Schneier on Security
A blog covering security and security technology.
« Fever Screening at Airports |
| Carrier Pigeons Bringing Contraband into Prisons »
June 26, 2008
Pervasive security cameras don't substantially reduce crime. There are exceptions, of course, and that's what gets the press. Most famously, CCTV cameras helped catch James Bulger's murderers in 1993. And earlier this year, they helped convict Steve Wright of murdering five women in the Ipswich area. But these are the well-publicised exceptions. Overall, CCTV cameras aren't very effective.
This fact has been demonstrated again and again: by a comprehensive study for the Home Office in 2005, by several studies in the US, and again with new data announced last month by New Scotland Yard. They actually solve very few crimes, and their deterrent effect is minimal.
Conventional wisdom predicts the opposite. But if that were true, then camera-happy London, with something like 500,000, would be the safest city on the planet. It isn't, of course, because of technological limitations of cameras, organisational limitations of police and the adaptive abilities of criminals.
To some, it's comforting to imagine vigilant police monitoring every camera, but the truth is very different. Most CCTV footage is never looked at until well after a crime is committed. When it is examined, it's very common for the viewers not to identify suspects. Lighting is bad and images are grainy, and criminals tend not to stare helpfully at the lens. Cameras break far too often. The best camera systems can still be thwarted by sunglasses or hats. Even when they afford quick identification — think of the 2005 London transport bombers and the 9/11 terrorists — police are often able to identify suspects without the cameras. Cameras afford a false sense of security, encouraging laziness when we need police to be vigilant.
The solution isn't for police to watch the cameras. Unlike an officer walking the street, cameras only look in particular directions at particular locations. Criminals know this, and can easily adapt by moving their crimes to someplace not watched by a camera — and there will always be such places. Additionally, while a police officer on the street can respond to a crime in progress, the same officer in front of a CCTV screen can only dispatch another officer to arrive much later. By their very nature, cameras result in underused and misallocated police resources.
Cameras aren't completely ineffective, of course. In certain circumstances, they're effective in reducing crime in enclosed areas with minimal foot traffic. Combined with adequate lighting, they substantially reduce both personal attacks and auto-related crime in car parks. And from some perspectives, simply moving crime around is good enough. If a local Tesco installs cameras in its store, and a robber targets the store next door as a result, that's money well spent by Tesco. But it doesn't reduce the overall crime rate, so is a waste of money to the township.
But the question really isn't whether cameras reduce crime; the question is whether they're worth it. And given their cost (£500 m in the past 10 years), their limited effectiveness, the potential for abuse (spying on naked women in their own homes, sharing nude images, selling best-of videos, and even spying on national politicians) and their Orwellian effects on privacy and civil liberties, most of the time they're not. The funds spent on CCTV cameras would be far better spent on hiring experienced police officers.
We live in a unique time in our society: the cameras are everywhere, and we can still see them. Ten years ago, cameras were much rarer than they are today. And in 10 years, they'll be so small you won't even notice them. Already, companies like L-1 Security Solutions are developing police-state CCTV surveillance technologies like facial recognition for China, technology that will find their way into countries like the UK. The time to address appropriate limits on this technology is before the cameras fade from notice.
This essay was previously published in The Guardian.
EDITED TO ADD (7/3): A rebuttal.
EDITED TO ADD (7/6): More commentary.
EDITED TO ADD (7/9): Another good survey article, and commentary.
Posted on June 26, 2008 at 1:18 PM
• 65 Comments
To receive these entries once a month by e-mail, sign up for the Crypto-Gram Newsletter.
"their deterrent effect is minimal."
Exactly how do you measure the number of crimes that weren't committed?
Did you s/z/s/ or did the editor? :)
"But the question really isn't whether cameras reduce crime; the question is whether they're worth it."
How much would the price have to fall before they became worthwhile, or do the arguments about civil liberties automatically trump those about economic effectiveness?
I think cameras would be better in dark locations if they had motion sensors on them that honed in on the movement, getting a narrower shot. This would HELP reduce the graininess problem, along with motion detector security lights that would decrease the camera's ISO when they came on.
I locked up my bicycle on the street within 10 feet of a security camera. It didn't help. The seat was stolen twice, once in broad daylight by two neighborhood kids that I caught walking away from the scene of the crime (they left it and ran when security came by to see why I was yelling at them), and once overnight, so I had to buy a new one. Then the cable was cut and it disappeared, leaving only the severed remains of the lock behind. I reported it to the police, and reported to the university security, who owned the camera. They said they would check the tapes but I never heard back from them, and never saw my bike again. Very effective.
> Combined with adequate lighting, they
> substantially reduce both personal attacks
> and auto-related crime in car parks.
The International Dark Sky Association may have a different take on what 'adequate lighting' means. One of their points: more light =/= more safety.
I have been saying this on your blog (and in other pllaces) for so many years now, what finally brought your view point around?
I also identified the main excuse for why politicos could get away with saying that CCTV did work (which is short term studies of issolated systems just after instalation).
How about doing the next stage which is to follow the money around to find out which political parties get the donations from directors of system operators like Crapita who get business whilst having an absolutly hopless performance record. But it was all above board for one of their directors to sub over an extreamly large donation to the current UK encomberats (Labour) who are so close to being financial bankrupt it is almost impossible to belive they are allowed to continue...
Then there are the civil servants and government ministers involved in awarding the contracts who leave their jobs and take up very high paying posts with the successfull company (or another that owns or is financialy associated with it) well within the times that the codes of conduct say they should not...
I think spying on politicians is a public good.
One point you didn't cover. Many of the "cameras" you see on British streets "protecting" shops etc are fake. The shop may have a working camera inside covering the till, but the outside camera is purely for show.
These cameras do little to prevent crime. We often see the images on TV when the police ask the public to identify the criminals. I for one would much prefer not to be attacked in the first place rather than have the police identify my killer from the pictures.
Street CCTV is just security theatre.
"Exactly how do you measure the number of crimes that weren't committed?"
You measure crime rates in areas before and after CCTV installations. You measure crime rates in areas with cameras verses areas where there are none. There are lots of ways to evaluate the impact of cameras on events.
at least, i sure HOPE there is. I mean, the whole point is that the cameras reduce the numbers of crimes committed. If it is impossible to measure crimes that are not committed, than no one would ever be able to prove that cameras work. "Cameras prevent crime" would be an unfalsifiable claim and therefore totally worthless (if you give a hoot about scientific rigor, that is)
"You measure crime rates in areas before and after CCTV installations."
No. That's measuring the crimes that WERE committed during one time period, and comparing that number to the number of crimes that WERE committed during another time period.
How do you measure the number of crimes that were NOT committed (that were deterred)?
"You measure crime rates in areas with cameras verses areas where there are none."
Well that's obviously of limited value, since two different areas can involve many different, possibly intervening factors.
The point is, what of the instances in which the would-be criminal was deterred? When he thought to himself, "I feel like grabbin' that woman's purse! Free money! Oh #&*$! There's a camera watchin' me. I better wait and go someplace else, and commit my crime where there aren't any cameras!"
How are you measuring those instances? (Hint: you're not, because you can't.)
And yes, I realize the crime was just "moved on" to some other place. That belies the camera's supposed lack of deterrent effect.
" "Cameras prevent crime" would be an unfalsifiable claim and therefore totally worthless"
It is silly to claim that the cameras work as a deterrent. But it's just as silly to say that they don't. You can't measure what didn't happen (very reliably.)
Bruce knows this (since I've posted this fact in his comments numerous times), but he makes the contrary claim anyway, not because he believes in what he claims, but because it fits his anti-camera agenda as an effective though disingenuous rhetorical tactic.
I'm afraid your article is counterproductive. I believe you actually intend to persuade the reader to reject widespread automated surveillance regardless of how effective it is. Making your argument rely on a critique of its effectiveness risks undermining your point, which is that we don't want to live in a surveillance society.
In the future, please imagine someone such as L-1 replying, "Well, we've added all this new tech, like facial recognition and a database of all citizens, that makes unmonitored CCTV effective in prosecuting crime after the fact!"
(I'm referring to the Rolling Stone article you linked to about a month ago)
There is no limit to human ingenuity, and it can solve the effectiveness problem with more cameras and more technology.
Keep up the good work.
@anon: "Did you s/z/s/ or did the editor?"
I assume you're referring to "well-publicised".
It's correct the way it is. The original was published in the UK.
@Same old song:
"Well that's obviously of limited value, since two different areas can involve many different, possibly intervening factors."
Ahhh yes, and obviously researchers cannot in any way identify some of these factors. In fact, researchers cannot in any way identify any factors relating to crime whatsoever. It's not possible to compare neighbourhoods in terms of crime committed. In any way. That makes any sense. Whatsoever. Right.
Furthermore, just for the sake of it, let's assume the following: in a given area, say somewhere in London, you track crimes committed. The area isn't too small and so you don't have too many extremes in the data you pull from this area. You can start to correlate fluctuations in crime with arrests and other factors. You get some idea of how the neighbourhood works. Now, cctv is installed. You keep tracking the area, keep tracking the various factors that you have previously identified, and then you start considering the effect of the cctv. Is it impossible to measure the crimes that are not committed? Perhaps, but that doesn't matter, since what you're interested in is the amount of crimes committed and by whom.
The lesson: cut the bs with the semantics.
Edward Palonek comments on CCTV Cameras. These Cameras are here to stay, even if it s not for security its for liability. They also keep vandals out. In addition these cameras do create a sense of security in people's minds. Even tho the research show otherwise it is one's perception that really counts and will ultimately dismiss such research. Edward Palonek Security http://www.paloneks.ca/
what you're talking about, obviously, is experimental design which is essentially a solved problem - albeit not an _easy_ problem. Were it not a solved problem drugs trials could not proceed since 'how do you measure the illnesses that didn't occur ... obviously you can't'.
The same logic used in drug and clinical trials applies to measuring the effectiveness of CCTVs.
"The same logic used in drug and clinical trials applies to measuring the effectiveness of CCTVs."
Making an accurate statement about the effectiveness of security cameras as deterrents to crime, requires measuring that effectiveness.
So, using the same logic used in drug and clinical trials, how do you measure the number of crimes that weren't committed (deterred in the sense of "There's a camera watchin' me. I better wait and go someplace else, and commit my crime where there aren't any cameras.") ?
From one of the articles:
""Her life has almost been ruined, her self-confidence entirely destroyed by the thought that prying male eyes have entered her flat."
No, that wasn't what destroyed it. It was that anyone ever let her know, that destroyed it.
What you don't know, can't hurt you. But it can be kinky. :D
"Pervasive security cameras don't substantially reduce crime."
Hey you missed a study:
"Murders went down within 250 feet of the cameras, but the reduction was completely offset by an increase 250 to 500 feet away, suggesting people moved down the block before killing each other."
Yes, and pervasive firewalls (including host-based) don't substantially reduce crime.
More to the point, even pervasive police offices don't substantially reduce crime, according to consultants/researchers:
"Five of San Francisco's 10 police districts should be eliminated as part of a major redrawing of police district boundaries, a nontraditional move that could reap huge public safety rewards, according to the report prepared for city officials by a Massachusetts consulting firm."
Effective management of the technology and resources seems to be the key to the problem, eh?
Advocating for limits of cameras makes no more sense than trying to limit firewalls just because people do not know how to deploy and manage them properly.
@Same Old Song
You almost got me. I was going to feed the troll, until I got to the end and you started in on "anti-camera agenda".
@ Same Old Song,
' deterred in the sense of "There's a camera watchin' me. I better wait and go someplace else, and commit my crime where there aren't any cameras." '
By the logic you have so far shown (such as it is) it is very easy to show the answer to you question above.
Simply look at how the criminals have responded to the CCTV.
There is a general historical trend in areas where and around CCTV is installed. Initially crime rates change marginaly then,
1, the arrest rate goes up in the CCTV area but stays fairly constant in surounding areas.
2, Some criminals are arrested and detained and the word goes around and others are discouraged.
The consiquence is that the crime rate in the CCTV area drops. Then,
3, The less smart criminals carry on in the area and continue to be eaisly identified and arested.
4, Others move into arears not covered by CCTV some are adjacent but others may be many miles away (shoping centers etc).
5, The crime rate in these other areas rise.
Which enabled the less honest report writers to say "arrest rates are up" in the CCTV area and "crime rates are declining whilst crime in surounding areas remains increasing" thus apparently substantiating the claim that "CCTV was an effective method of dealing with street crime".
Generaly at this point monitoring of the crime rates is politicaly ignored, which is unfortunate because something interesting happens...
6, The crime rate starts to rise again and the arrest rate drops.
7, the CCTV records some of the crimes but is no longer of use in identifying the criminals.
Basicaly the criminals have responded to a static threat and initialy avoided it, then developed a defensive stratagy which involves hoods caps and multiple layers of quickly changed cloths etc that a gang will swap amongst themselves to further confuse identification.
In otherwords the criminal has shown what appears to be an evolutionary response. The preditation of less wary criminals has been noted by the more wary who have taken a simple defensive measure of avoiding a static threat. Then the smarter ones have developed a counter defence of disguise, the others recognise this as a successfull stratagy and fairly soon the crime rate is comparitivly back where it was (also helped by the fact that CCTV is instaled in the surounding areas.
This behaviuor by the criminals is important to their survival and also should be for those responsable for dealing with street crime.
The following points should be noted,
A, static deterants even if initialy effective will eventially fail.
B, static deterants actualy make the criminals more of a problem to deal with.
Point A needs a little clarification in that I am using static in more than one sense.
Initialy the criminals take their activity away from the CCTV which is a simple spatial defence for them.
Secondly the criminals develop a stratagy against a technology that either does not change or changes more slowly than the criminals can to compensate for it.
Which has two consiquences,
Spatialy static technology is only of use with spatialy static targets.
Slowly developing technology actualy "evolves" the criminal, thus obsoleating not just the technology but other technologies as well as a consiquence.
As an example of this a criminal who knows that they are likley to be observed by CCTV starts to wear a defence that hides their facial ID and also swap cloths with other gang members to make their physical body shape change. They will also realise that they can be recognised by the shape of their hands etc (due to CSI and other similar programs) and therfore with little or no extra effort start to carry gloves. Which obviates fingerprint technology and also to some extent DNA technology which cloth swaping further hinders or makes irelivant.
After some thought most people start to recognise that the only effective deterant to street criminals is dynamic deterant that is capable of developing new stratagies as fast if not faster than the criminals.
Which in a round about way describes well trained police officers out on the streets...
To conclude CCTV is only of deterant use against fixed target crime where a rapid response is assured. It's secondary use is simply to provide a visual record, that is rapidly becoming worthless.
Worse it has side effects that weaken or obsolet other existing technologies as a consiquence of "forced evoloution" of the criminal.
Further continued development of CCTV systems such as "gait analysis" is also destined to fail. Like the Saber Tooth Tiger it is now over evolved and rapidly heading (one hopes) towards exstinction.
Due to the latness of the hour here in London I forgot to mention one thing.
In my previous post by "criminals" - ment those who go out in a premeditated way to comit street crime.
There is also a second class of person who comits street crime after ingesting quantaties of intoxicating chemicals. They are the stars of your "caught on camera" programs. For these "idiots" CCTV is always going to be effective as they either have no knowledge of it or do not care.
The frequent portrail of such "idiots" on TV is one ot the things that bolster the myth that CCTV is effective in fighting crime.
There are a few other points to consider but it is now early morning and I bound for my bed...
"You almost got me. I was going to feed the troll, until I got to the end and you started in on "anti-camera agenda"."
Heh. My city is in the process of installing CCTVs in the CBD http://www.wellington.govt.nz/haveyoursay/meeti ngs/committee/Strategy_and_Policy/2008/18Jun0915/pdf/2_Courtenay_Place_Lighting.pdf >. Partly inspired by this blog post I wrote them an email. The response I got back was in part, and I quote:
"If people are concerned about privacy I would suggest that 8 cameras in a city this size compared to probably around 200 - 300 that exist in the likes of banks, car parking buildings and a myriad of other places is insignificant in comparison."
This from the Manager - City Safety, Citizen Engagement Directorate.
Interesting. I guess the difference between public (money and spaces) and private (money and spaces) doesn't matter. Also, this seems a variant of "You've got no privacy. Get over it." >:(
Jon Sowden's city manager:
"If people are concerned about privacy I would suggest that 8 cameras in a city this size compared to probably around 200 - 300 that exist in the likes of banks, car parking buildings and a myriad of other places is insignificant in comparison."
Ask if the next time you get a ticket for running a red light you can have it dismissed on the grounds that compared to the hundreds of times a day people run red lights in your city your transgression is insignificant?
Of course, event the two incidents cited don't say anything about crime *prevention*.
Cameras are the bread and butter of infrastructure protection and corporate security. Period, full stop, end of story.
They are an audit tool that allow investigators to analyze patterns of activity and adapt security techniques and methods to fit. Occasionally they deter crimes, occasionally they catch criminals, but their true value is in recording exactly what the #!@& happened to the company's property.
In infrastructure protection, you can go back and spot the bad guy's surveillance efforts once they have been noticed by your employees.
A fence with a hole in it every ten feet is almost useless at containing sheep. When the fence is continous the pasture is controlled and the sheep don't wander off the ranch.
The analogy is that when you can't go away from the camera, when the camera can follow you home because there is a camera everywhere, then you can be controlled by the government.
Worse, when your granny can watch you chat up boys and fink on you to your mommy.
@Same Old Song
Let's assume that crime statistics are worth something (they aren't really, but that is for another discussion).
Then you just take a look at the development of the overall crime rate a larger area, including the one watched by cameras and the surrounding areas where the deterred robbers, muggers, pinchers etc. will go to commit their crimes.
If the numbers do not differ vastly, then the crimerate did not go down. Congratulations, you just moved crime about 200 yards for app. a lot of money.
Measuring different times (last year vs. this year) is taken care of the way it is done with all time-comparative statistics, by allowing a margin of error.
Spent batteries are the bread and butter of fighting poverty and hunger. Period, full stop, end of story.
See, adding "Period, full stop, end of story" does not make a statement true.
"In infrastructure protection, you can go back and spot the bad guy's surveillance efforts once they have been noticed by your employees."
I have yet to see this happen. No one doubts that in an industrial complex with a defined area walled off from the public cameras that are closely watched provide some security, simply because all people moving into the watched area are suspicious, but thats about it.
@Same Old Song
"Exactly how do you measure the number of crimes that weren't committed?"
Of course, there is no scientific way to do this, but you could compare crime figures against areas without cameras.
For those arguing privacy -- the second you are outside of your home, you are in public and your actions are public knowledge. Arguing "civil liberties" is not valid (and I am one who is very much into privacy).
If you want to argue against the cameras as a value item, fine. I'd say that if a police jurisdiction sets them up and it moves the crime outside of that jurisdiction then it has done the job.
As far as guaranteeing the cameras stay on public areas (i.e.: not looking in on someone's apartment), that's a different argument and one that would need to be addressed. This could theoretically be done with software built into the camera to prevent it going at certain angles based on its GPS and inherint knowledge of the buildings and usage of those buildings based on documents within said jurisdiction. Again, there'd have to be protection, but it is feasible.
The two examples you give in your first paragraph aren't exceptions. The cameras didn't *prevent* the crimes, they merely helped figure out whodunnit afterwards.
Well, we know that emotional state is part of committing crimes, so clearly the sense of security engendered by CCTV is going to reduce anxiety and thus the impulse to commit a crime. So in an area where there are CCTVs there will be fewer crimes that weren't committed. But the same number, give or take, that are. There you have it. Bulletproof logic.
"If the numbers do not differ vastly, then the crimerate did not go down."
Well, measuring the crime rate is not the same thing as measuring the number of deterred crimes.
You could measure the crime rate for Month 1, before any cameras were installed, and for Month 2, after cameras were installed.
If there were 8 purse-snatchings during Month 1, and 9 during Month 2 (after the cameras were installed), you might conclude that not only did the cameras not decrease the numbers of crimes that occurred, the cameras actually INCREASED the crime rate.
But you still haven't measured the number of deterred crimes.
Unbeknownst to you, a local gang had had decent luck snatching purses in the area in question for several of the previous months, and were planning on ramping up their activities in that area during the next month (what turns out to be your Month 2.) They sent 15 snatchers into the area on Day 10 of Month 2, all of whom reported to the kingpin at the end of the day that "There's a whole bunch of newly-installed cameras at each of our favorite corners! I had at least a dozen opportunities to snatch a purse, and I refrained from taking advantage of any of them, because I didn't want to be caught on camera! We'll need to move our operation elsewhere!"
So the real "crimerate" for Month 2 not only did not increase, but rather decreased from a total of 79 aborted purse-snatchings, down to only 9, a huge reduction in the crime that would have occurred, but was deterred by the presence of the cameras.
Your measurement of crimerate would completely miss such an improvement.
The problem is, unless you're a member of that gang, you won't know that such an improvement occurred, because the improvement consists of budding, planned crimes that didn't come to fruition, that didn't happen.
So when one praises of derides deterrence, the question remains:
Exactly how do you measure the number of crimes that weren't committed?
79 crimes didn't happen that day, thanks to the presences of the cameras. Or was it only 76? (Or 0?) How do you know?
For all I know, Bruce is unaware of the fact that, in this context (crimes aborted at the last moment) you can't measure what didn't happen, and he sincerely believes that cameras have little or no deterrent effect.
But I would like to know how he's measuring that.
@Same Old Song
"But you still haven't measured the number of deterred crimes."
The number of deterred crimes is -1 in your example. This is how the phrase is used in this context.
That you are deliberately ignorant of this isn't very interesting. Apparently not even to yourself.
Just by coincidence, I watched the film "Red Road" last night. Based on the premise that there is a master control room for all these CCTV cameras with banks of monitors and people actually watching them in real time and stopping crimes.
"The number of deterred crimes is -1 in your example. This is how the phrase is used in this context."
No. A crime is not deterred if it happened. That's the opposite of a deterred crime.
1. to discourage or restrain from acting or proceeding: The large dog deterred trespassers.
2. to prevent; check; arrest.
@Same Old Song:
If you have crime rates for a very long time (which we do), then you can make out patterns that are statistically valid.
I know this is difficult to grasp for a layman, but there is a whole branch of mathematics doing nothing else then developing models etc. - its called statistics.
Since crimerates are a well researched area it is known which parameters have an impact on crimes happening - and which have not. Putting cameras at place A will in some cases lower the crimerate there, but it will increase the crimerate at place B. In the overall context the same amount of crime happened in the world.
To add to your confusion: crimerates do not tell us as much as we would like, since they only include crimes that happened _and_ the police is made aware of. This is why a factor called "Dunkelfeld" (estimated number of unknown cases) is added to the equation, which in itself depends on factors like type of crime, area, people living in an area etc..
You can get a lot more information on this at your local university. I would recommend statistics and criminal law, where this is explained én detail.
"If you have crime rates for a very long time (which we do), then you can make out patterns that are statistically valid."
Absolutely. But you're missing the point, Kaukomieli. The crime rates (which you've had for a very long time) are rates of crimes that *actually happened.* The fact that an estimate of the unknown instances of crime (dunklefeld) is to be factored into a rate of crime, nonetheless addresses crimes that again, actually happened. These last are crimes which were not deterred, but which went undetected or unreported.
Measuring these is not the same thing as measuring the number of deterred crimes, which are crimes which did NOT actually happen (as opposed to being merely unreported.)
No mathematics at all (advanced or even simple) is required to grasp the simple concept of "a sequence of events that terminates in an aborted final event, such that the ultimate purpose was not achieved." For example:
A purse-snatcher prepares to commit his crime, sneaks up behind an old lady, and just before he reaches for her purse, restrains himself from reaching for her purse, and walks away, having decided at the last moment not to commit the crime as a result of something having discouraged him from proceeding with his plan.
That was a deterred crime.
The question is, how do you measure such "deterrences" (without interviewing the would-be criminal and getting him to confess that he almost committed the crime, but was deterred) ?
Since no such interview is likely to take place, you won't ever know of this instance of a budding crime that was deterred. So that instance goes unmeasured, as do all the other instances of deterred crimes.
But just because the instances go unmeasured, doesn't mean the instances didn't exist. It simply means that you can't measure them effectively, and so you can't know how many crimes were deterred.
I doubt anyone would dispute that statistics is a wonderful tool for analysis and interpretation of complex data. But it's decidedly less useful when applied to an absence of data.
>I think spying on politicians is a public good.
The STU [Special Training Unit] had its own single-engine Piper Lance, and had obtained a BigEye surveillance pod for it. The BigEye was a gyro-stabilized combination video camera for daytime use, and infra-red camera for night use. An operator in the plane could put the camera's cursor mark on a stationary or moving ground target and the camera would lock on to it even as the plane circled high above, out of sight and sound of its quarry.
The extensive use of light planes was a tradition in the ATF going back decades; from the time when the "revenue agents" had flown them to spot bootleg liquor stills from the air. These pilot-qualified agents bragged that for them ATF stood for 'agents that fly'. The numerous flying special agents and ATF light planes often permitted them to reach the scenes of federal crimes involving illegal firearms or explosives before any other agencies. Any one-horse Podunk town with a dirt landing strip nearby could usually have ATF agents on the ground in a few hours at most. The ATF was independently air-mobile to a greater degree than most other agencies at the light plane end of the aviation spectrum.
After a brief familiarization period with the BigEye Malvone gave his air team the addresses of a dozen senior government officials who were in a position to help the STU. They hit pay dirt on a Sunday morning in June when the Piper was flying lazy eights over Fairfax County Virginia, and they noticed activity at the estate of Deputy AG Paul Wilson. A Mercedes arrived with a young couple who turned out to be Wilson's daughter and son-in-law. Mrs. Wilson then left with them to attend church services.
Soon after the driveway's automatic gate closed behind the Mercedes, Paul Wilson had appeared in a bathrobe on the back patio of the mansion by the swimming pool, accompanied by someone else. The stabilized zoom lens of the Big Eye then recorded in intimate detail the white-haired senior federal official and a black-haired girl playing in the Jacuzzi, with no detail left to the imagination for the next fifteen minutes. Upon further investigation the girl had turned out to be the 16 year old daughter of the Wilson's Costa Rican housekeeper, who had taken the day off.
Malvone was smiling broadly at the memory. "As soon as I saw that tape I knew we'd own Wilson, we'd have him in our pocket. When the time comes he's going to go to bat for us, big time, and we'll get the Special Projects Division approved."
"The FBI's going to fight it. They'll never let ATF have a new division with that much power."
"That's where you're wrong Joe, the STU or SPD or what ever we end up calling it is going to be seen as a dirty outfit for dirty jobs, and the FBI won't want any part of it. If the SPD falls on its face, the stink won't rub off on them. They'll be glad to let the ATF have it, and let the ATF take the hit if things go wrong. By the time they figure out what's really going on, the Special Projects Division will be too big for them to stop."
The discussions about the pro's and con's of video surveillance have always revolved around the arguments based on does it work, or does it not.
The very short answer is it depends what you want the cameras to do, and more specifically, how you actually do it.
Politicians and to some degree journalists will; frequently cite the performance of "CCTV" without really having that much insight into what it is they are actually referring to.
As a very simple example, let's just consider whether vehicles serve their purpose. If you were in charge of a city's mass transit system, you'd need vehicles to move the populace around, so if the fleet comprises two seater sports cars instead of fifty seater buses, the chances are that subsequent studies might conclude that the vehicles don't work, rather than a more accurate assessment, which is they do work if you choose the appropriate vehicles for the job, and then use them correctly.
Now if anybody ever asks me whether CCTV works, my short answer would have to be, only if the correct technology is applied using the most appropriate techniques in order to fulfil the desired objectives.
Has that been the case thus far ...? Regrettably no!
Whilst sticking a camera on a pole may well produce pictures and recordings, in most situations it won't necessarily fulfil operational objectives, whilst at the same time taking account of very valid and frequently disregarded issues relevant to civil liberties.
Security cameras may well be here to stay, but whether they are used appropriately and efficiently any time soon, is in practice a completely different ball game.
i'll invite anyone to go to http://www.break.com or http://www.liveleak.com and have a look-see at all the "purse snatched on camera!" or "gas station clerk shot on camera" or "woman beat up by police officer on dash cam" links. this is by no means a scientific sample, but it's clear that in many instances cameras don't stop even police officers from breaking the law.
the reason, as is seldom stated, is that these are crimes of rage or desperation. a crack addict in need of a fix doesn't give a crap about getting caught. a cop who's lost his temper isn't going to calculate all the possible outcomes. cameras witness these kinds of crimes, but they do nothing to deter them.
if the purpose of cameras is to be able to catch criminals after the fact, fine. but punishing people isn't the goal of law enforcement, preventing crime is.
@Same Old Song
"I doubt anyone would dispute that statistics is a wonderful tool for analysis and interpretation of complex data. But it's decidedly less useful when applied to an absence of data."
This is when it is most useful, it gives you a way to create a most picture of what is most likely happening, from data that can be correlated with the data that is absent. So you have some hope of making the right decision.
You seem closed minded and just want to blabber on about how intentions can't be measured and hand wave about cctv being a good idea. Rather than trying to back a case either way with data.
Having searched the market and looked at various suppliers for CCTV for our small business in Shropshire we decided to approach Interwatch Security, who are also based in Shrewsbury. The security system is ideal for our business. Their system allows us to be able to view our camera via their website, live and recordings. There is also the ability to receive text messages and phone call alerts.
We use 3 Axis cameras for our shop in Monkmoor.
Same Old Song:
Your statistical argument is just bizarre. What kind of magic do you think people used to discover that quitting smoking or taking a baby aspirin every day decreased your risk of heart attacks?
However, let's imagine that somehow, the impact of cameras on crime is impossible to determine. In that case, we're spending a bunch of money, and introducing a bunch of privacy violations, for something about whose impact we have no idea. Why would that make sense?
Interesting time of my reading this post. We have a substantial CCTV security system at church - much more than I would think is necessary. A couple weeks back, a purse was snatched.
On video, we caught the guy going methodically room-to-room during service looking for left-behind's from Sunday School. We were able to follow him out of the building and into his car with a clear view of his plate. An arrest was made.
I still wonder, if this is the extent of the purpose of the camera's, if they were worth the investment. Hopefully nothing larger ever occurs to prove their worth.
One area where I have seen CCTV camera being useful is where they are coupled with a load speaker and a monitoring station. In this context they are effective at deterring crime and this effect is directly measurable. It also provides an immediate response (albeit a verbal-only response, not a physical presence) which traditional cameras do not provide.
As an example, my brother had his car stolen from a car park but from the part of it that wasn't covered by the cameras. We spoke to the people who monitor the camera and they said that they regularly scare off kids who are looking for their first hotwire/joyride experience with the loud speaker. They just get freaked out to think that someone is actually watching them, even if that someone is 60km away. Mostly they don't know and think that the person monitoring them is close by and has backup.
Unfortunately this is just an example of moving the crime somewhere else... and not very far away in this case but with an extra camera or two they could have moved it away from the remote and lonely car park and into the much more difficult and dangerous (for the criminals) suburbs where they are less likely to succeed.
I'm not sure I like the privacy implications but I think that CCTV cameras could be used more effectively than they are now and they could actually deter crime if used this way.
In 1996 Australia's largest retail department store chain was suffering a shrinkage rate of around 2.5% TO 3.0% at its largest site.
It decided to install the largest CCTV system in an Australian retail environment, at a cost of around AUD 3million. Plus invested in the wages and staffing to monitor the system almost continuously and to have loss prevention officers able to respond to crime in progress. In 1997 and 1998 it suffered a shrinkage rate of around 2.5% to 3.0%. Since the system had realised no impact, they did not pull it out, but ceased to invest in the staffing to monitor the CCTV system.
(As far as I am aware, this data has not been published elsewhere.)
Before, during and after the CCTV experiment the store maintained all the traditional means of loss prevention.
And, yes, the shrinkage rate in the Australian retail scene is much larger than in the USA or UK.
@Same Old Song
"A purse-snatcher prepares to commit his crime, sneaks up behind an old lady, and just before he reaches for her purse, restrains himself from reaching for her purse, and walks away, having decided at the last moment not to commit the crime as a result of something having discouraged him from proceeding with his plan."
For measuring if there are deterring effects of CCTV-coverage you look at months A-E, have a crimerate of 100 per month in a certain vicinity, mount CCTV and look at month F. If the crimerate is down to 60 (after factoring in all other known influences to crimerate) you can assume that CCTV was the deterring factor, as otherwise you can safely assume crimerate would again be 100.
No one cares if during months A-E some purse-snatcher was deterred by some car driving by or due to bad conscience - because it has nothing to do with effectiveness of CCTV.
Of course you would have to measure more time and factor in the changes in the crimerate of other vicinities around to look for crimes that just moved.
What would really be interesting is, if the "Dunkelfeld" gets smaller due to CCTV-coverage, as a higher percentage of crimes get noticed.
Today's paper reported of a bicycle owner who was told to review the CCTV footage himself so that he could find out who stole his bike. This can't be a good thing...
"Today's paper reported of a bicycle owner who was told to review the CCTV footage himself so that he could find out who stole his bike. This can't be a good thing..."
That depends on who you are. If you are the person who had their bike stolen and it was the only avenue open to possible recovery then to you (and most others) then it would apear to be a good idea.
However when it becomes general knowledge the system will get manipulated by those who simply want to get access for whatever reason.
The extent of the manipulation can of course to a certain extent be controled but like you I suspect that it will not untill something significant goes wrong...
"you can assume that CCTV was the deterring factor"
You can assume it, yes. You'd be wrong to do so, as it's an unfounded assumption. But you could assume it nonetheless.
"otherwise you can safely assume crimerate would again be 100."
Not safely, no.
Your arguments suffer from a logical fallacy: affirming the antecedent.
In your statement "If the crimerate is down to 60 (after factoring in all other known influences to crimerate) you can assume that CCTV was the deterring factor, as otherwise you can safely assume crimerate would again be 100.", you propose as an assumption the very proposition that is being questioned: whether or not the CCTV is a deterring factor.
Shockingly, you've missed the point once again: that you can't measure the crimes that didn't occur; the crimes that were almost committed, but that weren't; the budding crimes that were nipped in the bud by the (possibly) dettering effect of the CCTV cameras.
"No one cares if during months A-E some purse-snatcher was deterred by some car driving by or due to bad conscience - because it has nothing to do with effectiveness of CCTV."
Well, the would-be victims may well care. But no one here has attributed any deterrent effect to "conscience or cars". The (possible) detterent effect referred to by me, is that of the CCTV cameras.
"Of course you would have to measure more time and factor in the changes in the crimerate of other vicinities around to look for crimes that just moved."
Again, changes in the crimerate don't measure crimes that didn't take place by virtue of having been dettered (by the presence of CCTV cameras, for instance.) (See any of my responses to your posts above.)
"This is when [statistics] is most useful, it gives you a way to create a most picture of what is most likely happening, from data that can be correlated with the data that is absent."
Hardly. Statistics is most useful when you have no data to apply it to? Come now.
You can correlate the data of crimes that "the data that is absent" (if such exists--remember, the point is we don't know if these instances occurred, so it's not necessarily "absent" data; it could very well be non-existent data), but any conclusions you draw from it will (obviously, I hope) be much weaker and much less substantiated that conclusions that you would draw from applying statistical methods to actual data.
"...hand wave about cctv being a good idea"
Well, it may be a good idea, and it may be a bad idea. But citing "deterrence" or "lack of detterence" as evidence for against, is silly. Measuring the deterrent effect, if any, will be very difficult if not impossible (see my above posts.)
Does that really seem like handwaving about CCTV being a good idea?
"What kind of magic do you think people used to discover that quitting smoking or taking a baby aspirin every day decreased your risk of heart attacks?"
Well, your analogies aren't apt. To use your first example, when a statistician compares the rate of heart attacks of those who quit smoking to the rate of heart attacks of those who didn't quit, and perceives a decrease, he doesn't pretend to be able to know the number of heart attacks which didn't happen. He concludes only that fewer, or none, occurred in those who quit smoking.
In the case of the deterrent effect of the presence of CCTV cameras, claiming that there is or isn't such an effect pretends to know how many budding crimes were stopped (e.g. "36 would-be muggings were nipped in the bud before they occurred, as a result of the would-be muggers aborting their mugging upon noticing the cameras, and presumably fearing having their crime videotaped").
Few people are silly enough to make those claims (or claim their opposite-little or no deterrent effect, which Bruce claims), since it should be obvious that it's very hard to measure how many budding crimes were aborted. (see my example in an above post.)
To use your second example, when a statistician compares the rate of heart attacks of those who take baby aspirin to the rate of heart attacks of those who don't, and perceives a decrease, he doesn't pretend to be able to know the number of heart attacks which didn't happen. He concludes only that fewer, or none, occurred in those who take baby aspirin.
For him to claim "The baby aspirin takers avoided 36 heart attacks each!" would be silly, since his data doesn't show that.
Now, for him to claim that, since those who didn't take aspirin suffered, on average, 1 heart attack per year (let's say), and those who took aspirin had no attacks, that therefore those who took aspirin probably avoided 1 heart attack per year, may well be a well-founded conclusion.
But the detterent-effect-of-CCTV-cameras, either for or against, is of a different kind. It claims that we can know what we can't ("36 muggings avoided!" or "No muggings avoided at all!"). Such arguments are analogous to a statistician observing data from baby aspirin's effect on heart attack rates, and then claiming that baby aspirin probably prevented "not just 1 heart attack per year in those who consumed it daily, but 12 attacks per year avoided!"
Aspirin may well have prevented 12 heart attacks in those who took it for that study. But to conclude that from that data is unwarranted.
(Keep in mind also that if the data showed an INCREASE in the average number of heart attacks (from 1 to 2, for instance) in those who took aspirin, that would not necessarily mean that, had aspirin not been taken, there would not have been 10 heart attacks in those subjects. In other words, the aspirin may have been working wonders (preventing 8 hearts of the 10 that would have happened, and allowing only 2 of those 10 to happen), but we don't know. Again, it can be very difficult to measure what did NOT happen, and no statistician will claim that he can produce as valuable of a result from working with an absence of data as he can from working with data.)
"In that case, we're spending a bunch of money, and introducing a bunch of privacy violations, for something about whose impact we have no idea. Why would that make sense?"
Who said that it did?
I agree with Bruce very much. Thanks.
I believe your critique is unfair and ignores key elements about video surveillance.
For instance, "By their very nature, cameras result in underused and misallocated police resources." Cameras have been used effectively for many years and by many organizations as an effective force multiplier. Indeed, they are recommended by Sandia Labs for proper physical protection systems. Cameras may be misused but it is certainly not by their very nature.
Also, the comment: "“Criminals can easily adapt by moving their crimes to someplace not watched by a camera and there will always be such places.”
This ignores the reality that different assets are more valuable and different locations are more conducive to crime. Shifting crime away from high value assets with attractive locations makes crime less attractive. You do not need to put locks or cameras everywhere to improve security.
I am a cctv expert and I have an extensive response at: http://ipvideomarket.info/review/show/139
Regarding your July article on CCTV Cameras, I think there is a big angle on
this question that you fail to address: What about large networks of CCTV
cameras? I think there will come a time (soon) when cameras and their
supporting software are interoperable, where it will be possible to follow
complex activity (such as crime, and also parades, traffic, protests,
various "social rhythms" of urban areas, etc.). Some of this is being done
today over Baghdad and similar places, by aircraft/UAV with large-area CCTV
systems. These camera networks will provide both great opportunity and
great Orwellian danger, and therefore your question of oversight and
limitations and management is even more important.
I do think historical procedures that have been required of police
(administrative warrants to secure phone records, etc.) can be leveraged
here. I can see a day when an operations center is watching the city as a whole, with a limited police team as part of the center, and an "operational warrant" system in place approving police investigation of incidents seen by the operations center.
I applaud this posting and other studies you have done in contributing to a body of knowledge in security management. I only have a few brief comments:
1. Point of reference: CCTV is viewed by the vendors/integrators selling it, consumers unaware of its potential (good or bad) and by knowledgeable pesons such as yourself making its use more understandable to the consumer. It is absolutely necessary that myths concerning the use of cctv be avoided.
2. No cctv installation should be done without first conducting a careful analysis (risk assessment) of WHY it's necessary. CCTV shoud be responsive to foreseeable risks and careful integration with the user.
3. CCTV may be very useful - especially if interactive with other technologies and those at risk (e.g. emergency phones in parking lots, gaming tables in casinos, etc.).
4. It should never be viewed as a panacea as this creates a false sense of security and can mitigate the awareness and responsiveness needed by persons who should be part of the solution.
I'd have to concur with the points made in Ira S. Somerson's last post.
That said, it's probably also worth mentioning that a Risk Assessment Survey (RAS) is IMHO only part of the preparatory work necessary in order to achieve an appropriate and hopefully effective level of CCTV deployment.
As a key stage in helping to develop a workable System Profile (SP), it should aim to ensure that any eventual CCTV System Design is based squarely on actual operational objectives, rather than just naive technological aspirations.
Two things are needed for a crime to occur: opportunity and need. We can't do anything about a criminals need, but we can do our best to eliminate the opportunity. Cameras will reduce the opportunity...the need is a whole different topic.
Murder on CCTV
Crytptome.org has posted details of an interesting Dubai TV news cast, made up of CCTV excerpts of of a group of European passport holders who are said to go on and assassinate a Hamas Arms dealer, on behalf of Mossad. Copies of passports are shown and details of cell phones and other forms of surveillance. It certainly shows the comprehensive, after the fact, nature of CCTV evidence. Some of the CCTV footage seems to be deliberately tracking some specific members of the group, which could mean prior knowledge of intentions or associated monitoring of the murdered Arms dealer meetings.
But the detterent-effect-of-CCTV-cameras, either for or against, is of a different kind.
Schneier.com is a personal website. Opinions expressed are not necessarily those of Co3 Systems, Inc.