor Mark Pomerantz. Pomerantz sought NYPD records following reports by two unions of police officers that NYPD commanders were downgrading crimes from felonies to misdemeanors to create misleading crime statistics. As far as we know, this is a kind of corruption. Kelly illegally refused to release the records, wrongly saying it was none of the commission's business. Thus there was a conflict between Pomerzntz and Kelly. In this conflict, Bloomberg said and did nothing, which helped Kelly. Pomerantz quit.

                   According to our understanding of < a HREF="http://nypdconfidential.com/columns/2010/100215.html" TARGET="_BLANK" TITLE="Brooklyn crime statistics">"Two Schoolcraft Mysteries." (Leonard Levitt, 15 February 2010), Adrian Schoolcraft supposedly reported downgrading crimes, for the purpose of creating false statistics about crime, by the NYPD (New York Police Department) in Brooklyn's 81st precinct (as a result of which he supposedly was illegally punished). The 81st precinct is Bed Stuy, a high-crime area.

                   According to "CHECKING THOSE STATS." (Leonard Levitt, 8 March 2010): PBA delegate Frank Palestro claimed supervisors were downgrading crimes at the 42nd precinct in the Bronx, and police officer Adil Polanco made similar claims about his bosses in the 41st precinct.

                   In "Retired Officers Raise Questions on Crime Data" (William K. Rashbaum, 6 February 2010, New York Region section, New York Times), the NYPD is correctly accused of unethically creating false statistics about crime (in part, incidentally, to make Compstat statistics that will make the city look good). Incidentally, we guess that the same unions which gave information to Pomerantz, arranged to have criminologists do the research which Rashbaum reported.

                2. Consequences of Bad Statistics

                   If a sincere person contends that Compstat worked well in New York City in modern times (for example, 1999-2010), either he does not understand how Compstat works or he does not know that New York City's Compstat uses false statistics.

                   We guess that it's a state felony to try to induce someone to write a misleading crime report, and a federal crime for the city government to knowingly supply false crime statistics to the FBI. The FBI's crime statistics report for New York City for the last several years should be blank (perhaps with a note that there are no statistics available, and with further information available by telephone that no statistics are available because the city government has been concealing the true amount of crime for years to make the city look good). Criminologists and others use FBI crime statistics. Those users should not be misled by New York City crime statistics supplied by the FBI.

                   Consider someone who moved to New York City because he was tricked by the NYPD's fraudulent statistics about crime. He was then the victim of crime in New York City. Can he successfully sue the city government for his moving expenses or crime expenses?

                   We do not know of any jurisdiction which measures the external validity of its crime reports or crime statistics. This makes it easy to lie that there is less crime than there is.

                   There is a question of how the NYPD deploys its resources (for example, patrol officers and detectives) to work against crime. In the interest of brevity, we will crudely oversimplify. If there is more crime in precinct one than in precinct two, there probably should be more police resources (for example, patrol officers and detectives) in precinct one. By studying records of crime (and of police resources deployed against that crime), a police department can learn how to minimize crime. The NYPD intentionally keeps dishonest crime records. No one knows how much crime there really is in precinct one and in precinct two. Thus, it is impossible to productively deploy resources between the two precincts. The result is far more crime than the city would have if it kept honest records (and therefore could non-ignorantly deploy resources). Above in this paragraph, we provided an extremely simple version of the problem in New York City. There are many other ways that the NYPD's terrible records cripple the NYPD's work against crime. For example, there are is a questions of which work shifts to which to assign officers. If there is a long-term trend of crime growing on one shift in some precincts, it might be sensible to transfer resources (for example, patrol officers) to that shift in those precincts. Crime records might also be analyzed before deciding off days (non-working days) and vacations. Furthermore, some kinds of crime decline more in response to patrol and investigation than do other kinds of crime. There will be unnecessarily high crime if, as in New York City, police resources are not deployed in response to careful study of honest crime statistics.

                   It is important to understand the seriousness of a jurisdiction's crime. It would be wasteful to use detectives to investigate minor crimes, such as jaywalking. It is appropriate to use detectives to investigate serious crimes, such as murder. It would be wasteful to use a laboratory (for example, to match DNA) for minor crimes but it is appropriate for serious crimes. What percent of the NYPD's sworn officers should be detectives? It is impossible to answer that question because no one knows how much serious crime the city really has, because the NYPD has been falsifying its crime records for years. A reduction in serious crime in New York City should cause less money (than otherwise would be in the NYPD budget) for detectives and for crime laboratory work. The NYPD budget cannot sensibly respond to how serious the city's crimes are, because the NYPD avoids knowing.

                   Police departments are not the only government agencies which work against crime. Honest crime statistics might show that the best ways to drive crime lower chiefly involves agencies other than police; for example, to have more (or less) incarceration, and to have more (or fewer) probation and parole officers and prosecutors. Crime statistics are necessary to efficiently deploy resources among government agencies, not merely within police departments.

                   Above are some examples of how the NYPD's work is damaged by lack of honest statistics. The examples can easily be understood. (There also are sophisticated ways to use crime statistics, none of which are done by the NYPD.)

                   If the NYPD suddenly started keeping crime records honestly, there would seem to be a big increase in the amount and seriousness of crime.

                   It would be interesting to try to study how much crime New York City has had every day for the last decade (for example, how many burglaries, robberies, and rapes) because NYPD resources are not deployed in response to honest crime statistics. For example, if a crime is downgraded or is not reported at all, there may be no investigation by a detective, so the criminal may avoid arrest, so the criminal may do another crime. If there seems to be less crime in a precinct, it may get fewer patrol officers, so there may be less patrol, so there may be more street crime. How much crime resulted from the NYPD's false records?

             3. End

                We do not provide Compstat. Elsewhere on this page, we suggest a considerably more sophisticated set of programs.

       
       
       

       ---

2. The paragraphs above contain the mere introduction (Ipedmis) to our idea. We have not stated the idea yet.