My city recently redesigned a nearby intersection for higher safety and efficiency. It looked good on paper, but in practice the bike lanes turned out to present some difficulties. I ran through it a few times to get a better understanding of it; this post presents video of the runs, some analysis and observations, and finally a suggestion for how to improve it.

(Summary of conclusions: This is still a car-optimized intersection. The bike lanes are unpleasant, difficult to use, and have 2-3x the conflict points of the main lane; many cyclists would benefit from using the main traffic lane. Recommendation: Paint "super sharrows" in the loop.)

Update 2021-12-23: I've adjusted some conclusions around safety to clarify which groups I think are currently the least well served, and to better compare the old design against the new one (as opposed to just several elements of the new design against each other.)

Background on the redesign

The city of Somerville, Massachusetts recently redesigned a intersection near where I live, the Powder House roundabout. They did it as a "quick build" project, only using paint and poles, with no changes to paving or curbs. The stated goals were to improve safety (after several pedestrians were killed by drivers) and to reduce traffic backups. The roundabout also has a reputation of being pretty terrifying to navigate, with six roads joining together at varying angles in a broad, unmarked expanse of pavement and cars coming from every direction, and I imagine that one of the unstated criteria was to just make it less confusing.

The proposal looked pretty nice, with a ring bicycle lane protected by posts, a narrower main traffic lane, better traffic funneling, shorter crosswalks, and no stoplights. (I was a little disappointed that there would no longer be crosswalks to a tiny park at the center of the circle, but as quirky as it was, I really only visited it once. It's not really a pleasant spot to hang out.) They also moved the bus stops out of the circle. In theory it looked pretty nice!

Here's the design that was proposed:

So, how did it actually turn out? I took a few rides through to see what it was actually like to use from a cyclist's perspective.

(I should note that the build isn't quite finished; the city still has a lot of posts to put up. But the paint is down, and drivers seem to be respecting it.)

Update 2021-10-26: I recorded a new pair of videos where I don't fiddle with the camera during the ride. Should be much clearer. I also rewrote my analysis, although my conclusions haven't changed much.

The bike lane

I took a ride through the new roundabout design using the bike lane ring, making a loop starting at Broadway on the east side:

The first thing to observe is that in the bike lane, for each street I crossed I had three conflict points involving merging or crossing traffic and a complicated path to follow. I had to 1) make a sharp left turn from the "bypass lane" to the "crossing lane", 2) immediately look to my left and negotiate with circle-exiting traffic so that I could start crossing, 3) switch to looking right to negotiate with circle-entering traffic, 4) keep looking right to watch for cyclists entering the circle, and finally 5) make a sharp left turn back onto the bypass lane.

In the bike ring there are 18 conflict points total: 14 cross, 4 merge. (I don't care about diverging points.) None of the merge points have yield indicators to indicate whether bikes in the ring or the bypass lane take priority. About half the turns are sharp, due to the limited space and the need for the bike lane to cross the main lane at roughly right angles.

At four or five points I had to stop and make eye contact with drivers in order to make sure they were yielding or slow down to find out whether they were even exiting.

There was also a truck blocking the bike lane at one point, and in general this is a pretty common sort of occurrence and is one of many reasons you see cyclists avoid bike lanes. However, it's possible that the flex posts will prevent this once fully installed. Similarly, the bus at the end started pulling left as I left the bike lane to pass it, but this could happen for any bus stop of this design. It's not specific to this roundabout. But the car stopped on the bike path near the beginning of the video? That's analogous to the way confused or rude drivers will block a crosswalk, and that's going to be happening a lot here.

The main lane

And then, for comparison, I took the same route but in the main traffic lane, in line with the cars:

In contrast to the bike lane ride, I only had one conflict point per street (circle-entering traffic, where the drivers already expect to yield) and moved in a very simple circular path until my exit. There was no need to stop or slow at any time. In order to "cross" each street I had to: 1) watch that circle-entering traffic were yielding to me, and... that's it. There's just one step.

For the main circle, there are only 6 conflict points total; all are merges with a clear yield indicator. (There are two additional, illicit crosses where someone entering the ring might enter and pause before they have a proper gap to enter.)

Personal comparison

In the bike lane I was overwhelmed by the number of things I had to keep track of and the sharp turns and changes in direction of focus. I had to keep looking left, right, right, left again, over and over, and felt like I was always about to overlook a car coming at me—or that I was insufficiently visible to the cars. Repeatedly and rapidly changing between "riding" mode and "crosswalk" mode was disorienting.

In the main lane, the only time I had to look anywhere but (roughly) forward was when I was entering the circle. Not only was the experience smoother and less confusing, it was also faster: Measuring the time between entering and exiting (as defined by crossing the crosswalks), I spent 64 seconds in the bike lane run and only 27 seconds in the main lane run. Even if you include the time spent waiting to enter the roundabout in the main lane, the bike lane run still took twice as long.

Just speaking personally? I felt more comfortable riding in the main lane.

But what about the general population?

OK, but I would be classified as a "strong and fearless" cyclist according to Geller's categories, a group of less than 1% of the total population and perhaps 10% of those biking regularly. The largest population of people biking regularly are the "enthused and confident", with the "interested but concerned" being a majority of the population but with only a small portion biking regularly. For the purposes of this post, though, I might actually break it out differently: How experienced/competent people are, and separately how confident they are. How would this roundabout work for them?

I'll grant that it's possible that an inexperienced cyclist might feel safer in the bike lanes in the new design, and perhaps would benefit from it as well. Imagining the current design without a bike lane, people who are still nervous about biking around moving cars would likely stick to the outside of the main traffic lane rather than taking the center of the lane, increasing the possibility of "right hooks" from exiting drivers. (I wonder if these right hooks would have been less likely in the old design, where biking might have felt scarier but actually been safer. I'm not entirely sure, though.) Apart from safety, these cyclists might feel safer and less rushed in the bike lane loop. If this increases cycling uptake, then I suppose I have to respect that. However, I have serious concerns that the sheer complexity of navigating this design will cause inexperienced cyclists to either avoid it or get confused and fail to negotiate the intersection safely, e.g. by failing to check in the correct direction on each crossing.

I'd also claim that the bike lanes could present some danger for fast-moving cyclists, which are a sizeable portion of the cyclist population, if they opt for the bike lane. At higher speed, the higher rate of conflict points creates more danger and confusion. Faster cyclists are more likely to be more skilled, but they'll need to rely more heavily on that skill in order to handle the labyrinthine path at speed. Segregated bike paths also create social conflict; even though bike lanes are always optional, a small but loud fraction of drivers get angry at cyclists not using an available bike lane. So while I could use the main lane—and will—I also expect to get some bullshit from the more entitled-feeling (and less clueful) drivers.

Also important to consider are those with heavy loads (e.g. hauling kids or larger loads of groceries in a trailer) or less stable bikes (such as those encumbered with a trail-a-bike or top-heavy load). From experience, these would have more trouble navigating the stop-and-go, tight-cornered bike loop. Bikes are unstable when moving slowly; creating stop-and-go conditions increases the chance of minor injuries.

From all this, I think it's clear that a certain portion of the cyclist population should be using the main lane—namely, those capable of high speed, carrying heavy loads, riding a less stable bicycle, or confident enough to take the lane. But there's no particular support for this population.

Overall, I think the new design is still heavily prioritized in favor of car efficiency and experience with bicycle and pedestrian safety in mind as a close second. Bicycle efficiency and experience are still second-class, and bicycle safety in particular may suffer in actual practice for certain subpopulations, and create either more danger or increased social conflict for fast-moving, experienced cyclists depending on whether they choose the bike or main lane.

I'll grant that overall this is probably an improvement over the previous design, which was probably worse for all but the fastest and slowest cyclists. However, the balance is still in favor of car traffic.

A suggested improvement

I think there's a fairly simple solution, though. The main risk of cyclists using the main lane is that they'll feel uncomfortable taking the center of the lane, either for safety concerns or due to imagined or real social pressure. Because the lane is so wide (due to its curvature), there's enough room that cyclists might stick to the right of the main lane, but that of course increases the chance of right-hooks. However, there's an existing solution to this: Super-sharrows. These are markings down the center of a main traffic lane indicating that bikes should ride in the center of the lane, rather than at the outer edge. These communicate to drivers and cyclists alike that bicycles belong in the center. From my experience commuting through Allston (also in Massachusetts) this seems to work quite well.

I believe that with this modification, cyclists would feel empowered to use whichever lane was most appropriate for their needs.

(As a side benefit, this may even increase efficiency for cars, since it would reduce one exiting-traffic conflict point for drivers. By allowing in-circle traffic to exit with greater ease, the flow within the circle would be more continuous and less prone to jams.)

I still haven't tried out the circle as a pedestrian, but I suspect it is largely an improvement due to the shorter crosswalks, larger medians, and better-defined traffic flow. However, as I've learned from this experience, it's tricky to predict the on-the-ground experience from a diagram. I'd love to hear from others about their experiences with the new design—whether by foot, bike, or car.