Red Light Cameras

Red Light Still Cameras - Great in Theory, Not in Execution

Like anyone who drives, I know that people run red lights. Some people run them because they enter an intersection on a yellow which just happens to turn red before they can cross. Illegal? Yes, but not a big deal in my book, especially if screeching to a halt and getting rear ended was more dangerous. Lately, however, I've seen more and more people purposely run reds at turns -- typically by following the car in front of them that was entering on the yellow. Personally, I see this happen a lot at intersections that are very crowded, which tells me the problem isn't people who run reds, but rather a traffic light that's timed stupidly. For example, if there's a main entrance to a company -- say Apple or Google -- that most employees use, the line of cars waiting to make that turn might be so long that people are willing to run the light versus having to wait through 1, 2, or 3 more cycles.

Enter the red light still camera. In theory, I suppose they work great, but as anyone with any modicum of maturity and intelligence realizes, lots of things work well in theory only to fail miserably in practice. Personally, red light cameras are emblematic of something that's come to bother me more and more in recent years: a society where public policy is set by the infantile and the immature. Is safety important? Absolutely, but it's by no means the be all end all goal for which everything else -- like cost, fairness, and common sense -- should be sacrificed. And it sure as hell isn't something so important that we should tolerate a technological solution that works great in theory, but in the real world creates a financial windfall for a select few while penalizing the guilty and innocent with equal reckless abandon.

As an engineer, I can imagine that red light cameras are fascinating pieces of technology, but I still do not support their use for these 7 reasons that I just thought of:

Reason 1: Cameras Can't Exercise Judgment
A red light camera is dumb. It just shoots a picture when its proprietary and privately-owned and developed software algorithm says that a car has run a red. Has the car actually run the red? Maybe, but you'd have to be a supreme moron to think that software is without flaws. On the other hand, a police officer who monitors an intersection can judge when a light has turned red and whether it was excusable for the person to run the light anyway. For example, numerous studies (google "red light camera rear end accidents") show that rear-end accidents are more frequent when red light cameras are installed. Thus, while a person may have technically run the red, a cop can elect not to write a ticket if, for instance, they would have been rear-ended if they hadn't.

Thus, the argument can be easily made that all red light cameras do is trade one type of accident for another. Property damage still results. Injuries still result. And, of course, if the rear-end accident is severe enough, the rear-ended car is pushed into the intersection anyway, mowing down any pedestrians or cars that might be present. Some people argue that rear-end accidents are less lethal than the broadside accident that would otherwise result from a run red. I can see the logic in that argument in the abstract since there's more length to a car than width, but the funny thing about the abstract is that it doesn't matter in the real world.

First, the speed of the crash matters. If you're rear-ended while stopping suddenly at a red, I'd bet the car that hit you was going close to the speed limit of through traffic. In some cases, the cross-traffic vehicle will also be going pretty fast, but in many other cases, the cross-traffic vehicle was stopped at the light and is starting off from zero. In other words, it's going at lower speed. Is a high speed rear-end preferable over a low speed broadside? I'm not sure it is.
Second, the type and age of the cars involved make a huge difference. Broadside someone in a new car with a side airbag and they might be perfectly fine. Rear-end someone in an older car without a safety cage and they'd be anything but fine.
Third, the seating arrangement of people in the car matters. If the broadside occurs on the side of the car where no one happens to be sitting (e.g. passenger side impact on a car with a solo driver), the lethality goes way down. Contrast that to rear-ending a car with young children sitting in the rear crumple zone, er, I mean the 3rd row seat .

Anyway, here are 3.5 scenarios that I, personally, am not sure a red light camera is smart enough to differentiate:

First, let's say you pull into an intersection to make an unprotected left turn. You enter on a green light and wait for through traffic coming the opposite way to clear. However, traffic is so heavy that you're still waiting for it to clear when the light goes yellow. You want to make your turn, but, of course, you can't because there's always some car coming the other way who insists on running their red. Thus, by the time traffic actually clears and it's safe to make your turn, your light has already turned red. Did you break the law? Yes, you turned against a red. But under the circumstances was that the safest thing for you to do? Yes. Would a camera know the difference? I doubt it.
Second, let's say you're turning right on a red non-arrow light. Where do you stop? California Vehicle Code 21453(a) says you're supposed to stop "at the marked limit line" if there is one. Does that mean your bumper can't go past or front tires can't go past? Does a cop care? Sure, but the margin there is pretty big so the cop likely won't ticket your ass if your front wheels go 5 inches past the limit line. 36 inches over? Sure, he'll get you. Does a red light camera's software know the difference between 5 inches and 36? I doubt it. Go 5 inches over at 2 miles per hour or blow through the line entirely at 40 and the camera will most likely nail you all the same.
Third, here's a variation on the turning right on a red scenario: how long do you wait before making the turn? The last sentence of California Vehicle Code 21453(b) says that a driver must yield "until [he] can proceed with reasonable safety". A cop knows (or should know, hopefully) what reasonable safety means. Does a camera know? Probably not -- reasonability is inherently a judgment determination based on experience as well as the totality of the factual circumstance existing at the time of the act. This, of course, means that what was reasonable in one instance likely won't be 5 seconds later when traffic changes. The best we can expect of a camera, I think, is a line of software that says "if a driver stops at a red light for more than 2 seconds, it's reasonable. Anything under 2 seconds is unreasonable". Period full stop.
Continuing in that vein, here's another concern when deciding how long to stop before turning: camera response time. The cameras I've seen issue tickets with speeds (e.g. 5, 20, etc) which is, theoretically, the speed at which you allegedly ran the light. That speed is measured by some mechanism involving hardware and software machinery, like a radar gun. Machinery of any kind takes time to respond. What is the response time for the speed measurement on a red light camera? Because if it's anything sizable, that speed on the ticket is not the speed of the car in the photo. In other words, the car that was traveling at 8 mph (perhaps during deceleration) when the reading was taken could very likely have come to a full stop by the time the picture was taken.

Reason 2: Cameras Do Not Provide Immediate Notice of the Offense
If a cop sees you run a red light, he'll pull you over immediately. As a result, you know what you did because you just did it and can defend yourself accordingly. With a camera, you get a notice in the mail several weeks -- does it need to be this long? -- after the fact about an incident you likely don't remember at all. Instead of being proven guilty, the burden falls to you to prove your innocence which, conveniently, you can't because you don't remember what you did.

Reason 3: Cameras Lack Transparency
I'm not aware of any manufacturing, reliability, accuracy, or maintenance standards for red light cameras. For example:

How often do cameras give false positives or negatives? What's an acceptable false positive or negative rate?
Do cameras from different manufacturers perform similarly? For example, supposed city A used cameras from American Traffic Solutions and adjacent city B used cameras from a competitor, Redflex Traffic Systems. Would city A's cameras nail you for an act that city B's cameras wouldn't? (or vice-versa)
If there are standards, does the city or local police department actually follow them by, for instance, keeping up with software upgrades? Governments (and their agencies) are not known for being at all adept when it comes to technology. For example, consider the FBI and their botched Sentinel system from earlier this decade?

Lastly, the leasing arrangement (see below) between red light camera owner-makers and municipalities is governed by a contract. I've never tried it, but do you think that a member of the public can walk to city hall or the police department and get a copy of that contract? I doubt it. This matters because California Vehicle Code section 21455.5(g) states that red light camera contracts "may not include provision for the payment or compensation to the manufacturer or supplier based on the number of citations generated, or as a percentage of the revenue generated, as a result of the use of the equipment authorized under this section." Do municipalities follow this law? Quite a few do not (scroll down to Defect #10, Subsection B)

Reason 4: Cameras Disrupt Traffic
I used to drive limousines and, as part of that, spent a lot of time in Millbrae, the city just south of San Francisco Airport. The city of Millbrae uses red light cameras, particularly at the intersection of Rollins Ave and Millbrae Ave, where I used to turn right from northbound Rollins onto eastbound Millbrae against the red.

This particular camera goes off all the time, about once every 90 seconds day and night in my estimation. As a result, most drivers choose not to risk it and simply wait until the light turns green, even if it its clear to make the turn against the red. The line of cars used to stretch down Rollins to the next block (Adrian Road) and back up into the car wash and In-N-Out burger complex. There'd be lots of honking as yelling as well as someone who wanted to make the turn was inevitably stuck behind someone who didn't want to risk the ticket.

The irony here is that the system at Millbrae and Rollins actually covers all 4 corners. According to the Millbrae Police Department (I called them once), this means that the flash (which is what most people notice anyway) goes off whenever (a) through-traffic runs a red, or (b) any traffic turns against a red, including right turns on red at any of the 4 corners. The result? People were paralyzed by camera flashes that likely weren't even intended for them.

Reason 5: Cameras Encourage License Plate Crimes
Obviously, red light cameras work by snapping a picture of the offending car's license plate. However, if a car has no license plates, then a red light camera is pretty useless. Even though it is technically illegal in most states (I think) to drive a car without license plates, I imagine the penalty for doing so is less than that for running a red light. Is this an incentive to drive without a license place? I think it is.

There are limited circumstances (e.g. brand new car, etc) in which driving with no license plates is okay, which is perhaps why, to deter the police, license plate theft is more common now, particular in cities that use cameras to enforce congestion charges.

And I haven't even mentioned the plethora of license plate obfuscation devices (e.g. covers, sprays, etc) that all claim to prevent the camera from snapping your plate. Coincidentally, license plate covers that cover all of your license plate (the number and the registration stickers) are illegal in California under the last sentence of California Vehicle Code section 5201(f)(2). Are these covers all over the place? Absolutely.

Reason 6: Cameras Create a False Sense of Security
This one is more of a personal pet peeve: red light cameras create a false sense of security. I as a driver don't need to take personal responsibility for my own safety by being actively aware of my surroundings because I can just trust the camera. I can just turn off my brain and drive straight through any green light like an ostrich with it's head in the sand without looking first to see if the intersection is clear. (Seems to me that this would violate every driver's duty to operate their vehicle in a safe manner). Our society needs more personal responsibility, not less.

Obviously, I won't come out against safety. But next time you see a red light camera, take notice of what type of intersection it's placed at. Is it a sleepy uncontrolled intersection in a residential neighborhood where someone might walk or drive into an intersection without noticing a car? Most likely not because of simple economics: the camera won't make any money there. More likely than not the camera is at a busy intersection where there's lots of pedestrian and car traffic. In other words, it's an intersection where you as a driver or pedestrian should be paying attention anyway.

Reason 7: Where does the fine go?
Shockingly, the answer is not to the municipality. At least not all of it. For some reason, a lot of -- if not all -- red light cameras are leased instead of sold and the company that made and owns the red light camera gets a share of each ticket generated. Think it's a small share? Congratulations, you're an imbecile. For example, for the cameras in the city of San Rafael, CA, that share is 28%. That's not a typo -- each red light ticket averages $500 and the city of San Rafael gets $140. Now I'm the first to admit court fees are not cheap, but 72% not cheap? I don't think so. Here's a clue as to where that 72% goes to: the owner of the red light cameras in San Rafael -- Redflex Systems of Scottsdale, AZ -- gets a monthly rental fee of $5,900. Per intersection. San Rafael started their program with 10 intersections monitored, meaning a roughly $59,000 monthly income stream -- $708,000 per year -- for Redflex. And that's just for one city. Of around 58,000 people.

Remember that the next time someone says "Red light cameras aren't about money".

(As an aside, Redflex is part of a company called Redflex Holdings Limited which is publicly traded on the Australian Stock Exchange under the symbol RDF)

From the municipality and camera company's perspective, it's fantastic. The municipality doesn't have to pay to put cops on traffic duty. Not only that, but it can sit back and literally watch the money roll in $140 at a time (in San Rafael's case). The camera company reaps a huge and steady income stream of $5,900 per intersection per month. Private companies profiting off of illegal behavior. Interesting. Can profiting off murder and rape be next?

Onto something else, camera leasing. Why not just buy the camera outright? It's a stand alone piece of equipment. Yes, it has hardware and software but so do a lot of other things -- your car, your computer, your cell phone, your digital camera -- and you don't lease those, right? (well, some people the car, but not the latter three usually). As far as I know, there are two reasons why you lease something: (a) you can't afford to buy it outright, and (b) a newer version will come along shortly (i.e. the product has a short life cycle relative to its cost). I don't know what camera systems cost -- shockingly, the makers don't publicize that information -- but nothing comes to mind about why they should be so expensive as to be unaffordable. It's a red light camera, after all, not a helicopter, fire engine, or ambulance. The life cycle argument doesn't go very far either. Did people run red lights so differently 5 years ago versus today that camera software from back then won't work today? I doubt it. Will people run red lights so differently 5 years from now that camera software from today won't work then? I doubt that too. Unless the underlying event (i.e. what constitutes a run red) changes due to governing statute, it seems the software and hardware of today will work just fine.

Assuming this doesn't happen though, what's the upgrade need? None that I can see, but then again, I'm only using common sense.

Common Arguments In Favor of Red Light Cameras That Are Not Convincing

My brother/mother/sister/friend/neighbor/son/daughter/barber/roommate/dog was hit and injured by someone who ran a red light.

You know what? Car accidents will happen, regardless of whether red light cameras are used because the fundamental physics of driving involves big heavy objects traveling at high speed in close proximity on trajectories that take them into the same space in sequence. When that happens regularly, they sometimes crash and momentum takes over. Thus, someone's brother/mother/sister/friend/neighbor/son/daughter/barber/roommate/dog will always be getting hurt in an accident of some kind. People will continue to drive automobiles so, as a result, no one but you cares.

I'd submit that the more effective solution is not to issue tickets based on a faulty device, but rather to (a) to drive defensively assuming everyone else on the road is a drunken idiot trying to kill you, (b) pay attention and watch both sides of traffic whenever you cross the street, and/or (c) buy a safer car.

Lastly, I'll say what everyone who hears this argument is already thinking: People who's brother/mother/sister/friend/neighbor/son/daughter/barber/roommate/dog was hit by a red light runner always make arguments based on emotion. Emotion-based arguments are neither credible nor convincing, and all the less so when coming from a biased source. We're sorry your brother/mother/sister/friend/neighbor/son/daughter/barber/roommate/dog was hit, but it happens.
Just don't run lights.

Such a view fails to distinguish between people who run reds on purpose and those who run them by accident. I think most reasonable people would say the former deserve to be ticketed while the latter do not. How does one run a red by "accident" you might ask? I'd say a very common way is coming up to a green light only to have it suddenly turn yellow on you. The dilemma then becomes do you (a) brake suddenly, or (b) speed through the intersection?

The simplistic among you may say braking suddenly is the thing to do, but remember, sudden unexpected actions -- like braking suddenly -- can catch other drivers off guard, thereby leading to accidents too. Also, even if you do brake suddenly there's no guarantee you'll stop before entering the intersection which could set off the red light camera anyway. And, of course, if you decide to speed through the intersection, you could make it just fine or you could hit another car or a pedestrian. So both choices could work out either very well or very badly and remember, you have only a split second to make the choice.

On a more personal note, people who subscribe to the overly-simplistic fantasy view of "Just don't run lights" are utterly useless with no ability to comprehend or process the shades of gray that accompany life in the real world. In their fantasy world, diets always work, everything goes according to plan, and nothing is ever late or over budget.

Bottom line: unless you're a moron, it's always easier to talk about something in the abstract than it is to actually deal with it in the real world.

If you believe that red light cameras only catch bad drivers who habitually speed and run red lights, you understand the real world even less than I thought. As I said above, red light cameras are triggered by some combination of software and hardware meant to replicate an officer's eyes and brain. The camera has no sentience or autonomy. It can't make a judgment about whether running the light was justified in a particular circumstance. For example, from April to December 2007, speed camera in Montgomery County, Maryland, issued 224 speeding citations... to police cars. 76 of those 224 (34%) tickets were dismissed when it was discovered that they were issued to on-duty police officers responding to emergency calls. All the camera knows is what's programmed into its software which is -- in crude terms -- something like this: 'if a car is in the intersection after the light turns red, take a picture.' And that's assuming the camera is working correctly and aimed so that all relevant facts are in the camera's limited field of view.

Not surprisingly, the two biggest red light camera makers in the US -- Redflex and American Traffic Solutions -- both make speed enforcement cameras as well as red light cameras, which tells me that the software for the two types of cameras is not that dissimilar.

Alternatives to red light cameras

Be observant. If you see red light running happening a lot at a particular intersection, here's an idea: call the damn police, that's what you pay taxes for. They'll (eventually) send a cop to watch for runners. Like any type of policing, traffic enforcement requires the community's involvement. However, know that calling the police after the fact to report a particular red light runner is pointless. The cops need to see the offense first and cannot issue a ticket based on just your word. Also, you must accept that people will always run red lights, just like they will always commit fraud, robbery, murder, and assault.
On a related note, here's another idea: if you see a yellow, instead of braking suddenly or speeding up, slow down instead. Something I see all the time is someone speeding (usually through a turn) to catch a light that is just turning yellow or red.
Be a more defensive driver and assume that everyone else on the road is a drunken idiot out to kill you. Expect that someone will run the red and -- oh, I don't know -- look both ways before entering the intersection, just like how your parents taught you as a child to cross the street.
If you're particularly daring and you know someone is going to run the red light, you can enter the intersection anyway (being prepared to brake, of course) and honk really loud to scare the other driver when he or she crosses your path. People tend to remember what scares them.
If you insist on using technology, put a video camera at the intersection instead of a still camera. Start shooting video when the light turns yellow, while the light is yellow, and for a few seconds after the light turns red. This could theoretically show mitigating circumstances a still photo doesn't. Unfortunately, you'd have lots of short (~8 seconds) video clips since lights cycle thousands of times a day. (Note, I never said this was a perfect solution).

Alternative Low-Tech Solutions:

Solution #1: Paint a yellow line on the road leading up to the red light camera.

I know it doesn't seem like it, but traffic lights are not programmed randomly. In California, vehicle code section 21401(a) requires that traffic control devices (e.g. a traffic light) must conform to standards promulgated by the California Department of Transportation (aka Caltrans). If you look at tables 4D-50, for example, (see page 750 of the 72.7MB "California MUTCD (Entire Document)" link), you'll discover that yellow lights must be timed so that someone traveling the posted speed limit can stop if the light turns without running into the intersection. Thus, if you know how long a light will be yellow (say 3 seconds) and what the posted speed limit is (say 35 mph or 51.3 feet/second), you can work out what distance a car with well-functioning brakes will travel in the time the light is yellow (154 feet). Paint a yellow line at that distance.

Thus, if you're driving that 35 mph speed limit and you see the light ahead turn yellow, look to see where that yellow line is.

If it's behind you, you can proceed through the intersection and be guaranteed not to get a ticket because you entered while the light was still yellow.
If it's still in front of you, you won't make it through the intersection before the light turns red (unless, of course, you speed up) so brake normally and you should be able to stop just fine.

But then again, this would improve safety without increasing revenue and we can't have that, can we?

As an aside, some cities decided to shorten the duration of the lights at some intersections while simultaneously installing red light cameras at those same intersections. Result? The tickets so issued were illegal and the cities had to refund the fines collected. In 2005, Union City, CA was forced to refund over $1 million in fines when a motorist proved that the yellow lights at the monitored intersections were too short. And that leads in to...

Solution #2: Have longer yellow lights.

The intersection of Mohave Drive and Mission Boulevard in the town of Fremont, California, used to generate about 20% of the red light camera ticket revenue in Fremont, roughly $900,00 out of $4.7 million. This was the most profitable intersection in all of Fremont. I say was because when the traffic lights at the intersection were programmed to give yellow lights that were 0.7 seconds longer, traffic violations at Mohave and Mission dropped. By 81%. (Link)

Solution #3: Use a timer like in the picture below.

It works just fine for pedestrians.

Last edited: Sunday, May 15, 2011 01:10:56 PM