The goal: No more deaths or serious injuries at Dexter and Thomas.
This seemingly unremarkable intersection has been the site of at least two tragedies in recent memory. Mike Wang was killed by someone making a left turn there while biking home from work in 2011, and Jaahnavi Kandula was killed by a speeding Seattle Police officer while walking in the crosswalk in 2023. Though this project was planned before Kandula was killed, its opening this week feels like the city is saying, “No more. Not again. Not here.”
The new design includes several features Seattle has not used previously, but they are all designed to maximize safety by slowing motor vehicles, shortening crossings, separating modes of travel, and creating redundant safety buffers. It is something of a showcase of safety features, a test of what a high-budget, complete rebuild intersection project could look like. There are curbs and separators all over the place, carefully placed to slow turning traffic, improve sight lines between all users, and make it clear where everyone is supposed to be.
I observed the new protected intersection at Dexter and Thomas for a while Monday, and it seemed like most people had no problems figuring out what to do. The biggest issue I observed was that some drivers wanted to turn left, which is no longer allowed. One person drove in the bike lane to do it, and another sat there confused for a while before finally deciding to turn right.
But even the person who drove in the bike lane to turn left didn’t create a super dangerous situation because they were moving so slow when doing it. Obviously it was not ideal, but that’s the big takeaway I had from riding through it from every direction and observing others use it: There are redundant safety features. For example, let’s say someone in a car wants to turn right. First, they have to take a sharp, slow turn so they are already moving relatively slowly. Then there is a long maybe ten-foot buffer area where their car is turning but they have not yet entered the path of the bike lane. This buffer area accomplishes two important things: 1) The person driving has a lot of extra time to see people in the bike lane and crosswalk and yield to them. 2) The people biking and using the crosswalk have a lot of extra time to see that someone is turning and to look out for them. So even if the driver completely fails in their duty to yield, everyone else has a lot more time to avoid a collision. That’s what I mean by redundant safety features. Several things need to all go wrong in a row before someone is hurt or killed.
Redundancy is a vital feature of any safe system, whether we are talking about a factory or an airplane or a public street. People make mistakes, so we need to design systems in a way that accounts for those mistakes. While the U.S. has embraced a safe systems approach to workplace safety (thanks in large part to labor unions) and air travel (insert Boeing joke here), our public roadways are almost entirely designed so that even a small mistake by a single person can end in tragedy. At most intersections, it only takes a moment of inattention for a person turning right to strike someone in a crosswalk or right hook someone on a bike. Maybe the person was hidden for a moment by the pillar next to the car windshield or maybe the driver was looking to the left for a break in traffic and started moving before checking to the right. There is no backup safety process to prevent a tragedy.
People biking probably have the biggest learning curve here, especially if they want to turn left. Doing so now requires making a two-step turn by crossing straight with the green bike light, then turning and stopping in the waiting area on the far side until the signal changes again. It’s easy to figure out, but it does require waiting potentially for two full signal cycles to get through. That’s the trade-off for all the extra separation. People who don’t want to wait and who feel comfortable biking in traffic can always turn a block before or after using the left turn lane, but now there’s a complete solution for people who don’t feel comfortable merging across traffic or who are willing to wait in exchange for increased safety and comfort.
There is still a lot of work to do on Thomas Street, but it is now mostly ready to play the role of the best biking and walking connection between South Lake Union and Seattle Center. I’m remembering back a decade to before the Mercer Street bikeway under Aurora opened when it was next to impossible to bike between these places. What a difference.
It would be very pricey and take a very long time to turn every intersection in Seattle into what Dexter and Thomas is now. It would be amazing if this style of intersection becomes part of the standard street rebuild process for high-budget projects. I am also hopeful that by learning from this design, SDOT can take the most important elements and apply them to lower-budget projects that can be built much faster across the city. Because while this one intersection had seen far more than its share of tragedy and heartbreak, it’s just one of thousands.
If you want to see the intersection in action, check out today’s video from Best Side Cycling:
Comments
15 responses to “Seattle’s first fully-protected intersection is now open at Dexter/Thomas”
Hmm it seems Americans still don’t fully understand a protected intersection.
In Europe the bike stop line before the zebra crossing would be a yield (shark teeth) and everyone would ride their bikes up to the inner stop line. This is safer as cars can see you ahead of them and you cross immediately ahead of the cars when the light turns green.
Same for pedestrians, you just look both ways, cross the bike path and wait at the inner part.
Also, they neglected making every bike intersection 2-way which would have resulted in only having to wait once for a red light when making a left turn (if it is red going straight, immediately turn left and then right again). Of course you can just use the zebra crossing, but why not make it obvious this is the best way to cross?
Sure, perfect is the enemy of good, but why make suboptimal changes? All of this is established design, which works well all over Europe.
regarding right turns, sure, at a red light it’s best to pull forward and be visible. i think the real problem is during the green light. then, a bicyclist might zip through and a turning car might not have a good chance to see them.
exactly this was reported by someone on nextdoor recently. she was in slow traffic and just barely noticed the cyclist as she started her turn. now, i don’t know if she had her turn signal on or how attentive the cyclist was. i think that’s tom’s whole point about redundancy.
I agree Mark, but there are a lot of variations. If you look at what NACTO recommends, this is actually more of a “dedicated intersection” (https://nacto.org/publication/dont-give-up-at-the-intersection/dedicated-intersections/) then a full fledged protected intersection (https://nacto.org/publication/dont-give-up-at-the-intersection/protected-intersections/). The latter is recommended when you don’t have enough space. It appears as if this is a little bit of both, which seems like a weird mix (and confirms your point). We appear to have enough room for a “bike queue area”, but we don’t have “bike yield signs” (the shark teeth). As a result, it isn’t clear if it is OK for a biker to go to the bike queue area, even though there appears to be plenty of space for that (and it is ideal from a safety standpoint). That being said, I am less worried about drivers turning right on a full stop then I am drivers turning right on green. I think this lead so more awkward stops (a driver pulls out, starts to turn, then realizes they should yield) then it does full fledged collisions. The problem can also be ameliorated by giving the bikers a bit of a head start (which is common now for pedestrians).
As for turning left, that again is an alternative listed as a “High-Capacity Protected Intersection” (https://nacto.org/publication/dont-give-up-at-the-intersection/protected-intersections/variations/). In this case they likely didn’t want to dedicate that much space for bikes. To be fair, the intersection is not huge. In many European cities (where such things are common) the streets are very wide. (European cities often have very wide streets and very narrow streets which is actually a pretty good combination when it comes to managing cars.) This is not especially wide, which may explain why they didn’t make the bike lanes wider (to allow bikes to go both directions to cross the street).
Oh, somebody PUH-LEEZE look at implementing this at N 50th St & Green Lake Way N! That 5-way intersection is dangerous, has lots of traffic (car, bike and pedestrian), and seems to be absolutely begging for this type of treatment.
As for the point about making the bike intersections 2-way, I concur with the point above – it seems we have to drag traffic engineers kicking and screaming away from their preconceived notions one step at a time. Still, this is a step forward.
“The biggest issue I observed was that some drivers wanted to turn left, which is no longer allowed.”
I don’t think it has been allowed for years. This is the Google Maps picture from December 2021: https://maps.app.goo.gl/iNFmR6THiV45wsA87. In terms of what is allowed, it is basically the same. The big difference (what makes this truly unique in Seattle) is that there is that AND the little islands which reduce the danger when cars are turning right (as you mentioned).
“it does require waiting potentially for two full signal cycles [for a cyclist] to get through.”
Realistically though it is just one cycle. If you pull up to the light and it is red, that means the crossing light is green. You can always get off the bike and push it across the crosswalk. Even if it is full of pedestrians I don’t see much of a problem. Just as the confused driver didn’t cause a problem because they were driving slow, a bike rider using the crosswalk while going really slow (pedestrian speed) is just fine.
“It would be very pricey and take a very long time to turn every intersection in Seattle into what Dexter and Thomas is now.”
Yes, but I think we could something similar on a lot of streets. We already have done this in a few areas. Maybe not as nice, but same basic idea. In my opinion the best part of this is the limitations for drivers. No left turns. No going straight across Thomas. To me, that right there gets you most of the way there. This is what they have at 92nd and Aurora (https://maps.app.goo.gl/LQYmSDp4P2ry7CAh7). There are a lot of similarities. Drivers are restricted in exactly the same way (no left turns, no going across 92nd). Drivers turning right onto 92nd have to make an unusual turn. They are no islands, but they are forced into the opposite lane, which by the very nature will cause drivers to slow down (and make it clear that you are now on a residential street).
The biggest difference is that there are no bike lanes on 92nd. To me, this what we need. More intersections like this, and more bike lanes on residential streets and minor arterials. That is the basic idea here: https://bikeportland.org/wp-content/uploads/2023/10/2023-1001_UT_FINAL_email.pdf.
Awesome!
If the construction is over on Dexter then I guess I should switch to Dexter while whatever nonsense is happening on Westlake near the Galer overpass is going on (I can’t find any information about it online).
Dexter continues to be the street where Seattle tries every street and intersection design idea. In this one there’s a reasonable and supported way for us to make all the turns, which makes it better than maybe 95% of the intersections along PBLs that SDOT has designed in the last decade — here it’s cars that can’t make all the turns. I’m very skeptical of our ability to replicate this fully, with enough space that we can wait for turns without stopping dead in someone else’s way, on streets that aren’t as wide as Dexter. We might be able to do smaller versions of this in other places, but… to some extent this is the price demanded by accommodating all these cars. It only gets simpler and cheaper when more of the city needs fewer car trips.
After going through this a few times yesterday I realized I under-estimated what a hodge-podge Dexter is now.
Within a few blocks of here we have an island bus stop, an in-bike-lane bus stop, and a bus stop where the bus crosses over the bike lane. We have this “fully protected” intersection where there’s a traffic signal despite barriers that prohibit cars crossing the median (i.e. eliminating most of the conflicts), then a block south Dexter/John is a free-for-all. We have a signal-phase-mediated right turn at Harrison, a pre-turn cross-over at Denny, this curb-mediated right-turn at Thomas, and those weird merge pockets elsewhere.
I know there’s a history to it… maybe looking at that history could help us avoid similar messes elsewhere…
I’m afraid that is the part of the problem. I think people get excited when they add something that appears to be world-class (hey, look at us — we are like Europe!). To a certain extent, I am guilty of this as well. But 90% of this was done a while ago, when they prevented cars from going straight on Thomas and prevented left turns. At that point the intersection became much safer. Of course (biking) left turns were awkward, but as folks have pointed out, those remain awkward if you are arrive at the intersection and the light is red. You are using the crosswalk (just like you would have a few months ago).
I don’t want to be too critical of SDOT. I think it is good to have models, so that we (as a society) can debate whether something is worth it or not. But I also think priorities are just as important. I would rather build a lot more intersections that looked like Dexter & Thomas did a few months ago than make this (relatively minor) improvement. This intersection has gone from good to great. But how about we make more good ones. It is hard to see why Dexter & John is not like that, given the similarities with this intersection. Why are you allowed to drive across Dexter at John? Holy cow, why do they allow unregulated left turns?!! It seems like we should have fixed that problem long before we touched this intersection.
When I look at Dexter I think there are a few distinct stretches with distinct histories:
– From the bridge down to Comstock, the design from over a decade ago with bus islands and buffered lanes inside the parking spaces has held up fine. Parking is low-turnover, so there aren’t too many conflicts with that, and trying to do the “parking-protected” style would make it harder for drivers making unprotected lefts to see oncoming cyclists, and vice-versa.
– From Comstock down to Aloha that design was later changed, moving the bike lanes outside of the parking. This made sense in light of higher parking turnover accompanying development in the southern part of the corridor. I have no problem with this change, on either side of Comstock.
– Between Mercer and Denny SDOT tried to apply an experimental design from NACTO. That’s the part that really went off the rails. The bus stops, with huge distance from the travel lane to the curb, have been really bad for everyone. The right-turn pockets, which had some logic on paper, just haven’t worked out — whatever the logic behind them, whatever the reason, drivers just don’t use them correctly, resulting in some of the worst right-turn conflicts in the city despite fairly low volume of turning traffic. Because this design has worked out so badly we’ve been re-doing it intersection-by-intersection whenever we have the chance, with continually more expensive and heavy-weight solutions.
To me, it was 2014 when SDOT really screwed up, applying the NACTO design without really thinking about how all the traffic flows and conflict points would work. And the NACTO design was just bad on its own terms… it was supposed to be a standard set of design tools but a bunch of them were experimental and untested. Maybe those right-turn pockets should have been tested at one intersection (preferably in a city populated exclusively by people I don’t like) before making it into a design manual to be applied on entire corridors.
Please REMOVE this terrible design. Never repeat it. This is a lawsuit to the city waiting to happen and I will ride in the road forever more. Ridden this 100s of times over 15 years – totally wiped out w the new (stupid design) could have been very injured. How is it safer to install a diagonal unmarked curb?
Some of the curbs on the soon-to-open PBL on Pine, on the way down from Capitol Hill, are even worse in this regard. I didn’t have a problem with this one when I went through it a few times yesterday but I could see it being tricky for unfamiliar riders when visibility conditions are bad (common enough here).
Lane-edge hazards have been a common problem in recent Seattle lanes. As a fairly experienced rider that’s used to keeping my eyes up for safety, it’s been weird to see a lot of new designs that force you to look down at the pavement.
That’s a good point. If you are just cruising along Dexter, then the little jog seems unnecessary. I think this is pretty much universal with all protected intersections. I think there are two issues here:
1) This is something new.
2) Is this really necessary *here*?
This is not an ordinary intersection. It is one in which the crossroad (Thomas) does not go through (for drivers). Drivers can not take a left at all. Thomas is minor compared to Dexter, which means that my guess is, most of the time the light is green for those continuing on Dexter. Those riders have to do this extra jog and it seems unnecessary given the threat by cars.
In contrast, Dexter crosses both Mercer and Denny just a few blocks from here. Both of those streets are major. Riders naturally realize that, and understand that being able to simply cruise through the intersection (on Dexter) is unrealistic. Thus riders making a jog (in the name of safety) is quite reasonable. The intersections are also very wide — perfect for this sort of thing.
I bike commute through this intersection every weekday and only wish the city would have redirected the millions spent on this one intersection to repaving the pot-holed riddled streets that pose a much greater hazard to bikers.
As for the improvements, I watched this morning as a truck attempted to turn right (westbound) on Thomas from Dexter, but the new Thomas Street is so narrow the truck couldn’t execute the turn with an eastbound vehicle on Thomas waiting at the light. So…vehicle traffic on S bound Dexter and Thomas, the bike lane and the crosswalk were blocked while the car reversed and the truck navigated the turn. Way to go SDOT!
It’s really quite a preposterous use of our resources.
Looking at the street design it appears that would have happened a few months ago as well. I don’t think they made it worse. If anything it is better, as the stop lines on Thomas are now a lot farther from the intersection (as shown by the picture on this website versus the old Google Street View https://maps.app.goo.gl/4ar2r7qBGdLR7pN27). My guess is the driver simply pulled up to far, in an attempt to make a right turn. Either that or the truck driver miscalculated.