Menlo Park Expands Red Light Photo Enforcement Program

IIHS red light running crash photo
Red light enforcement cameras are effective in reducing the number of injury collisions, according to the Federal Highway Administration. Photo: Insurance Institute for Highway Safety

Last week, the Menlo Park City Council voted to extend its red light photo enforcement program for an additional five years and add a fifth red light camera at the intersection of Bayfront Expressway and Chilco Street, where 20 collisions have resulted in 14 injuries and one fatality since 2008. A temporary test camera installed at the intersection on March 11 recorded 217 red light violations by drivers in only 12 hours.

On August 24, 2011, 64-Year-Old Richard Buckley was killed in a collision with a car while riding his bicycle across the six-lane Bayfront Expressway at that intersection during this lunch break. Buckley, who worked at nearby Tyco Electronics, began cycling for exercise after suffering a heart attack a few years earlier.

The intersection of Bayfront Expressway and Willow Avenue, the entrance to Facebook’s headquarters, was the site of two fatalities resulting from side-impact collisions in which drivers ran a red light.

In November 2009, 6-year-old Menlo Park resident and Laurel School student Lisa Xavier was killed when her family’s car was struck in the intersection by the driver of a Ford Mustang heading north on Bayfront Expressway who ran the red light. The driver, suspected to be the Mustang’s owner, local resident Shannon Fox, was never apprehended. In April 2007, Pulitzer Prize-winning journalist David Halberstam was killed as the passenger in a side-impact collision, after his driver, UC Berkeley graduate student Kevin Jones, ran the red light while turning left onto Willow Road. Jones was charged with vehicular manslaughter and sentenced to 200 hours of community service.

Lisa Xavier & David Halberstam portraits
Lisa Xavier (left) and David Halberstam (right) were both killed in side-impact collisions resulting from drivers running a red light at Bayfront Expressway and Willow Road in Menlo Park. Photos: San Jose Mercury News (left), San Francisco Chronicle / Michael Maloney (right).

A total of four red light photo enforcement cameras were installed a year after Halberstam was killed, one at Bayfront Expressway and Willow Road, two at El Camino Real and Ravenswood Avenue (facing north and south), and one at El Camino Real and Glenwood Avenue (facing north).

Heyward Robinson, who served on the City Council when the cameras were installed, described them as “a cost effective means of enforcing an important traffic law,” in an email to the current City Council. “They operate 24/7 with no salary, overtime, or benefit costs. The bottom line is that our roads and community are safer with these cameras than without them,” he wrote.

The rate of red light running has indeed dropped steadily every year since the cameras were installed, with the number of citations issued dropping from 6,381 in 2009 to 3,898 in 2012, a reduction of about 40 percent. The number of collisions that have occurred at the camera-enforced intersections dropped from 141 in the five years before the cameras were installed to 103 in the five years after they were installed, according to Menlo Park police.

The police believe the improvement to public safety has been even greater than the statistics alone indicate. “With the red light cameras in position, officers were able to work other locations in the city,” said Sergeant Sharon Kaufman. “It did seem to make a significant difference in the number of collisions.”

Menlo Park’s results are in line with a series of national studies on red light cameras conducted by the Federal Highway Administration (FHWA) and the Insurance Institute for Highway Safety over the past decade. The most recent and extensive such study, published in 2011, compared crash data from all 99 American cities with populations over 200,000, and concluded that the rate of fatal red light running crashes in the 14 cities with red light cameras was 24 percent lower than it would have been without cameras.

Council Member Ray Mueller requested that a stricter privacy policy be included in the city’s contract with RedFlex Traffic Systems, the Phoenix-based company that installs and maintains the cameras, so that photos or video taken by the cameras could not be used by any other agency than the Menlo Park Police Department unless obtained under a court-ordered subpoena.

“This is a two-for-one, these cameras are not just red light cameras,” Mueller said, explaining why he was concerned with the privacy implications. “These are 24-hour surveillance cameras that are used by our police to go ahead and investigate crimes in our community.” Police Commander Dave Bertini confirmed that the cameras have been used to investigate crimes in addition to red light running, including hit-and-runs, homicides, and burglaries. “We’ve been able to figure out a suspect vehicle that was fleeing the area,” he said.

Like Menlo Park, neighboring Belmont and Redwood City also signed five-year contracts in 2008 with RedFlex Traffic Systems, but those cities allowed their contracts to expire earlier this year.

The Belmont City Council voted 3 to 1 on June 11 to remove the city’s two red light cameras, both located on Ralston Avenue, where a car crash occurs once every five days on average.

Belmont City Council Member Dave Warden
Belmont Council Member Dave Warden called himself "a victim" of red light cameras because he was caught driving through a red light by one in Newark.

Council Member Dave Warden, who was caught driving through a red light by a camera in Newark five years ago, said that the cameras “aren’t solving any problem,” and felt that Belmont was potentially losing sales tax revenue from red light runners who might “feel unwelcome” in Belmont and refuse to shop in the city after receiving tickets.

Belmont Police Chief Daniel DeSmidt recommended that the city continue its red light enforcement program, but his report lacked compelling evidence that the cameras were effective, stating only that an average of 173 red light citations per month were issued. No collision data was provided.

Two red light cameras were also installed at Whipple Avenue and Veterans Boulevard in Redwood City, and  their effectiveness was reviewed by the City Council in October 2010. Sergeant Ashley Osborne explained that the number of red light citations issued per month had dropped by 19 percent and 30 percent at the two camera locations, respectively, and he credited the automated enforcement for that reduction. The number of collisions also decreased.

However, the Redwood City cameras were removed by Redflex when the city’s five-year contract expired this March. No further analysis of the system’s effectiveness was presented to the City Council, and the council was not asked whether it wished the cameras to remain operating.

Twelve of the Bay Area’s 101 cities and towns currently operate red light enforcement cameras: San Francisco, Oakland, San Rafael, Napa, Fairfield, San Leandro, Newark, Fremont, Daly City, Millbrae, South San Francisco, and Menlo Park.

  • Brad

    I will never understand the argument against an easy way to make our streets safer. Running red lights is illegal. Why have a cop there to write tickets when cameras can do it for you? Concerned about privacy? Don’t drive. As soon as you get out of your car, you’ll quickly understand why these red light cameras are necessary. I’d like to see it expand to speed cameras in certain zones as well. Like, around schools, and anytime a bike route gets dumped into a highway interchange.

  • Anonymous

    1. Motorists need to know about Snitch Tickets, the fake/phishing camera tickets sent out by the police to bluff registered owners into ID’ing the actual driver of the car. Over 30 California cities -including Menlo Park – use them, and in some cities the fakes are more than half of everything the city mails out. Snitch Tickets have not been filed with the court, so they don’t say “Notice to Appear,” don’t have the court’s address and phone # on them, and often say, on the back (in small letters), “Courtesy Notice – This is not a ticket.” Since they have not been filed with the court, they have zero legal weight. You can, and should, ignore a Snitch Ticket. Skeptical? Google: Snitch Ticket.

    2. Been to the LA area? A REAL red light camera ticket from ANY city in Los Angeles County can be ignored, as the LA courts do not report ignored camera tickets to the DMV. This was revealed in multiple LA Times articles in July 2011. Skeptical? Google: Red light camera no consequence.

  • Brent Sleeper

    Are you are an anti-red-light-camera posting robot, or just a dedicated AstroTurf salesman? 283 comments on Disqus, all boilerplate text on the same topic.

  • Anonymous

    The problem is that red light cameras cost money to buy and operate. They’re quite effective at changing behavior… but as this money comes from the inevitable traffic fines, the cameras eventually become unprofitable. Once they become a money pit, the cities let the contract expire and the cameras are removed.

    The solution that a number of cities (Union City comes to mind) have come up with is to shorten the length of the yellow cycle below the minimum threshold. This means that drivers going the speed limit won’t have time to react. The result is an increase in both revenue (tickets) and accidents. Until red light cameras are treated as something other than a profit center they will remain a scam.

  • Anonymous

    Are you the pot (a cop or red light camera company employee) calling the kettle “black?”

    You – and the author of this streetsblog post – should take the time to read the 20 page report that SaferStreetsLA wrote about the Menlo Park program. The report showed that before the cameras were put in there was no accident problem but there’s one now, due to a big jump in rearenders.

    Signed: Not a Robot, Not Selling Anything, But Definitely Dedicated

  • Anonymous

    SaferStreetsLA wrote about the Menlo Park program. The report showed
    that before the cameras were put in there was no accident problem

    Tell that to David Halberstam and Lisa Xavier

  • Anonymous

    Belmont was potentially losing sales tax revenue from red light runners
    who might “feel unwelcome” in Belmont and refuse to shop in the city
    after receiving tickets.

    Sorry Dave, I don’t shop in Belmont because your town is a bike unfriendly dump. Maybe if you did something about Ralston I’d drop by. Ever wonder why Caltrain keeps threatening to close your station while giving extra service to the two adjoining stations? Could it possibly be that you’ve provided crappy station access?

    Though you did build a nice bike path that gets me *out* of Belmont if I do end up at your train station…

  • mikesonn

    Exactly. I avoid Belmont Caltrain station like the plague. San Carlos all the way if I am headed north from RWC.

  • guest

    Camera’s alone will not solve the problem, they are just another reminder to pay attention and STOP! I dont mind the cameras… essentially if you’re not doing anything wrong, then dont worry about it. But seriously, do we really need to identify the driver?! Just send the ticket to the owner of the car. Surely they’ll pass it on or well tada, you found my stolen car! The driver should own up or the owner should really be worried about lending their car to that person…

    But I really wish yellow lights were a smidge longer, or timed with the posted speed of the roadway. I know people really dont stop or even slow at a yellow, but if people cant follow the rules you need to add more buffer, just like a momentary all red phase.

  • Anonymous

    The obvious, most amusing thing here is the giant protest against red light cameras from the same people insisting we need to crack down on cyclists running stop signs. “the cameras “aren’t solving any problem,”

    Yet we have 2 dead in just this article from red light runners in tiny Menlo Park.

  • Anonymous

    “But seriously, do we really need to identify the driver?!”

    You’re joking, right? You do realize it’s the *motorist*, not the car, that is at fault when a law is broken. It is crucial that the correct person receives the point on their license. Also, in the event a crime is committed, clearly the driver needs to be identified. That’s really great that you think the owner and whichever party was operating the car during the incident will work it all out, but that’s like saying we expect thieves to identify themselves when a shop owner notices something was stolen.

  • Anonymous

    Yep, gotta agree that the urban planning of Belmont, not just the Caltrain station, is a disaster. I worked for 2 years right on El Camino in Belmont, and it is about as unwalkable, unbikeable, and car-centric as a community gets. Old County Road is the only somewhat bicycle-friendly street in the whole city.

  • Brad, it is true that running red lights is illegal. But is also true that the standard for setting the length of yellow lights breaks the laws of motion–Newton’s Laws. The standard for timing yellow lights itself breaks a higher law. It is a higher law you and I cannot break and therefore we must break the local ordinances by no choice of our own. The discrepancy between the standard and the Laws of Motion is what red light camera companies exploit. Imagine a law forbidding gravity. That is what we have. Everyone is guilty. Unfair and dangerous. The yellow lights are up to 5 seconds too short.

    The length of the yellow light, by federal standard, is half the time it takes to stop your car. It is the red light cameras that expose the magnitude of the underlying problem. People do not want to run red lights. No one wants to commit suicide or hurt others.

    Google “Derivation of the Yellow Change Interval Formula.”

  • Anonymous

    Well… I’m a physicist and your claim is total B.S.

  • tknelson: Yellow time = v/2a. That’s the standard. Because you are a physicist then you should know that “2” means drivers get half the time to stop from initial velocity v when decelerating at constant rate a.

  • Anonymous

    Complete rubbish. You have no idea what you are talking about.

  • Common Sense

    Who wrote this “article?” It looks to me a like a not-so-cleverly press release from a scamera vendor disguised as “news.” They always bring up fatalities (which are truly tragic) to play on the emotional side of the argument. When it comes to red light scameras, one needs to focus on facts – not emotion. None of the above-mentioned fatalities would have been prevented by a red light scamera.

    Crashes caused by red light runners are those in which the driver is distracted, driving under the influence, or maybe even fleeing/eluding because they just robbed a bank or committed some other crime. A camera on a pole will not fix any of this. It’s quite simple, really.

  • Matt Laroche

    Spending some amount of money each year so that people don’t run red lights? I’m OK with that. (Not a big fan of shortening the yellow, but if we spend some small amount of money each year on problem intersections so that almost no one runs reds, money well spent.)

  • mikesonn

    How many red light runnings don’t result in a crash? Maybe some driver texting blows lights regularly and then finally hits someone. Let’s say that driver gets a ticket the first time they run that red, instead of just playing the odds and waiting for a crash on the forth time.

  • It’s so unfair to go after red light runners. They just want to run their red lights in peace. If an innocent victim dies, it’s just a tragic accident. Nobody could have predicted it!

  • Anonymous

    Read the report. It’s available on the blog thenewspaper daht com, at the bottom of the article dated Aug. 22.

  • Anonymous

    Here’s the real physics problem: Bicyclists want to mingle with cars, often with just a foot or two of separation, and expect to come out OK when there’s the inevitable contact.

    Well, the laws of physics say otherwise. Even the smallest mishap on a bike involves the rider being thrown to the ground. If you’re a bicyclist, you assume a risk of injury millions of times greater than if you were in a car. There is no way to create a perfectly safe place for bicycling. But I am working on my airbag equipped riding suit.

  • Anonymous

    Actually, that’s not the equation. It’s not just physics but also driver reaction time (physiology) that has to be considered. The actual formula used for calculating yellow light duration in seconds, y, assuming a flat road and a constant deceleration of 10 ft/s^2, is:

    y = t + 0.0735 v

    where v is the speed of the vehicle in mph at the moment it starts decelerating and t is the reaction time of the driver (usually 1 sec). See pg 2 of…/3%206r6_10.pdf‎

    This formula, for example, gives about 4 seconds at 40 mph.

    FYI, if you start googling per Mr. Ceccarelli’s own request, you find out he runs a website called redlightrobber

    and has made this argument before. It has been refuted in a report done for the Institute of Transportation Engineers:

    “Mr. Ceccarelli’s general thesis is that the yellow interval duration calculated using the kinematic
    equation supported by ITE is not long enough for vehicles to stop at the onset of yellow. His
    thesis is based on a misunderstanding of the yellow change interval—that this interval is equal to
    the time needed for a vehicle to stop before the intersection before the yellow signal indication

    He’s also been refuted here by the Insurance Institute of Highway Safety:

    As @tknelson:disqus said: “You have no idea what you are talking about.”

  • jd_x. Because you have been to my site, then please actually read my “Derivation of the Yellow Change Interval Formula”. Actually read IIHS’s paper. Then tell me what you think.

    While you are at my site, also read the paper from Dr. Chiu Liu, a physicist and civil engineer working at CalTrans HQ in Sacramento. He wrote a 2002 paper that was published in the ASCE Journal of Transportation Engineering I simply mimic which states the proper formula so that all traffic can legally move through an intersection. His formula mimics mine . . . or I should say, I mimic his. By knowing physics, both Liu and I came to the same conclusion.

    While you are at my site, read the “Dos” and “Don’ts” of the yellow change interval formula. This paper is a letter written by the inventor of the yellow change interval formula himself, Prof. Alexei Maradudin. He wrote this letter to the CalTrans Signal Devices Committee two months ago to encourage CalTrans to stop misusing his formula.

    When you complete these tasks which I fear you won’t, I ask this of you. Do the right thing. I do not care if you apologize to me, but you should openly acknowledge the laws of physics. Physics is an open book. Because I actually do have a degree in physics, I know what I am talking about. You and tknelson dismiss me too easily. I understand why. Neither of you understand physics, for if you did, both of you would immediately understand the implications of the algebraic expression v/2a. Perception/reaction time, tknelson, does not overcome the problem caused by v/2a.

    Or you can trust IIHS, an insurance company conglomerate that makes money off of raising premiums based on red light camera tickets, for your physics. Your choice.

  • Anonymous

    Dago Enrique,

    Any person who rear ends someone stopping for a red light should have his or her neck lengthened two vertebra’s worth.

    And that probably includes you, O apologist for crappy, irresponsible, selfish driving.

  • Anonymous

    Another Driving Mafioso, I see.

  • Anonymous

    I question authorship of the article. Andrew Boone? Seems to have fingerprints of Menlo Park Pd and Redflex on it.

    the article says, “The rate of red light running has indeed dropped steadily every year since the cameras were installed, with the number of citations issued
    dropping from 6,381 in 2009 to 3,898 in 2012, a reduction of about 40
    percent.” Most of the reductions are due to Caltrans adding a mere 1/2 second to the amber light at Bayfront and Willow. It was set (incredibly) at 3.0 seconds for a 700 foot pocket turn lane on a 50 mph road! 3.5 seconds is still too short, but in the 12 months following the change violations went down from1,600 (12 months prior) to just 400. Now credit is being given to the cameras for the 40% reduction in citations. This is pure baloney.

  • Anonymous

    The article uses 3 fatal accidents on Bayfront to justify photo enforcement. The devil is in the details. The little girl was killed when one of two cars drag racing at the time ran the red light. Photo enforcement has no influence on this type of behavior. Even signal lights do not seem to influence it.
    The 2nd accident cited as not really a red light running incident. The college student driving Halberstam to Palo Alto was in the through lane which had a green light. He, at the last moment, realized the correct route to P.A. was turn left onto Willow. He swerved from the through lane to turn left and collided with a vehicle (which had a green) and was coming from the opposite directions.
    The 3rd accident where bicyclist and motorist collided at Chilco and Bayfront was investigated by police. It was never determined if the bicyclist or the motorist ran the red light.

  • Anonymous

    The article states, ” The number of collisions that have occurred at the camera-enforced
    intersections dropped from 141 in the five years before the cameras were
    installed to 103 in the five years after they were installed, according
    to Menlo Park police.” Overgeneralized stats prove nothing. Read 20 page treatment which can be found at “”
    Jay Beeber studied data – intersection by intersection and did not cherry pick data to show the results the Police wanted.

  • Anonymous

    One of my favorite quotes from the article, ” A temporary test camera installed at the intersection on March
    11 recorded 217 red light violations by drivers in only 12 hours.” Yes, and over 95% were folks turning right. Has there ever been a right turning accident there? These types of accidents are rare and not usually severe. If this is really important, why not make the right lane a dedicated turn lane where only a “Yield” is required. This is a common configuration in many cities. I have never heard anyone complaint about accidents occurring because of this engineering countermeasure. If Menlo Park, gets its way, a RLC at Chilco and Bayfront can easily achieve the dubious distinction of sucking over one million dollars out of the pockets each year from folks who work or live in the Belle Haven Community where Chilco leads.

  • mikesonn

    As a pedestrian, right on red is a horrible thing. Next worse thing, rolling right on red.

  • Anonymous

    The article says, “add a fifth red light camera at the intersection of Bayfront Expressway and Chilco Street, where 20 collisions have resulted in 14 injuries and one fatality since 2008” I have the CHP data for 2008 thru 2011 and there were 14 collisions at Chilco and Bayfront.
    9 Rear enders (which cameras do not seem to reduce but to increase)
    2 were “hit object”
    1 Head On
    1 sideswipe
    1 other
    (of course the bicycle accident is included and we do not know if the bike or car was in violation)

    Also, NONE of the 14 were described as a result of primary collision factor of “red light running” (21453 cvc). Unsafe speed was cited as primary cause for 9 of the 14. The article uses info provided by Police. Police are merely spokespeople for Redflex. They are not presenting cold hard facts and analysis to the City Council as staff should do. They are promoting their program like Redflex promotes their product. I would question anything Andrew Boone (the author of the article) says.

  • Anonymous

    Here’s the problem with all the time you have spent on this issue (and now the time you are making me spend on it … arghgh): you keep assuming the car’s deceleration must be constant. That is the massive flaw in your whole argument. In the situation which is the basis for your whole argument, where the driver is in the “indecision zone” and they decide to “go for it” and try to get through the intersection even though they are (unbeknownst to themselves) farther than the critical distance, the driver just has to decelerate harder than the assumed 10 ft/s^2.

    We *all* have experienced this, and that is the nature of good driving: you can’t always predict distances and timing exactly right, so sometimes you have to stop a little harder than you would have wanted. (Note: a *bad* driver responds to this situation by still trying to make the light … which is why they deserve the ticket from the red light camera.) Good drivers don’t run red lights because they are 1) paying attention so that their reaction-perception time is truly 1 second or less, 2) driving at a reasonable speed which is at most the speed limit but can be lower based on less-than-ideal conditions, and 3) in the event they misjudge the distance (in the “indecision zone”) they respond such that they err on the side of caution and choose braking harder-than-ideal rather than blowing through a red light and risking people’s lives (and their own). Of, if said driver who is normally a good driver does run a red light, it’s because they dropped the ball on one of these 3 things and, even though it’s accidental, they still deserve the ticket (it serves as a reminder to pay more attention next time before your carelessness kills or maims somebody). One of the biggest problems with driving is exactly this problem: that motorists drive 4000 lbs of metal with 150+ horsepower literally at the tips of their toes while their senses are severely dulled and don’t take the responsibility of such a situation seriously.

    FYI, almost all cars are capable of decelerating at a minimum of about twice the assumed rate in the equations of 10 ft/s^2. See, for example,

    where even in the worst of situations, a pickup in wet conditions, the vehicle can decelerate at 18 ft/s^2 (note though, that in wet conditions, the driver should be traveling slower than the speed limit). In the case where the driver starts out decelerating at 10 ft/s^2 then realizes they misjudged the timing and they aren’t going to make it and so brakes harder, the declaration is non-constant (going from 10 to something higher, say 15). Or, you could take an even worse case and say the driver goes for 1 second beyond the perception-reaction time at their initial speed and calculate what deceleration you would need to stop in the critical distance. Using the example in the video you pointed me to where the perception-reaction time is 1 sec and the initial speed is 40 mph, that means that you would cover 118 ft in 2 seconds of going 40 mph. The critical distance is 232 ft, so now you only have 114 ft to go from 40 mph to 0 mph. Solving the equations of motion, you get that deceleration must be 15 ft/s^2. Though a little harsh, this deceleration is easily manageable. And again, here is where the good driver just brakes a little hard rather than run the red light. We don’t need to try and upend the whole methodology of traffic light timing based on decades of data because you don’t understand the assumptions that underlie these simplified equations.

    The point of all this is that, as is often the case, people are trying to appropriate a more noble cause (the yellow light timing is too short creating dangerous situations!) for their own petty cause (not wanting to pay a huge fine for running a red light). In reality, if people really gave a damn about safety (and not just getting away with red light running), they would be on the streets protesting *not* against red light cameras, but for longer yellow light durations (in the case where the duration truly is shorter than the recommended amount). And why are they waiting to protest the too-short yellow light durations only when the red light camera is installed and they start getting tickets? The hypocrisy and pettiness of it all, while simultaneously undermining a real threat to people’s safety (motorists running red lights), is downright petty, disgusting, and dangerous, not to mention a waste of everyone’s time and effort.

  • Anonymous

    If you search for the author you will find multiple articles covering San Mateo county for Streersblog. Either Redflex installed a mole many months ago for the purpose of planting this article or Streetsblog has a beat reporter.

  • Anonymous

    FYI, also check out this paper

    which, as part of the conclusions, states that *shorter* yellow-light durations *increase* the probability of drivers in the “indecision/dilemma zone” stopping rather than going through the intersection. More evidence that this whole yellow light durations issue is just a distraction and an excuse.

  • JD, I appreciate the fact that you are looking into the matter deeper than before. Now you believe my argument (and the inventor of the formula itself) is flawed because one should expect drivers to just brake harder. Now you disregard the federal standard regarding the maximum safe and comfortable deceleration rate that engineers can expect from drivers. You now oppose the FHWA, ITE and every study ever done on deceleration rates. While drivers do decelerate at different rates, their average deceleration rate is 10 ft/s/s as measured by the Wisconsin 2007 Study and adopted by ITE. The 85th percentile deceleration rate is 11.2 ft/s/s (85% do less than this rate under normal driving conditions, 15% do more). 11.2 ft/s/s was adopted by AASTHO and the FHWA.

    The original and follow-up papers regarding this formula do account for the possibility of varying deceleration rates of a single driver on a single approach. The conclusion of all these papers is this “average” decelerate rate–the one you find in the formula. There is another paper by Dr. Chiu Liu which elaborates on a non-constant “a”. His conclusion does not change the original premise. There is an effective “a” which is predictable.

    JD, the most fundamental rule of engineering is that designs must accommodate human behavior, not oppose it. For example an engineer does not put a brake pedal on the ceiling. Engineers set the rate to 10 ft/s/s because that is what can be expected of drivers.

    A passenger car can decelerate up to about 28 ft/s/s with ABS brakes engaged. But it is not comfortable for a driver to do this. His air-bag would engage just before his head would have hit the windshield. Decelerating at this rate is “safe” but it is not comfortable.

    Likewise field tests have shown that 10 ft/s/s is safe and comfortable for 50% of drivers. While half of the drivers will decelerate faster than this under normal circumstances, it is not comfortable. Doing so can cause a rear-end collision. The person driving behind such a driver does not expect the driver in front of him to decelerate that quickly. Do you see this facet of the problem? It is not just a matter of the capability of a driver’s brakes, it is a matter of what is comfortable and expected. There is a lot to say about perception/reaction times. The most thorough treatise on the topic I have found is in AASHTO’s Green Book. AASHTO concludes that 2.5 seconds should be a minimum of 2.5 seconds.

    For some perspective, 15 ft/s/s is the maximum deceleration rate a commercial trucker can do under a hazard. 2.5 s is the minimum p/r time for commercial trucker–truckers have to worry about their passengers and jackknifing. 8.0 ft/s/s is the safe and comfortable deceleration of a commercial trucker. These values come from the CDL manuals of every State. Commercial truckers require an extra minimum of 0.5 seconds for air-brake lag time. JD, no State in the USA other than Maryland considers the requirements of commercial truckers when setting yellow lights. In the end most States force school buses to run red lights. Do you like that? Is this what you want to justify?

    The problem with this formula remains that v/2a. That v/2a only models the driver who can proceed unimpeded at the speed limit through the critical distance and into the intersection. The formula does not model drivers, like turning drivers, whose slowing down before entering the intersection requires several seconds more (as demonstrated in the YouTube video). The time to proceed through the critical distance is t_p + v/a. The yellow light is supposed to give drivers the distance to stop *and* the time to proceed. Do not forget the “and.” Going is the other half of the meaning of the yellow light duration.

    I posted the original fomula paper at my website too. The one by Prof. Maradudin and Denos Gazis. “The Problem of the Amber Signal Light in Traffic Flow.” Read “Analytical Considerations.” on page 2. This paragraph shows the intent of the formula. Traffic engineers have been pulling it out of context since 1965.

    As for drivers beating the light, ITE itself says that the formula requires some drivers to “accelerate to beat the light” as they approach. (p. 756, starting at bottom of first column):

    And JD my motivation spending all my time on this is not what you assume. I do not live in California. I live in North Carolina where a red light camera ticket is $50. $50 to spend researching this issue is not worth my time at all. I hope that you give me the benefit of the doubt that I am not the arrogant megalomaniac such that I need to win at any cost. I tell you truth, from what I have been thought these last few years, it is not worth it at all.

    I do the research to help people like you. People are dying because of the misuse of this formula. The misuse of this formula has bigger repercussions that the red light camera industry its misuse created. It is the death and the injuries what primarily concerns me. If it wasn’t for those things, I would not persue this matter as relentlessly as I do. It is true that millions of people are getting ripped off from $50 to $500. It is the biggest charlatan fraud scheme ever perpetrated upon man. But that is secondary to people dying.

    One would like to put their faith in the traffic engineers whose lives we depend on. But I have found that such faith is misplaced. I took the NCDOT’s traffic engineers to task on this. On my website are 4 legal deposition from the official experts of NCDOT. Absolutely none of them knows what the ITE formula means. None! One one them put it this way, “We don’t know what the formula means. We just use it like a light switch. We turn it on in the morning.”

  • And now with real data, direct from Redflex, watch Tim J. Gate’s claim refuted.

    One can track engineering changes to the yellow light duration using the Redflex citation data.

  • Anonymous

    It was never determined… Menlo Park subbed out the investigation to SFPD…

  • SSLA

    JD, the the quote from the paper you reference does not mean what you think it means. It is simply a measurement of the likelihood of stopping, not the likelihood of running the red light. Those are two
    different things. Since I regularly correspond with the author of the paper, Dr. Tim Gates, I posited this very question to him:

    Me: Do you have a measurement of how much more likely they are to stop? Is it just a little more likely or a lot more likely? Is it linear, a curve? Your study is a measure of the likelihood of stopping, not the likelihood of running the red, correct? From all the data I’ve seen (and also collected) the likelihood of running the red goes up as the yellow time gets shorter as compared to the ITE equation. In contrast, the likelihood of running the red goes down as the yellow time increases as compared to the ITE equation (probably there is some maximum effect which I suspect is somewhere along a curve). So it would mean that with a shorter yellow, while some larger percentage might stop (per your findings), those who don’t stop (whatever smaller percentage that is) are more likely to run the red, correct?

    Gates’ Response: …myself and a colleague have recently proposed an alternative analytical approach using nested logit methods to simultaneously analyze the three driver actions (Stop, Legal Go, Red Light Running). What we have found is exactly what you
    describe – that shorter yellows (i.e., those less than 4.5 seconds) have 1) a slight, but significant, increase in stopping AND 2) a large, significant increase in red light running. The model we used controls for all other salient site related factors.

    So according to the paper’s author, it is not, “More evidence that this whole yellow light durations issue is just a distraction and an excuse”.

    This whole area of inquiry is based on probabilities. The unquestionable fact is that as yellow times increase (within reason) even beyond the time calculated using the ITE formula, red light running decreases. It’s not linear, as you get decreasing returns the longer you extend the time (but you still get a return at much longer intervals). There is likely a reasonable amount to extend the yellow interval that gives you a very good return (in decreased red light running) for that increase in the yellow interval. It’s somewhere in the range of 1 s – 2 s depending on your starting point.

    Understand also that in California our yellow times are shorter than the minimum as calculated using the ITE formula because jurisdictions are permitted to use the posted speed limit rather than the true approach speed based on the 85th percentile and are not required to adjust for the grade of the roadway which can have a significant impact on the minimum yellow interval that is needed. This is really a problem in hilly areas such as SF but even a downgrade as little as 3% or 4% can significantly impact the amount of red light running which will occur.

    These factors are at play both at intersections with red light cameras and those without. The changes need to be made at all intersections where the yellow time is insufficient. Unfortunately, where red light cameras are present, there is often a perverse
    incentive not to fix the problem because revenue is being generated.

    Regarding deceleration rates, while passenger vehicles are capable of decelerating at rates higher than 10 ft/s^2, the likelihood that drivers will choose to do so starts dropping pretty rapidly as you get much beyond that. By the time you get up to about 15 ft/s^2 the likelihood of stopping approaches zero. Most
    studies have found that when faced with a yellow indication, when the required deceleration rate was 8 ft/s^2 or less, most drivers (about 90%) chose to stop while when the required deceleration rate was 12
    ft/s^2 or more, very few drivers (about 10%) chose to stop. Again, this is a measure of what actually happens in the real world, not what we would like to have happen. If you want to protect the cross traffic, you have to engineer according to what is actually happening on the roadway, not what you would hope would be happening.

    As for some of the other points you made,

    “if people really gave a damn about safety (and not just getting away with red light running), they would be on the streets protesting *not* against red light cameras, but for longer yellow light durations (in the case where the duration truly is shorter than the
    recommended amount).”

    I agree that if you really gave a damn about safety, you would be on the streets protesting for longer yellow light durations. I hope you will join us in that as everyone will be safer as a result.

    “And why are they waiting to protest the too-short yellow light durations only when the red light camera is installed and they start getting tickets?”

    Because large numbers of tickets (i.e. violations) indicates there’s an engineering problem, likely a too short yellow and/or other deficiencies, that need fixing. The red light cameras help point out where the engineering deficiencies are. However, as I stated previously, where revenue is being generated there is a perverse incentive not to fix the problems.

    If you think less red light running is a good thing because it leads to safer intersections for everyone, then there is no principled argument against increasing the yellow light time a little bit, about an additional 1 s – 2 s as that usually drops the red light running rate by somewhere in the 70% – 90% range.

  • Eric Nordman

    OK I did google it and the yellow light should be at least 4.7 seconds at 50mph according to the ITE formula. Not sure where you got your 5 seconds too short. I suspect it is set at 5 seconds on Bayfront Expressway (50mph speed limit)

  • Your assumption is that the ITE formula is correct. See that 2 in the denominator? v/2a. That means the yellow light lasts half as long as it takes to stop your car. For a derivation and explanation of what the formula does and does not do, go to

  • Anonymous

    Okay, first, you need to stop this thing about there being a violation of physics laws or that the formulas are wrong. The formulas and physics are correct (well, at least approximately … in reality, the deceleration is never exactly constant so really you need to solve the differential forms of the equation of motion, but that’s a second order effect here). The problem you have is you don’t agree with the value for “a”, the deceleration, that is assumed to be representative in determining the yellow light duration. And you also seem to think a perception-reaction time of 1 sec is too short. So I recommend framing your complaint correctly: you are arguing about the value of the numbers in the equation, not the equation itself. Big difference and it gives you more credibility to frame your argument as one of degrees not one of some big conspiracy that we current traffic engineering is “violating the laws of physics”.

    On that note, I really don’t care about increasing yellow light durations by 1 or 2 seconds. There are so many bigger problems with our car-centric society.

    However, as I’ve said before, I’m tired of people co-opting a possibly legitimate concern (yellow light durations being too short) and translating that into “red light cameras are bad” or worse, that they cause accidents. That’s admitting that one doesn’t give a damn about the safety of the intersection and really only cares when they are getting a ticket. This is because, *regardless* of whether or not there is a red light camera at an intersection or not, the intersection is *still* dangerous if (according to your theory) people are running the red light because the yellow light duration is too short. Yet, what is the *only* context yellow light durations come up? Yep: whenever we’re talking about red light cameras (or perhaps people getting busted for running a red light by a cop the old-fashioned way). Just search the internet and see how the overwhelming majority of discussion about yellow light durations is in the context of red light cameras. (Also, I’ve been following this website for years and have never seen this issue come up except whenever there is an article about red light cameras). This proves to me that there isn’t actually legitimate concern about intersection safety but about people trying to get away with behavior that, without automated cameras, they can mostly get away with. If we cared about intersection safety, people would be going to *all* intersections, even if they didn’t have red light cameras, trying to fix what is (according to them) a major safety issue.

    “A passenger car can decelerate up to about 28 ft/s/s with ABS brakes engaged. But it is not comfortable for a driver to do this. … Decelerating at this rate is “safe” but it is not comfortable.”

    I’m talking about a 15 ft/s^2; don’t double my number! And if you are truck, then you need to be driving slower than the speed limit, certainly as your are approaching an intersection that you know has been green for a while.

    Your whole idea of saying we need to allow for the fact that people speed is nuts. If people violate the laws of the road by speeding, then they risk their safety. AND, since they risk the safety of others, that’s why we have laws against this and cops to enforce these laws. You don’t get to say: “Hey, because I want to speed, you need to design your intersections to have longer yellow lights.” The biggest problems with cars is speed, especially to vulnerable road users like pedestrians and cyclists, and we should be doing absolutely nothing to encourage more speeding. In fact, we should be cracking down harder on it.

    Same goes for your claim about rear-end accidents. These are always the result of not leaving enough distance. And especially if you are coming up to an intersection with a yellow light, you need to be slowing down and leaving more space.

    So your argument comes down to this: I want traffic engineers to design roads to account for people speeding, driving distracted, not leaving enough space to the car in front of them, not adjusting to poor road conditions or visibility, and otherwise driving irresponsibly considering the several tons of steel in their control hurtling down the road. This to me, as well as everybody in the livable streets movement, epitomizes the type of anachronistic car-centric thinking that has created unsustainable cities, kills and maims several million every year through accidents, kills many 10s of thousands of more through contribution to air pollution and the obesity epidemic, and has dehumanized our cities. So in light of all that, if you want to worry about yellow light timing, go for it. But if you really care about making our roads safer, you just need to (other than getting people out of cars) get people to slow down (that’s another way of effecting increasing yellow light timing), drive without distractions and take the power literally at their fingertips and toes seriously, and support heavy enforcement of laws that discourage this behavior.

  • J Milan

    These red light runners are real violent crybabies, no?


SFPD and Health Department Announce Pedestrian Safety Campaign

Photo: Thomas Hawk The San Francisco Police Department (SFPD) and the Department of Public Health (DPH) recently announced a pedestrian safety program funded by a grant from the California Office of Traffic Safety (OTS, tag: "Toward Zero Deaths, Every 1 Counts"), a joint initiative that aims to increase enforcement of dangerous and illegal driver behavior, […]

SFPD Park Station’s Most Dangerous Intersections: Not on the Wiggle

New SFPD data indicates that the Park District’s most dangerous intersections have nothing to do with the Wiggle, where Captain Greg Corrales has devoted his station’s limited traffic enforcement staff to ticketing bike commuters who roll stop signs. Under SFPD’s “Focus on the Five” campaign, captains have pledged to target the five most dangerous intersections […]