Schneier on Security
A blog covering security and security technology.
« How Vulnerable Was Internet Explorer? |
| Recognizing a Suicide Bomber »
January 31, 2007
Cameras Protecting Other Cameras
There is a proposal in Scotland to protect automatic speed-trap cameras from vandals by monitoring them with other cameras.
Then, I suppose we need still other cameras to protect the camera-watching cameras.
I am reminded of a certain building corner in York. Centuries ago it was getting banged up by carts and whatnot, so the owners stuck a post in the ground a couple of feet away from the corner to protect it. Time passed, and the post itself became historically significant. So now there is another post a couple of feet away from the first one to protect it.
When will it end?
Posted on January 31, 2007 at 2:05 PM
• 67 Comments
To receive these entries once a month by e-mail, sign up for the Crypto-Gram Newsletter.
It will end when someone figures out that a camera should have a view of its objective and at least one other camera. Once there are enough cameras to make this possible, the rest will just be where salespeople conned government officials.
The world is officially a Monty Python routine.
Who watches the watchers' watchers?
With heavy enough coverage, it should be possible to have seen anybody before they got into position to take out a camera. Of course, with wide hats, hoods, and masks, this won't help much.
Vandalizing the cameras is why it's such a short step from "some" public surveillance to "ubiquitous" public surveillance.
Hmm.. a circular camera setup should work. An attack on any one camera would be visibile by at least one other...
Indeed, a third camera placed beside the first and set up to watch the second aughta do it.
Alternatively, a second camera set up in the line of sight of the first would work as well, but then the first camera would be dual-tasked to both catch speeders and watch the second camera, which may not be so effective.
The American military uses a similar setup for surrounding missile bases, except that they use infrared beams (or an equivalent) as motion sensors instead of cameras.
Surveillance cameras should be mounted vertically with a lens which gives 360 degree (horizontal) coverage. Then it could see any attack other than from directly underneath (or above if it faces down). Of course you might need special software so that humans could recognize the output, but thats only temporary, pretty soon the cameras themselves will have enforcement authority.
One 360 degree panoramic camera FTW!
"When will it end?" - as soon as the privacy costs are successfully enforced as a monetary renumeration in court, and substantially greater (preferably in both frequency and cost) than the benefits derived from the cameras.
Carlos Mencia did a bit about this on his show on Comedy Central, talking about people stealing security cameras in LA's barrios.
The US Air Force used a sonic fence around its Titan missile silos, one can be seen at the Titan Museum in Tucson, AZ. It's possible that they have IR equivs, I don't know.
The Titan Missile Museum is at http://www.pimaair.org/index.php?...
The worst part of the whole thing is that until the mid 1990s, the number of people killed and seriously injured on Britain's roads fell every year. Since then the number has levelled out to be pretty much 3,500 per year.
Speed cameras were introduced.
People now spend so much time looking for the cameras, they don't see the car in front has hit the brakes. Or the child about to run across the road.
Wayne, thanks for the reminder about Carlos Mencia. As soon as I read Bruce's article I KNEW I had already heard someone joke about setting up cameras to watch cameras to watch cameras, because I remember thinking how silly it was and laughing heartily. Guess the joke's on me.
Didn't Dr. Seuss have something about the bee-watcher-watcher-watcher-watcher?
@Andy: not just speed cameras - so called "red light" cameras here (now dual function as speed detection and red light in many places) often account for an increase in intersection accidents, for obvious reasons.
article: "There have been seven camera attacks in just three years, with machines being set alight, damaged or pulled over."
Given the number of cameras in Scotland, 7 "attacks" does not sound like much of a problem at all, let alone over a 3 year period. It sounds like someone might be trying to get a big fat contract for a camera company.
I used to live in York, but I'd not heard that story. Does anyone happen to know more details?
Stolen cameras would have a ready market, both as second-hand working cameras and as convincing-looking dummy cameras (with the video output not connected, the camera just for show).
A large number of cameras could be temporarily disabled by a masked kid on a bicycle with a paintball gun. This could be done to blind the watchers in an area, or to draw manpower to the area, thinning it out in the actual target area.
Camera vulnerabilities will always be a problem, especially if the response time to disabling is long, which would be the case when cameras are used to replace boots on the ground.
Andy: People now spend so much time looking for the cameras, they don't see the car in front has hit the brakes. Or the child about to run across the road.
Because of course keeping your speed below the speed limit, and therefore not having to worry about speed cameras, is completely beyond your control.
They already do this in Australia... but usually only on Speed Cameras that are near Pubs ;-)
"There have been seven camera attacks in just three years, with machines being set alight, damaged or pulled over."
I think the figure is higher than that:
Nobby Nuts writes: "Because of course keeping your speed below the speed limit, and therefore not having to worry about speed cameras, is completely beyond your control."
Well, I've been in an accident aggravated by an enforcement camera. In my case it was a red light camera. The light turned yellow and rather than risk the yellow turning red before I could get through--though I was close enough to the interesection that thought it would be safe to go through. I stopped because of the camera and was rear ended by a driver in a giant 1970's station wagon. Although the accident was his fault for following me too closely, the accident wouldn't have occurred if the camera wasn't there.
At least one study seems to show that red light cameras cause more accidents than they prevent--though I'm not sure that the accidents were comparable which makes the trade off less clear.
"When will it end?"
It will end with cameras being sniped at by irritated local drivers. Speaking from personal experience, some Scots reactly badly to being f****d about by stupid rules. I am serious about this.
"Although the accident was his fault for following me too closely, the accident wouldn't have occurred if the camera wasn't there."
Well, THAT accident wouldn't have occurred, but ANOTHER one may have: you T-boning a car that got the green light.
In Phoenix, AZ, I think they analyzed the before/after accident rates at a number of pilot-project intersections before deciding to deploy red-light cameras at other dangerous intersections. That was at least 5 years ago.
In Scottsdale, AZ, speed-triggered cameras on the interstate (Loop 101) caused average speed to decrease by 10 mph. When that pilot project ended, average speeds increased by 10 mph THE FIRST DAY the cameras were turned off. And yes, there were more rear-end accidents when the cameras were working, but those were less severe than the high-speed accidents that had been occurring.
I have mixed feelings about red-light and speeding cameras, but I don't think it's accurate to say that they invariably fail to do their intended job. Some work, some don't, and some just trade one kind of accident for another, which may or may not be of lesser severity. As with all security decisions, there are tradeoffs.
>Who watches the watchers' watchers?
A Viking chorale, sitting in a cafe, feasting on Spam, Spam, Spam, Spam...
We already do this in Australia. It might be worth remembering that the camera which watches the camera has no cmera watching it...
I've often thought that a 'ring' of cameras, all watching eachother's backs would be the best solution. Plus, this stops the camera from watching the cars, which is the best possible outcome.
I'm guessing that after you pass the speed camera, you just speed back up again..? That's what we do.
@Kendal: In the Netherlands the same issue is at hand. Though the officials try to keep up replacing destroyed cameras, it's really of no use. Everybody knows their position and hits the brakes before accelerating again. The only "good" they do is adding to the government's financials. That's why they are not positioned at hazardous places, but where they generate most income...
The stats - even the ones published by the government show that excluding actual major accident blackspots, all gatso installations have resulted in increased numbers of accidents, mainly through inattention as people look down at the speedo for a split second, whether they are under or over the speed limit.
Other stats published by the UK govt show that despite rhetoric to the contrary over the past few years, speed is only a minor factor if at all, in accidents - the main ones are inattention and lack of experience.
Visit www.safespeed.org.uk for lots more info.
"Well, THAT accident wouldn't have occurred, but ANOTHER one may have"
"May" have, but there's no need to speculate in general, since RTA statistics will show whether accidents tend to increase or decrease at junctions/intersections with traffic light cameras. Of course you need a good statistician to interpret them. "Reversion to the mean" has a significant effect on the apparent effectiveness of any counter-measure which is selectively applied only in the worst places. Traffic cameras are in this category: they're often put in places which have recently had one or more accidents, including where those accidents are a statistical anomaly. So even if cameras have no effect at all, you'd still expect a short-term drop in the number of accidents after they're installed.
I think it's fairly clear, though, that anything which causes drivers (rightly or wrongly) to hit their brakes hard, is going to increase the incidence of accidents. This too has to be taken into account when studying the trade-off represented by the camera. Traffic lights even *without* cameras cause rear-enders, but I suspect an analysis would show that on busy roads they save far more accidents than they cause.
Sure, people shouldn't drive too close to the car in front, but in point of fact they do. So speed cameras and traffic light cameras, no matter what accidents they prevent, will tend to cause rear-enders. The too-close driver is obviously at fault, so there's a tendency to dismiss such accidents as "irrelevant". Then again, a driver is at fault if he causes an accident by speeding or jumping a red light, and the whole point of the camera is to prevent that.
Also, I don't know about the US, but in the UK lights are synchronised such that for a "sufficient" period when changing priority, everyone has a red light. If the light changes to amber as you approach, and you're close enough that you'd have to hit the brakes hard to stop, then there's no way you're going to collide with someone whose light has just turned green (although you could collide with someone who has jumped their light early, of course). So it's never necessary for a driver to brake hard when he sees a light turn amber.
Unfortunately, the driver doesn't know when the camera is going to start taking pictures. So the camera might change the behaviour even of safe drivers, because they're concerned about being punished unfairly. Whether this fear is justified is another matter - maybe something could be done to reassure drivers such as Skate that there is no need to stop suddenly. This would reduce the cost of the camera in excess accidents, without reducing its efficacy against drivers who are deliberately running red lights.
What doesn't reduce the cost of the traffic cameras, is installing CCTV to cover them. There's no way that it's cheaper to install and run the CCTV cameras than it is just to replace the small number of traffic cameras which is vandalised. If cameras were being vandalised every day, then maybe this would change. If a particular camera is repeatedly damaged, then surveillance on the camera will catch what is probably a single repeat attacker. But until one of those things is the case, this measure will cost far more money than it saves.
I think this is more about the credibility of the cameras themselves. The government feels that it will be better-respected if it can catch just one camera-vandal. It might be willing to spend money on this out of all proportion to the physical assets being protected, because respect for the law is very much at risk in respect of traffic cameras.
I used to live in York and haven't heard that story before, and I can't find anything by googling. Any details please?
www.safespeed.org.uk aren't necessarily to be believed when reporting what the government "admits", especially since links to DfT stats seem to be a bit lacking on the site.
For example, you say "even the ones published by the government show that excluding actual major accident blackspots, all gatso installations have resulted in increased numbers of accidents"
What the government actually publishes, for example, is this: (http://www.dft.gov.uk/foi/responses/2005/sept/negativeinfoaboutspeedcameras/letteraboutanfoirequestforne2622)
" The independent evaluations of the national safety camera programme by PA Consulting and University College London show that speed cameras reduce road traffic collisions and related deaths and injuries at camera sites. This was substantiated in a review by the University of the West of England in which 14 studies from around the world all drew the same conclusion.
We are aware that sometime in October 2004 a coroner concluded at an inquest that the location of a camera may have distracted a driver and therefore contributed to the death of a pedestrian. However we do not hold information related to his view other than associated press cuttings that report the case. "
Now, you could be quite correct that cameras consistently increase accident rates. But in respect of what the government itself publishes, I think you're mistaken: it has not published statistics which show that cameras consistently increase accident rates. I don't think it has published anything to the effect that speed is a minor factor in accidents, either, since I know someone who currently works for the DfT, and I'm sure I remember him citing a study recently that excess speed was the most common factor in serious accidents, as assessed by RTA investigators. Insufficient speed was also up there on the list, mind you.
In my first comment, when I said:
" I think it's fairly clear, though, that anything which causes drivers (rightly or wrongly) to hit their brakes hard, is going to increase the incidence of accidents."
I meant "increase the incidence of rear-enders". How it affects the overall incidence of accidents, and the overall incidence of serious accidents, is less clear...
As others have commented, this is standard practice in Australia. I have read about at least one incident where vandals disabled the second camera before attacking the speed camera, but they are generally watching from the top of street lights where they can't be damaged by a gun-free society.
It will end when cameras are no longer needed to hand out speed tickets, or speeding is no longer possible:
I expect they will require some strong authentication to identify the driver and radio transmitter in each car to report to the authorities everytime you go too fast. Your offence is emidiatly charged from your account.
Satelite tracking of cars to trace and arrest worst offenders, and a self destruct function for worst case scenarios - no borring halt the car, let's just blow it up - and tampering with the box will destroy the car too.
Of course, this will create much more revenue for the authorities, than requiring by law a black box that limits your max speed to the maximum allowed on any given location.
I actually once managed to convince one in Spain that they were introducing a system to track cars by satelite and issue tickets for speeding anywhere in the country at any time.
I have a real good idea. If you can't control your car DON'T drive. Contolling the speed is a basic requierment. That involves cheaking the speedo, without losing control or putting anyone else at risk. Otherwise DON'T drive. Sheesh, you only need to hit the breaks because you are already breaking the law.
Dam If all these people were *not* speeding then these "need to check my speed" type crap won't happen.
When i see a cop or a camera I don't need to look down. I know what speed I'm driving. (Learning to fly improved my driving alot i think)
Its the LAW: don't speed. End of story.
So Perhaps a better method of inforcement is tracking the cars. With the correct privacy protections. So that your not already speeding in first place.
There's a speed camera (the highest earning one in our state) near where I live (NSW, Australia) which has a camera monitoring it.
@Erik N: "I actually once managed to convince one in Spain that they were introducing a system to track cars by satelite and issue tickets for speeding"
Here in the UK, the government definitely *does* want to introduce tracking of all cars. This would be done by a GPS receiver in the car continuously logging its position. The intended purpose is to enable them to charge drivers according to how far they drive, on what kind of roads, and at what times.
So far, this is just in the stages of being an aspiration to work towards; there's no timetable for its introduction.
I anticipate plenty of "mission creep" if it is ever implemented, though. Speeding prosecutions is one obvious extension. Of course, there are many more applications for knowing all your citizens' movements...
You´ve almost come close to the crux of the matter - that people drive too fast. The cameras are there, ostensibly at least, to encorage drivers to stay below the speed limit and thereby reduce the number of accidents.
The fact that accidents actually increase, due to human behaviour, is to be seen as a failure of the cameras, irrespective of the fact that people should be driving slower. An appeal to personal responsibility is redundant.
@Kendal: "I'm guessing that after you pass the speed camera, you just speed back up again..?"
That's the usual approach, certainly. However, there are now a few places in the UK with a different speed camera system -- two cameras a mile apart recording vehicle number plates and timing how long it takes each car to travel the mile between them. If your average speed over that entire mile is above the speed limit, you're nicked. So you would have to slow down for a much greater distance to evade this system.
And now that they're bringing in number plate-recording cameras at very frequent invervals along all major roads (and storing for two years the information on what vehicles go past at what time, but that's another issue), they will have the infrastructure to apply this technique very widely in future.
hey, cameras are great, but what about attack from the air? I think they need to consider radar too, otherwise, quite obviously, everyone will just hit them from helicopters, microlights etc.
Geeze people, think: "Defence in depth"
I'd like to see a 100m perimeter fence around all speed cameras, with dogs and armed patrols.
I think you'll find this will mean that vandalism tails off completely.
"Because of course keeping your speed below the speed limit, and therefore not having to worry about speed cameras, is completely beyond your control."
Well, if speed limits had a grounding in some sort of actual safety and efficiency science; then probably I would obey them. However (I am speaking of the US, now; the speed limits in UK were logical when I was there '94-'95) they are political footballs.
Every time you turn around they are messing with them. In the US there is every possible speed limit between 5mph and 80mph SOMEPLACE [in 5 mph increments]. And it changes every few blocks along the same road. If the US had a predictable speed limit based upon road conditions they could save enough steel in speed limit signs to put the WTC back up.
I read a study which concluded that the drivers most likely to cause accidents are the ones driving ABOVE the 90%ile speed (faster than 90% of the traffic) OR BELOW 30%ile. Yet speed limits in the US are consistently set at approximately the 30%ile point, meaning the safest drivers are subject to penalty, but half the dangerous ones are immune.
In other words local governments will study what speed drivers feel safe at (used to be called "comfort zone" back when I took driver's ed) then set the speed limit for the area at a point where a plurality of drivers are in violation.
And when speed limits change, speeds do not. People continue to drive at the speed they feel safe. Unless, of course they see a cop or a camera. Then they slow down suddenly, bunching everyone together. Once they are past the speed trap, they speed up to make up for lost time.
And on the rare occasion when someone is driving at the speed limit, they will be reading, shaving, dressing or something to occupy the time they are in the car because at that low a speed they are not invested in the driving process. Kind of like a pilot on a long flight with autopilot.
And of course large sections of rural autobahn in Germany have no speed limit, and the speed limits they do have are higher, and the population density is MUCH higher; yet they have a much lower fatality rate than the US (which I primarily attribute to the fact that in Germany drivers are required to know how to drive; it takes about the same level of effort and expense to get a driver's license in Germany as it does a pilot's license in the US)
Speed limits (once again I am only speaking for the US) are STRICTLY a revenue-generation device; they have nothing to do with safety.
The speed limits are set for a good reason. A recent series of TV ads in the UK explained this:
Car hits a child at 30mph - child has 80% chance of surviving.
Car hits a child at 40mph - child has 20% chance of surviving.
The child most likely steps out in front of the car without warning - no driver could avoid this accident. But its the driver's speed that determines whether the child ends up in the hospital or the cemetery. Focusing on the fines is kind of missing the point.
Sensible Driver wrote: The speed limits are set for a good reason. A recent series of TV ads in the UK explained this:
Car hits a child at 30mph - child has 80% chance of surviving.
Car hits a child at 40mph - child has 20% chance of surviving.
First of all this is a propaganda advert by the government, and everyone knows how much to trust the government.
Secondly, how did they test this. How many children did they kill to get the 80% and 20% figures? Or perhaps they estimated them from accident statistics.
I don't know of any country that has widespread black box type recorders in cars (HGVs and coaches have them in Europe), so there is no way to know what speed the child was hit at, except from witness reports. What is the chance of witnesses estimating a higher speed for a collision in which a child dies compared with one in which it doesn't. Also, are these figures just for standard saloon cars, or does it include vans, people carriers, 4x4s, etc, all of which have different collision characteristics.
Of course, the figures sound "logical", the faster you hit something, the more damage occurs, so everyone believes them, even when they are not based on any type of reliable or scientific basis.
"Of course, the figures sound "logical", the faster you hit something, the more damage occurs"
No that is logical since Newton and the laws of motion.
What is less logical to the human is that slowing cars down increases the number of cars in a given length of road (stoping distance is relate to the square of the speed).
So slowing traffic down counter intuativly increases the volume of trafiic carried so more people get there in any givien period of time...
It's pure physics and you can go calculate yourself. The idea is this: The distance needed to break is related to the velocity squared.
And there is reaction time to add in which you drive at normal speed - that time is longer than you may think.
Check here for total stop distances at different speeds, assuming the same reaction time and the same maximum deceleration:
Or you can check this one which let's you plug in the numbers in the unit of preference:
(but it doesn't include reaction time).
So if you drive at 30mph and see a kid at 24m you have time to stop, but at 40mph you'll run her down at 25mph!
There is no campaining in this, it's pure physics.
@Sensible Driver & Clive Robinson
At last, some common sense. Someone that Thinks Of The Children™.
At 20mph, even less children would be hurt. At 10mph, less still, along an inverse square rule. No cars = no injuries, let alone deaths.
There's still the issue of running with scissors, but camera technology is still nascent in this area.
"slowing traffic down counter intuativly increases the volume of trafiic" - Over any given period, at a discrete point in time and space? What about density and / or distribution? The statement doesn't stand up to any scrutiny.
Besides, more volume would equate to more cars per child, increasing the risk of accidents.
And what are children doing playing on roads where you might find speed cameras - Typically trunk roads in non-residential areas.
@Anonymous: "As others have commented, this is standard practice in Australia. I have read about at least one incident where vandals disabled the second camera before attacking the speed camera, but they are generally watching from the top of street lights where they can't be damaged by a gun-free society."
I've known some people who are pretty accurate with a slingshot. If it can kill rabbits, it can probably kill the average camera installation.
That's right, Erik N. As I tootle along at 60km/h or more and some kid jumps out onto the road, chances are excellent that the kid is dead. Sadly, this is true regardless of what little signs the Authorities deem fit to afix to said roadway, how many laws they may enact, etc.
Physical reality simply doesn't care what we may think.
Should we be concerned about this? Rather than making criminals out of millions of people, why not ask a more pertinent question: why was that kid playing next to such a road to begin with? We don't blame the polar bear when it chomps on the leg of a tourist who offers herself, why should we complain when a car can't possibly stop in time, as you have demonstrated?
Even if such a questions is disallowed because you find it abhorrent to suggest that Innocent Children or their Delinquent Parents are in some way at fault because of their own actions or inactions, the plain fact remains that of all the deaths that occur on the roads, the vast bulk are among drivers and passengers of the vehicles. Pedestrians and others are hardly noticeable from this perspective.
Now let us ask the question: why is this the case? Allow me to suggest an answer: it is because of the physical design and implementation of the roadways, not the stupid little speed limit signs the busybodies put up to maximize their revenue. How many pedestrians are there on 100km/h+ roadways? Basically none, because of the fencing and other barriers to entry to the road surface. How fast can you possibly go in a typical residential area? Narrow road, on-street parking, bends and curves, and vastly more obeyed traffic control mechanisms like stop signs and red lights put substantial _physical_ limits on the speed.
People above are talking about "comfort speed" and the like, and for damn good reason. There is even a famous theory called "risk homeostasis" -- held by almost everyone who looks at the numbers, but flatly rejected by the bureaucratic control freaks who need their revenue streams to maintain their occupations -- which not only explains this "comfort speed", but also offers a rationale as to why the introduction of ABS and other non-intrusive safety systems has not lead to widespread reductions in motality re: driving.
All of this combined suggests that instead of printing up another crop of ever lower speed limit signs, that we instead shove the safety systems right in the faces of drivers, make them utterly unignorable, and do this in the places where it would have the greatest effect: where pedestrian/vehicle interactions are at their highest. This would be a one-time capital investment, continuing to Protect The Children for decades, and would require not a single human law to be obeyed, let alone enacted.
Well, if speed limits had a grounding in some sort of actual safety and efficiency science; then probably I would obey them.
Oddly, in California (possibly elsewhere in the U.S.), since at least the 1960's, the authorities have helpfully color-coded the speed-limit signs. The ones that are black on yellow are actually based on physical characteristics of the road and surroundings. They are established by real traffic engineers, and are typically found before curves, road-narrowing, etc. The ones that are black on white (or vice-versa) are set by a political process owing much to local revenue enhancement. If you are mostly concerned about your wallet, you need to take them all seriously. If you are only concerned about safety, you only need to pay attention to the yellow/black ones.
Bob wrote "the speed limits in UK were logical when I was there '94-'95" but, to be blunt, he is dead wrong. While they may make more sense in general than those in some other countries, there are still gross errors all over the place. Errors in both directions. And surprise surprise, those where the limit is set too low seem to sprout more cameras every week, unlike those where the limit is too high. I distinctly remember some of those errors nearly causing me grief when I passed my driving test in '94. Those exact same errors still exist to this day.
qui cameriet ipsos camarones?
@Derek Derrickson Jr
""slowing traffic down counter intuativly increases the volume of trafiic" - Over any given period, at a discrete point in time and space? What about density and / or distribution? The statement doesn't stand up to any scrutiny."
The logic is as follows,
1, There are always more cars than there is road volume.
2, Everybody drives at the minimum safe distance
3, The safe distance has two parts,
3a, thinking distance
3b, breaking distance
4, The lane length occupied by a vehicle is Vlength + ThinkD + BreakingD
5, The ThinkD and BreakD are unusable space.
So the vehical length is speed invarient and at all but very slow speeds is negligable with respect to the Thinking and Breaking distances.
The Thinking distance is directly related to speed ie MPS / Tms
The Breaking distance is based on the square of the velocity and above just a few KpH is the dominant factor.
So for any length of road the number of vehicles that can be safely accomadated is related to the dominant factor of S^2.
@Clive Robinson: "The Thinking distance is directly related to speed ie MPS / Tms
The Breaking distance is based on the square of the velocity and above just a few KpH is the dominant factor."
Nice physical analysis, but essentially wrong.
Sadly, that is not how people really drive, in most circumstances. People *do* attempt to "maintain an interval", but the usual technique basically assumes that the vehicle in front of you will have a stopping-distance relatively similar to your own. Thus, it can be effectively disregarded.
In the US states I have held driver's licenses in, the technique promulgated by the authorities is "The Two-Second Rule". This basically states that there should be two seconds (or more) of travel-time between you and the vehicle in front of you. This works pretty well in accounting for reaction-time, and generally lets you stop before rear-ending the vehicle in front of you, even if they slam on the brakes.
However, it reduces the inter-vehicle interval to a linear function of speed, again. Thus, effective vehicle density over time remains constant. Naturally, this approach does NOT deal well with the "child/obstacle/etc. in the road" situation. As you pointed out, braking distance increases exponentially with speed. The end result is that most drivers using the Two-Second Rule are probably overdriving their visibility with respect to anything except other traffic traveling in the same direction.
But, the Two-Second Rule is easy to check while dealing with all the other attention-requirements of driving. You periodically mark a signpost/shadow/pothole as the car in front of you passes it, and start counting until you reach it yourself. Most people can do a fair approximation of counting seconds for short periods of time. Calculating your actual expected braking distance "on the fly" is probably beyond the capabilities of most people. Our brains just don't deal very well with non-linear expectations and relationships. Thus, most people don't adjust their intervals to properly correspond to expected braking distances (especially under less-than-ideal circumstances, like wet or icy pavement).
To paraphrase Feynman,
I wonder why they put the camera there.
I wonder why I wonder why.
I wonder why I wonder why I wonder why.
I wonder why I wonder why I wonder why I wonder why.
Law enforcement in our area is promoting private corporate cameras to watch the public. Government gets all the benefits without the costs, supposedly. Too bad the central business district is full of closed down shops. They are counting on businesses that will build a new network of these things and share information. Our buses have cameras and the whole transit system is going broke. They can't even afford the drivers.
"Hackers Rebel Against Spy Cams"
"Wired is running an article looking at the little ways in which Austrian technology users are striking back against surveillance. From the article: "Members of the organization worked out a way to intercept the camera images with an inexpensive, 1-GHz satellite receiver. The signal could then be descrambled using hardware designed to enhance copy-protected video as it's transferred from DVD to VHS tape. The Quintessenz activists then began figuring out how to blind the cameras with balloons, lasers and infrared devices. And, just for fun, the group created an anonymous surveillance system that uses face-recognition software to place a black stripe over the eyes of people whose images are recorded.""
'Braking' stops cars.
'Breaking' glass hurts.
I don't think any government sets speed limits to maximize revenue. If they want to tax driving there are far more effective means: Tax gasoline, tax the car, road taxes etc. All give a stable flow of tax money.
Rather speed limits are set to help the driver judge the physical conditions of the road. And the physical conditions are not just the road itself but also the people on the road, and by the road - including kids.
Certainly, some roads are safe at higher speeds, but people are bad at managing many categories, so all roads of the same class gets the same limit.
There can be perfectly good reasons that kids are by the road or crossing it. Near schools for example, or just outside home. They may not be playing on the street, but playing football in the yard the ball ends up on the street.
Kids are less aware of traffic and less experienced, and roads in areas with kids should be designed and limits set to reflect this.
Yet drivers don't care - statistics show a huge number of parents ignoring road safety near schools and the like because they are in a hurry to deliver their kids!
Fact is that most people are ignorant of the risk they pose to others, and they are very bad at judging the physical conditions of the road. Speed limits is a simple and effective way to help, and fines are the means to make people respect these.
Ofcourse, if people were perfect drivers there would be no need for limits or fines.
As already pointed out, your analysis depends entirely on safe distances being maintained between vehicles. Five minutes of observation lays that to rest, and no amount of roadside ironware is likely to alter that particular impatient human trait. More often than not, the distance is either linear (a fixed time between vehicles) or constant (a fixed distance), irrespective of their relative speed.
Slowing down traffic will only ever increase volume if all participants were behaving themselvces in the first place, which tens to suggest the premise is moot anyway.
@X the Unknown:
Braking distance increases quadratically with speed, not exponentially.
@Erik N: "I don't think any government sets speed limits to maximize revenue. If they want to tax driving there are far more effective means: Tax gasoline, tax the car, road taxes etc. All give a stable flow of tax money."
I don't know if it's still true, but the state of Texas used to have highway patrol troopers whose pay was generated directly by the traffic tickets they wrote. "Speed Traps" were a very important source of revenue for those officers.
@Taneli Huuskonen: "Braking distance increases quadratically with speed, not exponentially."
Thanks. My bad.
@Eric N: "Fact is that most people are ignorant of the risk they pose to others, and they are very bad at judging the physical conditions of the road. Speed limits is a simple and effective way to help, and fines are the means to make people respect these."
Many states actually use what they call "The Basic Speed Law". This specifies that a driver is speeding if (s)he drives faster than the current conditions safely allow. This is over and above any posted speed limit regulations. This is also a purely subjective assessment, which theoretically allows officers to pull over anyone who seems to be driving in a way that would not give them full control of their vehicle. Howev er, my experience has been that it is mostly applied retroactively: if you had an accident of some kind, ipso facto you must have been driving too fast for the conditions. Essentially, it is a law which holds people responsible (and liable) for recognizing the risk they post to others, and the effects of physical conditions on the road. I fully agree with you that it doesn't seem to work well in practice - people generally disregard the Basic Speed Law under all but the most extreme of circumstances.
At the risk of dragging this back to security, the story as presented ignores the fact that the first camera is about road safety, so it is positioned down near the road, focussed on a small area, takes still pictures and its output isn't continually monitored.
The second camera is probably a security camera stuck on top of a fifty foot high pole, has a wider field of view, continuously records its output and has someone back at control watching that output on a bank of monitors.
So the second camera is much less vulnerable and any interference will be detected quickly. That's why it doesn't need a ring of cameras, in most circumstances.
The writer of this report should have been able to figure this out, or have asked the authorities for comment, or have realised that they didn't understand the situation, but why let the details get in the way of a good story... particularly if it makes the people who manage the unpopular road safety cameras look dumb.
This is already happening in Somerset (UK), initially all the authorities associated with the camera, signage, roads and law enforcement denied any knowledge. They're finally coming out and admitting it's an official installation.
I can't believe so many of you are overcomplicating this so much. The solution is simple.
Just post a policeman on every corner, to watch the cameras. Problem solved.
I'm afraid there are quite a few flaws in your analysis.
> 1, There are always more cars than there is road volume.
This is only true on non-motorway grade roads in built-up areas, and even then often only true at rush hour. On rural roads, suburban roads, and most motorways other than at rush hour, the peak capacity is many times greater than the actual capacity. Oddly, the circumstances where we do approach saturation are precisely the ones where speed cameras are usually not installed -- because it is not possible to speed!
> 2, Everybody drives at the minimum safe distance
This is again not true, at least not as you later define minimum safe distance. In roads that are not approaching saturation, cars usually travel in packets (or peletons), with the number of cars in each packet random but dependent on road conditions and % saturation. The lead car obviously has a huge distance in front of it. At saturation, the distance is determined by a lot of factors certainly including safe stopping distance for that car but also cultural factors, bunching/stretching effects of traffic lights, etc.
> 3, The safe distance has two parts,
> 3a, thinking distance
> 3b, breaking distance
> 4, The lane length occupied by a vehicle is Vlength + ThinkD + BreakingD
Note, what you call "thinking distance" is normally called "reaction distance" because for alert drivers it occurs considerably faster than you can think about the matter. It is highly variable even within one individual.
Even so, this formula is not correct. In the scenario of saturation that you are painting here, a vehicle only needs to emergency stop because the vehicle in front has started to do so. As such, the vehicle in front is going to take some distance to stop too. Thus the safe distance a driver must leave is actually
Vlength + ThinkD + OwnBrakingD - PrevBrakingD
Here's a problem: since we don't know the performance of the vehicle in front, we don't know PrevBrakingD and thus don't know what safe distance to leave. So, should we be conservative and ignore that factor, as you did? Well, that's not what drivers do. They start off with some sort of "average guess" plus a bit of fudge factor, then at each braking event they gradually back off if the car in front seems to be driving badly. (If it is driving _very_ badly, instead of getting even further behind they may try to get away from being stucj behind it.) However, for the most typical situation OwnBrakingD ~= PrevBrakingD and the V^2 factor almost disappears from the analysis.
> 5, The ThinkD and BreakD are unusable space.
ThinkD is unusable, BreakD just confers a modicum of risk. In practice, most of it IS used.
> So the vehical length is speed invarient and at all but very slow speeds is negligable with respect to the Thinking and Breaking distances.
This is not necessarily true - depending, of course, on how you define "very slow speeds". Reaction times can vary from 0.1 seconds for a young, expert driver who is highly alert, up to about 1.5 seconds for an elderly or badly distracted driver. In the optimal 0.1 s case, reaction distance is less than a typical sedan's vehicle length for speeds of up to nearly 70 mph. While if that young, alert driver is driving a minibus, it's likely his vehicle cannot actually reach a speed where reaction distance will exceed vehicle length. Of course, this is a deliberately extreme example, but it certainly shows that "at all but very slow speeds [vehicle length] is negligable" is not true.
> The Thinking distance is directly related to speed ie MPS / Tms
Only at a given degree of alertness. However, there is some tendency for drivers to become more alert a higher speeds (well, even more so when driving in conditions with a large _range_ of speeds.)
> The Breaking distance is based on the square of the velocity and above just a few KpH is the dominant factor.
Sorry, once again, this is not true. With good tyres, good brakes and a good road surface, braking can act as if sliding on a surface with a kinematic coefficient of friction of 1, or even slightly greater. At that point, the braking distance at (say) 10 kph is about half a metre. For the reaction distance to be less than that, would need a reaction time of under 0.14 s, which -- as mentioned early -- is pretty close to as fast as a it gets. For worst case reaction times of around 1.5 s, and the same excellent braking conditions, speed can get up to 65 mph before braking distance even reaches reaction distance. Once again, I have taken deliberately extreme values -- minimum on one side of the equation, maximum on the other -- but again I think this shows that "above just a few KpH [braking distance] is the dominant factor" is an exaggeration.
> So for any length of road the number of vehicles that can be safely accomadated is related to the dominant factor of S^2.
There are several problems with this. first, as we mentioned your claimed domination of v^2 (changed here to S^2) is exaggerated under a wide range of realistic conditions. Secondly, this ignores the fact that in a densely packed peleton, braking distance almost cancels out of the equation. Thirdly, we were talking about traffic volume, and for a single lane traffic volume is equal to speed divided by vehicle density, which drops you from a squared factor to linear even if braking distance was dominant; but since braking distance in fact nearly cancels out in most real situations, we end up with effective volume being _almost_ independent of speed. (But the time it takes you to get home is certainly not...)
There is a delightful guerrilla tactic that the street kids have fastened on over there....Plant yourself in the road in front of the camera, tie an empty steel can to a string and whirl it around. Tip speed (where the can is) exceeds the speed limit.
Camera takes many, many pictures of grinning urchin and runs out of film. Up the cops! Bless their little anarchistic hearts........
Schneier.com is a personal website. Opinions expressed are not necessarily those of BT.