The 11th MUTCD is Exactly What We Should Expect...
...except for the damn 85th percentile speed trap.
tl;dr: We can’t be wrong and have no accountability for it. The new MUTCD helps to bring up-to-date the mismatch between policy and need, but I’m not yet seeing a process for accountability in writing or in spirit.
I awoke this chilly December 19, 2023, to find a
Notice that the MUTCD
had been updated and released to the public and now we know how to mark our roads
and I took the one less traveled
more slowly than 85% of my colleagues
and it made no difference at all
—Robert “Moses” Frost
Today, you get poetry and memes, dear subscribers, because the content here can get very boring sometimes. Policymaking is a s t r u g g l e between historical context and current need; by definition in a republic of laws, rules, and measures, policymaking is responsive to need, even if that need is proactive. This was certainly the case for autonomous vehicles in 2009 when USDOT via FHWA released its 10th edition of the “Manual on Uniform Traffic Control Devices for Streets and Highways,” otherwise known as the MUTCD. It is no longer the case with the release of 2023’s 11th take on keeping our roads marked for safety and navigation; we’re fully reacting now.
We’ve traveled 14 years since the 10th edition, and our travel landscape has changed, shifted, and dug in. To claim that we’re not better off is reductive as hell: so many advancements in technology have narrowed the rails that keep us safe to a loose singularity. We, as both a field and a representative body of our users—aka all travelers—know and can name the outcomes that will, broadly, ensure the future of access and mobility, no matter the technology, whether a motor vehicle is gas or electric, user-controlled or user-enjoyed. The how we get there from then and from now is very much up for debate and the MUTCD, whether it’s dutiful “we’re just a reference manual” punts on its role or not.
Back to AVs: companies birth, grow, and die in service of optimizing the technology that will, with optimism and stacked successes, replace the need to rely 100% on human action and reaction to get around. According to one AI-in-tech scholar, Renée Awesome1 Ray, “For an autonomous car to be safer than a human driver, it needs to avoid crashes at least 99.999820 percent of the time.” Whereas the MUTCD had once reflected broad rulemaking and public comment for how public jurisdictions should either mark their own rights of way or how they should direct partners to do the same in the near future, the newest version has a big benefit of hindsight and 14 years of iterative successes and failures. We have precise numbers; anyone has a >99.999% chance of surviving any given trip. We know more, but in a manner of speaking, the rules still don’t write themselves. We’re still losing thousands of lives on our roads every year; the pedestrian fatality crisis is proof.
Autonomy and a new edition of the MUTCD don’t automatically fix our problems, but they sure do give us shiny rules to hide behind.
But first: speed limits. I harp about the precise language we’re conditioned to use when talking about speed limits, because, in reality, there are five speed limits.
posted speed limit: the sign you see.
observed speed limit: how fast you were going.
designed speed limit: how fast the road was designed to go.
These three are often bundled together as a parametric way to talk about speed; it’s easy to see how they could conflict with one another.
the speed at which you get caught.
the speed at which you die.
These last two aren’t canon in planning or engineering, but I bring them up because they connect the technical (how) to the social (who). Keep all of these in mind when reading next. What speed limit are we really talking about?
No rule represents the uneasy wade into the updated document itself than the controversial and utterly ridiculous “85th percentile” standard for setting and maintaining posted speed limits. In short, irrespective of the design of the roadway, the acting jurisdiction—usually the state—should set its posted and legislative maximum “speed limit” at or around the 85th percentile of some average, finite observation of drivers during certain hours on certain days in a specific location. If SamDOT observes the average speed under these conditions to be 100 mph, the posted speed limit should be 85% of 100 mph—85 mph. Done and dusted. We’ve got to trust the statistics, right?
Wrong.
In long, the 85th percentile rule isn’t new and is so unbelievably outdated that its outcomes are definitions of the statistics maxim “garbage in, garbage out,” for at least two reasons, but I’ll direct you to the over 17,000 comments on the proposed rulemaking2 from this very document for many more.
Both of the reasons below directly oppose DOT’s stated “safe streets approach,” which the BIL/Justice40 initiatives SHOULD seek to achieve. Millions of dollars and millions of eyeballs are watching (not so) eagerly to either laud the investments or decry the whole infrastructure/environmental justice approach of the 117th Congress as a government overreach/overspend, at best.
The rule rewards drivers who break the law.
From our friends at Strong Towns, the logic is:
The 85th percentile speed is the speed at or below which 85 percent of the drivers will operate with open roads and favorable conditions. The assumption underlying the 85th percentile speed is that most drivers will operate their vehicle at speeds they perceive to be safe. Speed limits set above or below the 85th percentile speed will create unsafe conditions due to speed differential as some driver adhere strictly to the law while others drive the naturally-induced speed.
The response to a new set of observations—one that calculates that a majority of the drivers under our measurement scenario now routinely travel above the 85th percentile—is not to recognize that the road is unsafe for all users (drivers included!) and redesign it and/or lower the speed limit. The MUTCD and its 85th percentile speed would then call for raising the speed limit.3 We know that people die exponentially more often if/when struck by a vehicle moving fast. We just have to care.
2. The rule has not taken into consideration non-car trips in the past.
I will be accurate and not reactionary here: the reclassified rule, as codified by the 11th MUTCD,
…allows for roadway owners and engineers to consider a wide variety of other factors in the engineering study including road characteristics, roadside development and environment, pedestrian activity, parking, and crash experience. All these factors (including speed distribution) are analyzed as part of the required engineering study and it is through that comprehensive analysis that the appropriate speed limit is determined… (Emphasis mine.)
instead suggests that an “engineering study”4 might consider “pedestrian activity” in its determination of a posted speed limit. The rule, as agnostically as a rulebook would allow, doesn’t define what qualifies an engineering study or what “pedestrian activity” means.5
The new Support provision clarifies that a range of factors can influence the speed limit determined in the engineering study. These factors include land-use context, pedestrian and bicyclist activity, crash history, intersection spacing, driveway density, roadway geometry, roadside conditions, roadway functional classification, traffic volume, and observed speeds. The engineering study will determine which of the recommended factors will prevail in setting the appropriate speed limit and the new provisions are intended to ensure that practitioners consider all road users when setting a speed limit. (Emphasis mine.)
This paragraph is prima facie GOOD. See, not all bad! Except, the “guidance” and “Support” language is careful to not commit to a “must,” because safety is too pedantic for a rule document. That sentence is, truly, unfair. The point of the MUTCD is not to direct or preempt local authority, or even other rulemaking, but to guide road designers to consider certain factors when embarking on a project. In practice, professional engineers often hide behind the MUTCD as legal cover for both following local jurisprudence or bucking it all together.
Kea Wilson, an editor for Streetsblog expanded upon and summarized this idea in an article regarding this very rulemaking process, from 2021:
And when cities want to use those life-saving design elements anyway, they’re often scared off of doing it, lest they fall out of compliance with the all-powerful Manual—even though, technically, not all of its recommendations are legally binding, much like its companion document, the AASHTO Green Book. In part because remaining in compliance with the MUTCD may shield transportation agencies against lawsuits, many traffic engineers tend to treat it more like a Bible with strict commandments than a “recipe book” that encourages chefs to sub out the nuts if they’ll send the person who’ll actually eat the dish into anaphylactic shock.
In reality, however, commentary from Transportation for America says the quiet part loud:
US law explicitly calls for the development of street design standards and recognizes several standards guidebooks. FHWA then outlines which standards must be consulted for use on National Highway System projects in federal regulations.
This does not mean a legally binding restriction to a standard. But since these standards are directly referenced in law, lawyers and judges often erroneously believe that these standards are law, and defer to professional engineers’ judgment on how design standards are best applied. And as noted in our other explainer on design standards, they allow tons of flexibility.
From experience, I can imagine a scenario where none of the mentioned factors (land use, crash history, etc.) are appropriately and categorically applied for any number of legal, practical, administrative, or time reasons. Even so, good engineers and highway planners will find a way to prioritize safety; great ones will demand it. That begs the following question:
Where’s the accountability? The MUTCD is not a binding legal contract, nor is it an authority for mediation. NOR is it a poster for tort reform. There’s got to be something else to hold engineering judgment to account.
Lastly, before I get off my horse, there’s a two-way knowledge/temporal mismatch in the expectations set forth here. What do I mean? Many of our seasoned traffic engineers have never been asked to think about environmental justice as part of their statement of work; many of our younger engineers have and do. It will take time to replace the outgoing knowledge base with an incoming one that views access and mobility as not only, “Can and did this traveler make it to their destination on time?” but also, “Can and did this traveler make it at all?” We, as a society, don’t have the time to wait, because both death and autonomy are coming and we better get it right.
If we do get it right, AVs should know to adjust their speeds based on any host of factors and we’ll get a sixth speed limit: crowdsourced. The speed limit is reset every half hour based on real-time data. What will the next MUTCD have to say about it?
It’s Autumn, but it’s also just right there.
Not all of these comments are directly related to the 85th percentile rulemaking, but many are, including some from my former employer, Smart Growth America/Transportation for America.
Of course, judgment first.
This phrase does have a specific definition as does its corollary, engineering judgment, which I hate to relegate to a footnote because the combination of it and engineering study is the crux of the nuance here. Though this sentence on page 17 of the MUTCD makes me sadchuckle: “Documentation of engineering judgment is not required.” I guess the outcomes of such a judgment are expressed in said study.
The phrase is mentioned once, on page 22 of the 1100+ page document as part of a definition for “pedestrian facility…”
Pedestrian Facility—a general term denoting a location where improvements and provisions have been made to accommodate or encourage pedestrian activity.
I take this pedestrian activity to also include death, at least 8,000 times in 2022.