Slashdot Mirror
The Increasing Role of Predictive Analysis In Police Work
elucido writes "A growing number of law enforcement agencies, in the US and elsewhere, have been adopting software tools with predictive analytics, based on algorithms that aim to predict crimes before they happen. From the article: 'Without some of the sci-fi gimmickry, police departments from Santa Cruz, California, to Memphis, Tennessee, and law enforcement agencies from Poland to Britain have adopted these new techniques.
The premise is simple: criminals follow patterns, and with software — the same kind that retailers like Wal-Mart and Amazon use to determine consumer purchasing trends — police can determine where the next crime will occur and sometimes prevent it.'"
When prediction goes, one could always predict that, given a time frame, something will definitely happen - such as plane crash
If TPTB is really interested in saving lives, they could have done more to predict future plane crashes and then do something to prevent it from happening
Muchas Gracias, Señor Edward Snowden !
I thought Tom Cruise tried that already and how that worked out, huh?
nobody predicted this would be the first post. Or the second.
Instead they predicted it would be the third.
I imagine patrol cops go where they expect some action may occur (or to stops that offer cheap food/drink for the uniformed). This sounds like a higher tech version of that, basically taking the instincts out of the equation and substituting it with statistics. Perhaps adds more coordination at the central office level too although I'm sure that also already occurs.
I mean, how would we know that no one has been thrown into a jail cell because of that ?
There are thousands upon thousands of arrests been done every-single-day and while most of those arrests are justified, I can't guarantee (and I do not think anyone can guarantee) that all the arrests had been done in a kosher manner
Muchas Gracias, Señor Edward Snowden !
It all boils down to statistical analysis
Let's say you own a grocery store, and there's one particular item that shoplifters like to steal
You, as the owner, can do one of three things -
A. Stop carrying that item in your store
B. Keep that thing close to the counter to cut down on shoplifting
C. Fix a hidden vid cam near where you put the thing and start recording
Muchas Gracias, Señor Edward Snowden !
Yes think of all the friendships and cards left in 2/3rd world nations as this is exported. Over years many would feel very happy around the nice people from the USA.
Training in the states, hardware upgrades, new deals and notes on dissident expats shared.
Domestic spying is now "Benign Information Gathering"
Thieves will learn to mix it up, such as tossing dice on a map.
Table-ized A.I.
Ok, so first, if the crime doesn't happen, how do you know you prevented it? Maybe it just didn't happen.
Second, doesn't this seem like there will now be a market for anti-prediction? That is, find out where the cops think the crime will occur, and do the crime somewhere else. Because the cops will be somewhere else, your chances of getting caught are less.
If you need web hosting, you could do worse than here
The software checks if person of interest holds a Facebook account. Voilà! If he or she doesn't, it should mean he/she will commit mass murder. Couldn't be easier, I guess...
I imagine patrol cops go where they expect some action may occur (or to stops that offer cheap food/drink for the uniformed). This sounds like a higher tech version of that, basically taking the instincts out of the equation and substituting it with statistics. Perhaps adds more coordination at the central office level too although I'm sure that also already occurs.
Technically a good cop with good instincts is applying statistics. The human brain is built to recognize patterns and to use those patterns to make predictions. Some of this is done at a subconscious level. So its not that we are necessarily introducing statistics, its seems more that we are using a much larger data set to mine patterns from. Still, as you say, a high tech version of what we already do.
You are assuming here, that thieves are actually smart enough to do so. Most common crime thieves simply lack the brain power to randomize their victims, M.O. and such and not pee their pants at the same time.
Unfortunately, society can't or won't afford catching smart thieves and most methods used by law enforcement in general only catch criminals that make obvious mistakes. The smarter thieves usually end up in politics, banking or the stock market.
I was promised a flying car. Where is my flying car?
If you want to catch a criminal, you need to think like a criminal
To be successful, you, as a criminal, must know your victim's vulnerability - either they are alone, weak, or they are far away from others' ear shot .... and ... this is important - the place you commit your crime must be familiar to you - or it wouldn't be so easy for you to get away - and in cases involving murder - you, as a criminal, must also know where to dump/bury/hide the body before you commit the murder
Many times crimes were solved because of the sheer sloppiness of the criminals
As for throwing dice on the map - I'm afraid it would not be easy - unless of course, the new location happens to be a familiar spot for the criminal
Muchas Gracias, Señor Edward Snowden !
Start by rounding up all those suspicious Facebook abstainers!
- For the complete works of Shakespeare: cat
Wasnt this an episode of Numb3rs?
I hear this is working pretty well for them. They've already discovered that people who don't use facebook are mass murderers in training. The real challenge is trying to figure out who these people are, what they look like, what they are doing, and how much gold they have in Farmville, because the software is currently only able to figure this out for people who are on facebook. This is also made more difficult by the fact that people who don't use facebook are more likely to be intelligent, self-aware, exhibit behavior that does not conform to the patterns of the pack and exhibit a phenomenon known as "free will," which wreaks havoc on their predictive models.
“When police departments are laying more sworn personnel, they can do more with less."
I would never have thought to try that. If you get more personnel laid, they can do more with less? Just think how much more productive programmers could be under such a system!
Well, if they are using software similar to what retailers use to 'predict purchasing trends', then all they are looking at is statistics of past crimes, looking at graphs that plot locations, seasons, times, types of crime against a database of past criminals, their ages, past crimes committed, other personal info, but how are they going to account for all other metrics? What about advertising? Whatever is advertised in the media at the time will have more sales. Also what about people without criminal records, are they looking at everybody?
Obviously they are not looking at every possible home invasion, every possible burglary, every possible theft and every possible murder. They probably have some prioritised locations and people they are interested in, banks, shipping lanes, famous people, who are they looking to protect?
Are they excluding entire classes of people from their software, like are all of the politicians excluded? All the people connected to the ruling class? Because if they want to prevent REAL crimes, mass murders, theft on huge scale, global scale even, they wouldn't have to go too far, they just have to look at the people in power, running the place.
You can't handle the truth.
...rather than the 48 hours to solve a crime (before the chief busts them down to traffic duty so fast their heads will spin), they have 48 minutes.
Unfortunately individual police officers were drawing from a much smaller pool of data which was then put through their personal biases. If an officer had a racial or cultural bias then they may perceive an area as having more crime, when the actual statistics don't match.
Please, Songdo is a nice city with excellent restaurants and some great bars, but the whole smart city thing is a little over-hyped.
I sense some disagreement in the form of moderation... though I'm not hearing arguments? Surely, if you can click, you can also type while eating, so do enlighten me.
Ok, so first, if the crime doesn't happen, how do you know you prevented it? Maybe it just didn't happen.
.
You don't look at individual crimes, you take a selection of areas with similar crime statistics, implement the prediction system in some of them, then see how the crime rates change.
Second, doesn't this seem like there will now be a market for anti-prediction? That is, find out where the cops think the crime will occur, and do the crime somewhere else. Because the cops will be somewhere else, your chances of getting caught are less
Perhaps there will be a market for anti-prediction, but the types of crimes that this aims to prevent (or even just be more response to calls about) aren't usually done by sophisticated criminals. Any anti-prediction system would first have to be able to aggregate crime statistics then apply the prediction algorithm, then find areas outside the predicted zones. If you have all that already, you might as well just sell the prediction algorithm to the police rather than make an unethical program that has a very small (and secretive) user base that wouldn't pay much for your system in the first place.
A police state is a state policed. We all live in one or more. Of course they'll do all they can to better police their state. It's their job, and we pay and vote for them to do it. Corporations license new technologies to them to do it better. Maybe if all citizens just aspired to be "officers" of the institution they'd be happier. Maybe... Just maybe, one should simply succumb and obey one's master.
Buy your next Linux PC at eightvirtues.com
But we did predict this post would be by an AC.
"The likes of Facebook and WhatsApp are free to those whose privacy is of zero value."
Ok, so first, if the crime doesn't happen, how do you know you prevented it? Maybe it just didn't happen.
You look not at one single crime but at the crime rate for a specific location and crime category. If the rate decreases after you start your prediction-based policing and the crime rate for this category does not increase in another area during the same time (interestingly this is one step proponents of public video surveillance very often happen to overlook), then your approach very likely has prevented such crimes in the targeted area.
Rudolf Hess edited Mein Kampf. He was the very first grammar nazi.
The problem is someone at some time will realise that "Predictive analysis" has a racial and religious "bias". People will ask "why are most police looking for muggers in black areas and terrorists in Muslim areas"? The answer - that this is where most of the crimes occur, is something that police have known for some time, but been instructed to ignore. Do you really think that the PC brigade will allow it back again because a computer says so? Now I know that in the case of blacks it is due to social deprivation, etc. - I am not blaming them - but it is a fact that they commit more crime.
balls in ur face. faggot.
I missed that - is it a new interrogation technique?
Because I enjoy intelligent and sexy company, duh!
There was a movie about this: http://www.imdb.com/title/tt0181689/
Wow. I listen to way too little hip hop worth shit, so thanks for that. The video is pure genius, too.
http://www.youtube.com/watch?v=MDb5BOage8k
http://www.youtube.com/watch?v=FKwgCGIQxf8
If you have technology to predict where crimes are going to take place, selling it to the police is going to be way, WAY more profitable than using it to commit crimes.
"Ok, so first, if the crime doesn't happen, how do you know you prevented it? Maybe it just didn't happen."
Exactly. Cops have no interest in preventing crime since they are evaluated by how they apprehend the criminals and how many of them.
They'll just use any advance knowledge they'll get, to be in position when the crime occurs to catch the criminals, not to force them to do another crime in a less predictable way.
Cunning
Those who are cunning are less likely to end up in jail
Those who are less cunning are more likely to end up in jail
Assuming both of them committed similar crimes
Muchas Gracias, Señor Edward Snowden !
In many places conspiring to commit a crime, *is* a crime.
Technically a good cop with good instincts is applying statistics. The human brain is built to recognize patterns and to use those patterns to make predictions. Some of this is done at a subconscious level. So its not that we are necessarily introducing statistics, its seems more that we are using a much larger data set to mine patterns from. Still, as you say, a high tech version of what we already do.
Very true. I cop friend of mine often gets asked "how did you know it was me?" and his answer is "because you always commit crime x by doing y."Ashe puts it, most criminals are stupid, or at least we only catch the stupid ones. This analysis just builds on what people's brain does naturally, with a more robust data set as you point out. Plus, it never forgets a crime.
I'm a consultant - I convert gibberish into cash-flow.
Smart criminals would be able to hack this system. However, most serious crimes are committed by people too stupid to find safer ways of making money.
-- Support a free market in the field of government
So they copied and automated an idea from "Idoru" by William Gibson. Colin Laney, a guy who earns his living by manually sifting thru data to find dirt on media personalities. He finds a woman who is about to commit suicide, if I recall it correctly.
Going on the assumption of the predictative accuracy of this technology what ethical imperative does the government have to reengineer these situations to make crime not tempting or not an option?
instead of just sitting around eating donuts and having rough gay sex in dark alleys...
Speeding tickets since "sarcasm on" speeding laws are about safety and not revenue enhancement "sarcasm off"
Did you know 80 to 90% of the moderators on slashdot wouldn't recognize a troll even if one dragged them under a bridge.
Provided it occurs in the richer enclaves where the Officer Bob bully routine is easily done, yes. But I suspect even in moderately-sized urban areas, there are implicit no-go zones for beat cops, lest they get got.
We call it statistical 'profiling' and it happens to be illegal apparently.
-=[ Who Is John Galt? ]=-
Depends on your definition of 'new system' I guess, but my proposal(short, rough, off the cuff, debate welcome):
1. No more pleading guilty
2. You still have prosecution and defense, but now you also have 'truthseeker', who's position is determining the truth, regardless of guilt or innocence.
3. Professional juries are an option - serve for something like a year, are actually paid real wages. Any given trial that's NOT close to capital case will have a mix of new and experiences jurists.
4. Sentences will not be done in terms of punishment, but in compensation and rehabitulation. Let's say you stole and wrecked a car as a stupid skillless high school teen dropout. If you're a dumbass in prison it could be a life sentence. If you learn a trade and work at it, you 'pay for' the car in a couple years and get out. If you're a gangbanger you're stuck inside until you're NOT a gangbanger(prisons will need to be seriously reformed to do this).
5. One idea was to have 'counselours' who 'bid' on prisoners - they get a % of the convict's wages and/or bonuses if the convict stays out of trouble for good amounts of time. The bidding system is so that 'high risk' convicts are worth more, that people don't simply get into bidding wars over the Michael Vicks.
I'd keep the death penalty for those deemed unreformable. The worst gang-bangers, serial killers, etc... I'll note that this is triage - the right person can work miracles with reforming the worst gangers and even religious extremists. But we need to help reform the most possible. Warehousing unredeemable prisoners with our resources simply isn't possible without ending up with even more unredeemable because you're supporting more. It's a vicious circle.
I don't read AC A human right
Sounds like a broken window fallacy to me.
Cops do get rated on their ability to make arrests, but there's so much crime out there in comparison to cops, that there is no benefit to cops in allowing it to happen if they can stop it. There may well be small-town places where crime is low, but usually that is dealt with by having lower numbers of police staff or even different sorts of "quotas".
Further, if they are moving their units to high-risk areas to only stop crimes in progress or that have been committed, that increases physical risk to cops themselves. They still have to apprehend the perpetrators, and the perpetrators can be armed and dangerous.
There certainly may be situations where individual cops might try to game the system, particularly if they don't mind trading off risk for potential rewards, but since statistics are generally used to set concentrations at the higher echelons of the department, individual goals like that are not likely to come into play.
One way or another, no police officer or administrator is going to complain if the crime rate is non-existent on his watch, because that is a statistic that they can use to justify their existence too.
Ok, so first, if the crime doesn't happen, how do you know you prevented it? Maybe it just didn't happen.
The real world is slightly complicated, so it really depends.
In some cases, the guy who went shopping with his shotgun in hand is a pretty good indicator, even if after noticing the nearby cops he just buys a beer.
In other cases, like property crimes, you can wait until a predicted crime happens and arrest the criminal on the spot.
In many cases, statistics will provide the answer - if you double patrols in some area and crime rate drops considerably, you can check for other effects (say, unemployment in the area dropping) and, after correcting for them, assume that what's left is at least partially caused by the patrols.
Assorted stuff I do sometimes: Lemuria.org
It was my understanding that 'frisk and/or harass the undesirables' has been a well-understood police work technique more or less as long as there has been such an institution... Is this 'predictive analysis' with its fancy computer machines and numbers and things actually a genuinely novel angle, or is it largely the process of paying IBM to build a model that provides an objective, scientific, reason you can give for doing what you are doing if anybody complains?
The case of New York comes to mind, a city that has both been on the leading edge of 'computerized, data-driven' modelling/performance stuff and rather tastelessly overt 'stop and frisk' approaches toward anybody that they don't like the look of...
"Ok, so first, if the crime doesn't happen, how do you know you prevented it? Maybe it just didn't happen."
Exactly. Cops have no interest in preventing crime since they are evaluated by how they apprehend the criminals and how many of them. They'll just use any advance knowledge they'll get, to be in position when the crime occurs to catch the criminals, not to force them to do another crime in a less predictable way.
Police are evaluated on the crime rates, and the percentage of violent crimes solved, the only other arrest that are made public is for DUI checkpoints. Crime rates are used to evaluate a city and their police force arrests are simply not used by the public when deciding on police bonds.
Knowledge = Power
P= W/t
t=Money
Money = Work/Knowledge so the less you know the more you make
Hell, it scares me already, and I don't even know what a "face faggot" is.
crime revolves around low education, social inequality, and cyclical poverty from low employment rates.
I can write that into perl, bash, python, or even C if you like, and i accept cash or check.
Good people go to bed earlier.
They were doing it with computers in New York City since about 1980. It's called Compstat. http://en.wikipedia.org/wiki/Compstat
Before that, they were doing it with maps and pushpins.
There's a similar situation in medicine. Atul Gawande had an article in the New Yorker on using a Compstat-style system for hospitals to find the patients who had the greatest need. http://www.newyorker.com/reporting/2011/01/24/110124fa_fact_gawande
The doctor Gawande interviewed said, “For all the stupid, expensive, predictive-modelling software that the big venders sell, you just ask the doctors, ‘Who are your most difficult patients?,’ and they can identify them.”
Fortunately, the software will be able to detect future attempts to unlawfully hack it.
That's what Miss Cleo told me when I called her, anyway.
Oh, for god's sake are we back to twelve-year-old jackasses trolling this site again?
One day I feel I'm ahead of the wheel / the next it's rolling over me / I can get back on / I can get back on
No, they've improved on an idea that New York City first tried in the mid 1970's, and that several other municipalities also flirted with in the late 1970's and early 80's. There's been fairly steady work on it ever since.
Walmart tracks every purchase by each customer. They know minute by minute when to expect something will sell. Why then, are there only 2 out of 35 registers open when I go to checkout at 6pm on a friday? :-/
Mr. Marks, by mandate of the District of Columbia Precrime Division, I'm placing you under arrest for the future phone hacking of Sarah Marks and Donald Dubin that was to take place today, April 22 at 0800 hours and four minutes.
-=- Many seek good nights and lose good days.
Ok, so first, if the crime doesn't happen, how do you know you prevented it? Maybe it just didn't happen.
The simplest answer is a Double Blind Study
I use a similar modeling technique to tune engines. Their model must have police presence as one of its input factors, since it's the only factor the police can control directly. Before they implement this system, they should perform a "sweep" of this input to establish a correlation.(i.e. vary the amount of police presence) Once the model is created, an optimization algorithm can be used to determine the most effective use of police resources. (i.e. maximize arrests per officer)
A nice advantage of a system like this is the model can be continually updated. The more data you have, the better the model can become. If the criminals start using some sort of anti-prediction method, I would expect the police would see a drastic drop in their coefficient of determination, and the model would be useless. However, I don't think most criminals are that smart. I think most criminal activity is rooted in some basic sociological rules, too ingrained to be easily changed.
I don't see this method helping an individual cop on the street. They would do their job as usual. What will change is at police HQ in large cities. It will help them determine boundaries of precincts, staffing, and patrol routes.
One of our competitors trademarked the term "hypothesis". From now on, we will call them "boneheaded ideas".
That's not what the article said. It talked specifically about preventing the specific crime that was predicted.
Hey, I don't have to predict anything to lower crime rates. Find the places that have the worst crime, and flood the streets with cops. Done. That's the same thing that you're suggesting.
If you need web hosting, you could do worse than here
Just because some people come to the gas station, and have guns, doesn't mean they were going to rob the place. Until they actually commit the crime, you don't know what they would have done, you can only guess and assume.
If they commit the crime, even if you bust them right away, you didn't prevent the crime. You just caught them immediately.
If you need web hosting, you could do worse than here
The legal term you're looking for is called "shopkeeper's privilege," and it has evolved -- in some jurisdictions -- from the right to challenge shoplifters to being basically indistiguishable from police powers with immunity from lawsuits and prosecution.
If you've noticed the ridiculously increased militarization of your local mall security, this is one of the reasons why. When one of those Seth-Rogen-Wanna-Be's cracks your skull with a nightstick, this is why the lawsuits you file will be dismissed.
It's not that our country is headed in the wrong direction any more. It's that we've already arrived.
He put his boots up on the table and made a face. "The sig," he smirked. "You can waste your life in search of the sig."
Unfortunately, these days...when they try to do that, they get covered up with complaints of profiling or some other ethinc outcry.
At least with a system like this...they can now point to something 'fancy and computerized' that backs up the arguments they've had all along...
Light travels faster than sound. This is why some people appear bright until you hear them speak.........
Stay on that train of thought, and eventually you end up at the same place Thomas More did in 1516 with "Utopia." More noted that the Crown created thieves be depriving people of both education and any possible livelihood, and then punished those same people with shocked outrage when they stole the food they had no other means of getting.
More went on to note that people who have been made desperate often do desperate things, and that keeping large numbers of ill-restrained, overly-armed men scattered throughout the populace did more to destroy the peace than to keep it. He then made the radical intuitive leap that if people could be kept educated and occupied with productive work while being spared from the ravages of poverty then society as a whole would greatly benefit.
Naturally, they beheaded him in 1535.
He put his boots up on the table and made a face. "The sig," he smirked. "You can waste your life in search of the sig."
It is not the job of the police to prevent crime. Nor is it their job to protect you as an individual. (don't believe me? check with SCOTUS.) Regardless, the only reason anyone is even working on predictive analysis is because the public is demanding it.
The job of the police is to take reports, conduct investigations, and apprehend suspects. If an individual officer feels it is his/her duty to help someone out, or protect them, then that's wonderful and I'm glad there are officers willing to do that. However, it's not their job, and you shouldn't expect it. The problem is people do expect it, and the more we expect the police to look out for individuals, the less freedom we're going to have. That's the only way for them to do the job we're asking them to do.
The coming police state (some say it's already here) will be the result of ordinary citizens refusing to take responsibility for their own safety. By foisting the job off on the police, you attempt to make them responsibile for your welbeing. In doing so, you compromise your rights. So if you don't like fascism, then stop asking for it.
Cops have been doing this for some time. There is always a section in any metropolitan police department that does such analysis. They are expected not to catch crooks but stop crime. Such analysis can be effectively used to position police departments in the right area to deter crime.
It is also unfortunately used to move officers around to enhance traffic tickets by analyzing slumps in tickets they know when drivers have adapted and will move to fresh pickings.
I'd go on a Vegan diet but the delivery time from Vega is too long. --brownkitty
half of humanity interprets signals the wrong way, and this can probably not be hacked ... nor can it be predicted. . . would i dare use the word 'lazy' ?
Free speech was meant to be free for all... how can anyone grow up in a nanny state ?