January 18, 2007
On red lights and yellow lights

I received the following email from Helwig Van Der Grinten, who left a comment on this post. I thought it was worth sharing and commenting on.

In the discussions about red light cameras I've read in your blog and elsewhere, it seems that everyone is ignoring the fact that all drivers must make a snap decision when a light changes from green to amber. Cameras do nothing to help a driver make a correct one. They merely penalize an incorrect guess.

According to a report of the Virginia Transportation Research Council published in January 2005 (http://www.thenewspaper.com/rlc/docs/05-vdot.pdf), of 33,547 citations (I added them up from Table 8) issued for red light violations a very large majority (85%) were issued for vehicles that entered the intersection within the first two seconds after the light turned red. This indicates that a vast majority of red light violations were not deliberate. Most violators are just poor at judging the how much time they have before the light turns red. The report also indicates that the number of citations decreases after the cameras have been in place for a few months obviously because the risk associated with a driver's bad guess has increased by the threat of a ticket. Unfortunately, the incidence of rear-end collisions also increased because inexperienced drivers are more likely to jam on the brakes in a panicked effort to avoid getting a ticket. What is lacking is a clear way for drivers to make a correct decision to either stop or go when seeing a green light turn amber.

Here is a simple solution to this problem. This will enable a driver to accurately judge both time and distance so that there will be no doubt about whether he or she should continue at the speed limit or come to a stop when seeing a green light turn amber.

It's in two parts. First, the amber time must set be in accordance with the formula agreed upon by traffic engineers that pegs amber time to the speed limit and the slope of the road at the intersection. The higher the speed limit and the greater the down hill slope is, the longer the amber time should be. The formula even allows one second reaction time for a driver to realize he or she should stop. Secondly, taking this standard amber time into account, it is simple arithmetic to calculate where a yellow stopping distance line should be painted across the road. This line would indicate to a driver who sees both an amber light and the yellow stopping distance line ahead that he or she should use a normal braking rate to stop at the intersection. If the driver sees that he or she is beyond the yellow stopping distance line when the light changes form green to amber, it is safe to proceed through the intersection at the speed limit and not run the risk of violating the red light.

I propose to found "The Houston Coalition Against Red Light Cameras" to promote this solution and to oppose the unfair entrapment of innocent drivers that is now being perpetrated by all of the proponents of red light cameras.

Lets put an end to the time and distance game that all drivers are forced to play with all the odds stacked against them. STOP THE PHOTO-RED SCAM! What do you think?

I think Helwig is far more lenient on red light runners than I would be. I very seldom find myself in a position where I'm not sure whether I should brake or accelerate when a light turns yellow; the times when I do, it's usually because I'm going faster than I should. I don't think this is a "guess" that people are making, I think it's plain old impatience. Believe me when I say I have a lot of sympathy for that particular emotion, especially when driving. But if we all did what they taught us in Drivers Ed (and which they reinforce ad nauseum in Defensive Driving classes, a curriculum with which I'm familiar thanks to that impatience I mentioned previously) and made it a habit to slow down and be prepared to brake when the yellow light appears, we wouldn't be guessing and we wouldn't be running red lights.

I don't think it's a surprise that most red light runners do so within the first two seconds of the light turning red. I mean, when else would it happen? This is like saying that most speeders are caught doing less than 15 MPH above the limit. I don't buy the argument that these are not deliberate. Have you never hit the gas to make it through the light before you lose the yellow? I'd bet that's what a lot of these folks are doing.

That said, I've no quarrel with the suggestions Helwig makes. By all means, optimize the yellow light times, and paint yellow stopping distance lines (I'm more skeptical of these, as stopping distance is a function of speed, which is not going to be a constant at these intersections). I'd go farther and say let's make sure every camera-enabled intersection has Walk/Don't Walk indicators as well, since they provide an extra measure of how much time you have till the light turns on you. Heck, I've even seen them with timers, counting down the seconds in the cycle. You can even add signs warning people about the presence of the cameras, which some people are oddly enamored with, if you think they'll do any good. If all that will help, I say go for it. Use the revenue from the tickets generated to pay for any extra hardware needed.

If we do all that - if we make it as easy as possible for people to determine when they need to stop for the light and what will happen to them if they don't - can we then agree that the argument about red light cameras being nothing but revenue generators are no longer valid? I mean, at some point, we have to agree that it's illegal to run a red light, and that there may be consequences for doing illegal things like that. Right? Look at it this way: if all these enhancements really do cut down on red light running, then the cameras won't be such a gold mine after all. It's a win-win.

Believe it or not, I still haven't made up my mind about these things. I'm still not impressed by the revenue arguments about the cameras, and that seems to be the main point being brought up by their opponents. My concern is still about how the image data is stored and used, and what benefit (i.e., reduction in fatalities and serious injuries) we're getting out of all this. I'm still waiting to see some data from Houston's implementation. I'm perfectly willing to tinker with that implementation to address other concerns, but I see all that as a sideshow. This is what I want to know, and I hope we'll know it soon.

Posted by Charles Kuffner on January 18, 2007 to Planes, Trains, and Automobiles

As I've mentioned before, if you install all the engineering solutions, no company in its right mind would pay out of its pocket to install the cameras as is happening in most cities. Cameras give cities incentives to MAXIMIZE red light running to boost revenue. So you can say do both, but no one does and it's not economically feasible for the camera operators to do so.

Also, you can say you don't think people are guessing when the light will change, and maybe you don't, but all the transportation research contradicts your view. That's why researchers at A&M found a 40% decrease in red light running just by increasing yellow light times one second. Countdown timers virtually eliminate all but actually intentional red light running. Welcome back, btw.

Posted by: Gritsforbreakfast on January 18, 2007 7:08 AM

A bigger problem is some confusion about what you're supposed to do. When I learned to drive, I was taught that a yellow light meant that you should stop if it was possible in the distance between you and the light. What most of us actually do is go if we think we can make it. And, in reality, most drivers are including the first second or two red in the "if I can make it" interval.

Obviously, yellow lights have to be timed correctly, so that at the posted speed limit, it allows someone closer than the stopping distance at the speed limit to get through it before it turns red. Of course, if someone is driving 10 mph over the limit, they won't be able to stop even if they're at that distance (because their stopping distance is longer).

Funny thing is, I almost never encounter yellow lights that are clearly too short.

Which is why I'm ultimately highly unsympathetic to people who "misjudge." Seems to me in the majority of cases, the "misjudgment" is speeding and not stopping if they can. It's hard to argue that a ticket is unfair in that case.

Posted by: John on January 18, 2007 7:10 AM

I think Helwig is far more lenient on red light runners than I would be.

True, but you may not be the "average" driver. Your intelligence alone places you at least several standard deviations above the average. I'm not saying I necessarily buy into the argument, though.

As for the revenue argument -- I would hope you'd at least be concerned about the incentive for the companies operating the cameras to tweak the red light times, an incentive directly related to "the revenue argument." We all know that Houston's way of doing both political and commercial "bidness" sometimes isn't as transparent as we'd like. :)

Posted by: kevin whited on January 18, 2007 7:51 AM

"I very seldom find myself in a position where I'm not sure whether I should brake or accelerate when a light turns yellow." You just said a mouthfull. How often is seldom? And do you want to risk $75 for the privalage of not being sure? Big brothers Bill White and Harold Hurtt think it's your fault for being uncertain. Multiply that uncertainty by the number of drivers in Houston and we've got ourselves a very subtle and profitable entrapment scam.

Posted by: Helwig Van Der Grinten on January 18, 2007 8:43 AM

I am a traffic engineer and am convinced that the cameras are primarily about money.

Cameras were supposed to be a last ditch effort after significant enforcement and engineering improvements at the intersection. They have become the first effort to combat RLR at an intersection for many cities.

A common argument is that officer enforcement is unsafe. Yet very rarely has the City installed enforcement lamps to facilitate one officer enforcement in a safe manner. It is a 150 dollar improvement that allows an officer to not have to chase a violator through the intersection, yet it is rarely used in some areas.

I can point to numerous cameras that are at locations without significant crash problems (as defined by crash rates, not frequency), cameras at intersections that are extremely congested. Raw # of RLR Violations are directly related to the volume to capacity ratio of the approach volumes.

Most violations occur during the first 1-2 seconds of red, most crashes occur after 6 seconds of red.

The federal highway administration has stated that camera enforcement should not be used during the all-red portion of the signal phase. This is usually a 0.5 to 2 second period after the yellow change interval. This time varies by location and in many places may actually be set to 0.

Contrary to popular belief there is no formally recognized recommended formula for setting yellow times. There is a formula promoted by the Institute of Transportation Engineers that is for the Change Period (Yellow + All Red) and any folks use the common convention that the reaction time + first term of the equation = yellow and then the second term = all red. However this is not always the case.

Yellows are also frequently set based on the posted speed, even in locations where the engineers and cops know that the majority of drivers are exceeding the speed limit. This has the effect of creating a yellow that is too short.

My biggest beef with the rush to camera enforcement is that in most instances, the traffic engineers and law enforcement have NOT worked together to develop other solutions before implementing cameras first.

Posted by: Trafficnerd on January 18, 2007 9:19 AM

Van Der Grinten's comments also suggest that we add an orange light between the yellow and red light with appropropate justifications for interpretation, similar to the Christmas Tree at the drag strip.

Posted by: Charles Hixon on January 18, 2007 11:28 AM

I've posted about this before, too:


I'm a bit mystified at Scott (grits)' refusal to take seriously the argument that there's some inherent value in ticketing people who decide, ON PURPOSE, to run red lights. He in turn is seemingly mystified by my desire to do so. I don't know how to bridge that divide, and I get the sense you're more on my side than his...

I suspect one's position on this stems from where one lives (or learned to drive). In South Florida, it got so bad that the "engineering solution" had migrated to an all-red cycle of two or more seconds. That means the capacity of the intersection was being drastically degraded for the benefit of red-light runners (who, again, were doing so obviously on purpose).

That's what will happen if you lengthen yellows instead of ticketing the bad drivers, folks. Those willfully bad drivers will just push that envelope further out.

Posted by: M1EK on January 18, 2007 11:36 AM

Hi M1EK! :)

Mike and I have been around about this quite a bit before. Needless to say he thinks red light runners are all evil-doers who need punishment. To me, when engineering solutions work as well or better to reduce accidents, and they do, there's no need to resort to criminal (on in the cameras case civil) penalties. Best,

Posted by: Gritsfobreakfast on January 18, 2007 6:00 PM

Yeah GritsforBreakfast!

Posted by: Support Science to Reverse Global Warming, which Al Gore says is still possible on January 18, 2007 8:18 PM


There you go again.

I believe that citing red light runners leads to less red light running than the engineering solutions will. (The engineering solutions have been in place for a decade or two in South Florida, and all that happened is that the purposeful red-light runners - the majority - run even farther after their light turns red because they've learned that there's a 2 or 3 second all-red cycle).

Posted by: M1EK on January 19, 2007 9:19 AM


The all-red portion of the signal phase is not frequently used as an engineering solution to reduce red-light running.

An overly long all-red can (and does) reduce angle crashes at some locations. An overly long all-red at an oversaturated intersection can induce more red light running because of its impact on capacity and motorist expectation.

Posted by: Trafficnerd on January 19, 2007 12:48 PM


South Florida's traffic engineers first tried longer and longer yellow cycles. Didn't work; the people who were running red lights ON PURPOSE figured it out and adjusted accordingly.

Posted by: M1EK on January 19, 2007 3:58 PM

Charles, can we persuade you to now include red light cameras on your index? There's lots of good stuff here.

Posted by: Helwig Van Der Grinten on January 20, 2007 3:06 PM


Where is your evidence of the South Florida Traffic Engineers and increased red light running with longer yellows?

I'm studying to be a traffic engineer,and never heard any of this.

Posted by: rte on January 24, 2007 2:22 PM

"Research has consistently shown that drivers do not, in fact, adapt to the
length of the yellow."
Determining Vehicle Change Intervals - A Proposed Recommended Practice", ITE,

The Texas Transportation Institute reported in 2004 that their research showed
that when the yellow change interval was 1 second above ITE minimum reduced red
light violations violations by 53%. When the Yellow Change interval was 1
second below ITE minimum red light violations jumped 110%. They also reported
that increasing the yellow interval 1 second above ITE minimum yielded a 40
percent reduction in red light running crashes.

Richard Retting et. el reported in 1998 that almost 80% of red light entries
occured on the first second of red at the intersection he studied.

"Increases in the length of the yellow signal toward values associated with the
ITE-proposed recommended practice significantly decreased the chance of
red-light running. (Page 2.)"
Richard Retting

The Office of the House Majority leader reported in 2001 that "Today's formula
for calculating yellow times yields yellow times that can in some cases be
about 30 percent shorter than the older formula." The older formula being
calculated from the 1976 ITE Traffic Engineering Handbook

Posted by: rte on January 24, 2007 2:29 PM


It's anectdotal experience from growing up there - a long time ago by now; but the all-red cycles were, in fact, promoted as an effort to finally get the red-light runners to stop causing bad accidents (after trying for years to 'solve' it with other engineering solutions).

Pre-WWW, though; so there's no chance of finding a citation - you're just going to have to trust me on this one.

Oh, and the problem wasn't "and increased red light running with longer yellows?", the problem was that drivers DID adjust to the longer yellows and CONTINUED running red lights (after an adjustment period). As with Kuff, I'm less than impressed with the statistic that most runners do so in the first second after it turns red - big whoop; they're mostly still doing it on purpose, and they can therefore just as easily do it on purpose after an even longer yellow.

Posted by: M1EK on January 24, 2007 5:53 PM


Your letting your biases and experiences dictate your answer, and then seek to protect your hypothesis.

All-red was not an answer to red-light running problem.

All-red is an optional interval used by an engineer to clear the intersection with vehicles who have legally entered the intersection.

Posted by: rte on January 24, 2007 11:06 PM

Controlled scientific studies seem to prove your anecdotal evidence as incorrect.

Research has consistently shown that drivers do not habituate to the duration
of the Yellow Clearance/Change Interval.

"Research has consistently shown that drivers do not, in fact, adapt to the
length of the yellow."
Determining Vehicle Change Intervals - A Proposed Recommended Practice", ITE,

"It has frequency been claimed that if the yellow is "too long", more drivers
will use part of the yellow as green. More drivers - it was argued - would
cross after the yellow onset with long [RATHER] than with short
yellow."........"The data show that the percentage of last-to-cross vehicles
clearing the intersection (T+0.2) seconds or more past the yellow onset was not
appreciably changed by the extension of the yellow phase." ["The Influence of
the Time Duration of Yellow Traffic Signals on Driver Response",
Stimpson/Zador/Tarnoff, ITE Journal, Institute of Transportation Engineers,
November 1980, page 27]

"The percentages of these vehicles, that is of vehicles that could have been
involved in a conflict with cross-street traffic, were substantially smaller at
both sites and under all conditions after the yellow duration was extended. No
evidence was found at either site, under any of the conditions, that the
vehicles that were in potential conflict with cross-street traffic with the
extended yellow would have cleared the intersection earlier in the cycle if the
yellow had not been extended. Thus, the extensions of yellow duration employed
in this study substantially reduced the frequency of potential intersection
conflicts." ["The Influence of the Time Duration of Yellow Traffic Signals on
Driver Response", Stimpson/Zador/Tarnoff, ITE Journal, Institute of
Transportation Engineers, November 1980, page 28]

Posted by: rte on January 24, 2007 11:10 PM

Also something else,

In the vast majority of states, the yellow signal serves as a warning signal.

You do not have to stop for yellow, and it is legal to enter the intersection so long as it is yellow.

Legally Red is the indication that means STOP.

The problem is that when the Yellow light is too short, a motorist will find themselves in an awkward position at a certain region of the intersection.

If you brake abruptly you risk getting rear-ended, if you continue at the same speed,you risk running a red light, and if you gun-it you risk rear-ending someone.

I have been in that situation,and at an improperly timed intersection it is not pretty.

The options a motorists faces in this situation are limited and unsafe.

However, a good engineer will correct the yellow timing so as that a motorist will never face that impossible choice of whether to stop or go.

Posted by: rte on January 24, 2007 11:29 PM


No, the (quite long) all-red was an effort to clear people who were purposefully running the red (typically, on left-turns at least, as many as four or five cars would BEGIN their turn after their light turned red).

Again, I was there; I lived through the change; I read about it in the newspaper and watched it on TV. It wasn't me 'assuming'; that was, in fact, the publically stated reason for the change.

As for the legal implications of yellow - this may vary from state to state; but we were all TAUGHT "yellow means stop if you can safely do so; otherwise proceed".

Posted by: M1EK on January 25, 2007 10:16 AM

Well those scientific studies prove you incorrect.

Who is going to school to become a traffic engineer?

The purpose of all-red is as a clearance interval.

"The primary measure of effectiveness for the yellow interval is the percent of
vehicles entering the intersection after the termination of the yellow
indication; that is, during the red following the yellow." ...... "When the
percent of vehicles that are last through the intersection which enter on red
exceeds that which is locally acceptable (many agencies use a value of one to
three percent), the yellow interval should be lengthened until the percentage
conforms to local standards." ["Determining Vehicle Change Intervals - A
Proposed Recommended Practice", Institute of Transportation Engineers,
Washington, D.C., 1985, page 6]

Posted by: rte on January 25, 2007 2:42 PM

Well in Florida, where you speak of, it is legal to enter an intersection on yellow.

It is DANGEROUS to jam brakes for yellow, that is why I proceed.

I had a friend the other day who jammed her brakes at a red light camera intersection, and see get slammed and injured pretty badly.

Only two states have restrictions on yellow, but they are practically unenforceable. That is Michigan, and Virginia.

Florida Statues:

316.075 Traffic control signal devices.--

(b) Steady yellow indication.--

1. Vehicular traffic facing a steady yellow signal is thereby warned that the related green movement is being terminated or that a red indication will be exhibited immediately thereafter when vehicular traffic shall not enter the intersection.

2. Pedestrians facing a steady yellow signal, unless otherwise directed by a pedestrian control signal as provided in s. 316.0755, are thereby advised that there is insufficient time to cross the roadway before a red indication is shown and no pedestrian shall start to cross the roadway.

Posted by: rte on January 25, 2007 2:57 PM

The defensive driving instructors are wrong in teaching to stop for yellow at all cost.

I used to jam, until I realized how dangerous it was to do that.

I stop if I can safely, but if I can't,like I'm being tail-gated I go.

If the light is improperly timed that might mean clipping the light by a fraction of a second to a second, but it is better than getting mangled up from behind by stopping at all cost.

Posted by: rte on January 25, 2007 3:03 PM

By the way, how would extending red prevent red light running?

It is unintelligent to say that all-red would stop red light running. As one can not legally enter the intersection at that time.

So the Florida traffic engineers did not want to stop red light running, as all-red would not do that.

Posted by: rte on January 25, 2007 3:11 PM

Also another one, it is not illegal to be in an intersection on RED.

It is illegal to enter an intersection on RED.

Don't mistake people legally in the intersection on RED for red light runners.

Posted by: rte on January 25, 2007 3:23 PM


When you post five or six times in a row like that, it looks less like counterargument and more like "shouting down".

Again, in South Florida, the typical activity at the end of a green left arrow cycle is: arrow turns yellow, people keep going. Arrow turns red; up to four or five more cars ENTER THE INTERSECTION AFTER IT TURNS RED.

That's why they had to go to a fairly long all-red. Not because of people honestly entering on yellow who thought they could make it, or thought they couldn't safely stop in time.

Posted by: M1EK on January 26, 2007 9:22 AM

That sounds like an issue of congestion, the engineers need to tweak that turn arrow timing a bit instead of just adding all-red. That sound solve most of the issue.

Instead of posting one gigantic post,I just posted a number of small ones. Whats the problem, and the difference? They all contain solid evidence of a point.

Posted by: rte on January 26, 2007 1:19 PM

"That sounds like an issue of congestion,"

Yes. And there's no way to solve the congestion when you're talking about two roads, one with six through lanes and the other with eight through lanes; each with two left-turn lanes and a right-turn lane. Make the roads even wider? Sure.

It's the old "suburban sprawl doesn't scale" problem - and at this point, nothing but enforcing the law against red-light runners will stop the wrecks.


I completely understand the efficiency arguments. I'm an engineer myself. But lay off; there are a hell of a lot of people running red lights who are doing so because they know it will save them time; and the most efficient intersection in the world _still_ provides opportunities like that for people willing to put others at risk.

Posted by: M1EK on January 29, 2007 3:48 PM

M1EK your a transportation engineer? You ought read the ITE's "Tricks of the Trade". They have all kinds of solutions to congestion problems.

Red light running is a problem, but so are many other things. Its not our job, as engineers, to pad the towns budget, or to go out an apprehend violators. Were not cops. Our job is to apply our knowledge to make intersections safer.

If an intersection has a red light running problem, and it it is not because of an engineering issue, than coordinate with law enforcement.

Posted by: rte on February 1, 2007 5:52 PM

What do you define as red light running?

Most Americans could not define the illegal act of red light running correctly.

Posted by: rte on February 1, 2007 5:55 PM