May 16 is National Bike To Work Day
May 15, 2014 7:44 PM   Subscribe

Strava, the bike and run tracking system, is using their database to create Strava Metro, to sell to urban planners for commute data. But unless you're the Oregon DoT, London, or Alpine Shire, you might find the Strava bike and run heatmaps more useful.

Outside: Strava's Plan To Revolutionize Commuting
“If cyclists were cars, even in just the numbers that cyclists have now, and they were using the infrastructure that is generally provided to bikes, people would be up in arms about how bad the infrastructure is,” Horvath says. “So we think Metro is important because it shows just how much commuting is being done by bike and pedestrians. Even before the question of infrastructure, it’s about advocacy and awareness.”
Just in time, since May is National Bike Month

The Strava Files - ", the most successful, popular, and influential social fitness network to arise over the past half-dozen years, had launched soon after Flint started riding in 2009. Within a year, the cyclist would infamously come to epitomize its allure, value, and menace."
Why Cyclist Is A Dirty Word

Strava, previously
posted by the man of twists and turns (34 comments total) 15 users marked this as a favorite
Ooh, neat! Much more informative than trip counters on bike trails, as this identifies all infrastructure used by bikes (and pedestrians of all shades).
posted by filthy light thief at 7:54 PM on May 15, 2014

I love that the heatmap glows kind of bright right at the velodrome in my general neighbourhood.
posted by sparklemotion at 8:41 PM on May 15, 2014 [1 favorite]

I bike every day so my kids don't have to be "cyclists", just people who happened to ride their bike to work.

I also bike every day since the freaking antarctic ice shelf is collapsing.
posted by bottlebrushtree at 8:59 PM on May 15, 2014 [1 favorite]

I spent quality time finding little glowy spots not on roads near my house and IDing them. "OK, that's Maplewood Bicycle, that's Big Shark Bicycle, that's Schlafly Bottleworks."

The one I don't get is the bright blue cloud in someone's front yard in the middle of a residential street.
posted by Kid Charlemagne at 9:01 PM on May 15, 2014

I worry that, if this is used to guide government policy, it will be biased toward the needs of athletic 20-40 higher-income riders at the expense of those who don't fit that profile. And policy is probably already biased in that direction.

Still, pretty neat.
posted by alexei at 9:20 PM on May 15, 2014 [8 favorites]

I dunno, in the St. Louis area we lack a lot of trails infrastructure, but across the river in Madison County they have a really nice trails network. What was immediately obvious to me was that we have a lot more cyclists on busy streets on this side of the river, while on that side of the river (with much lower population density) their bike trails were brighter than any of our streets, but their streets were much dimmer. That's pretty in your face data.
posted by Kid Charlemagne at 9:27 PM on May 15, 2014

I just like riding bikes. If Strava makes it even easier, gives me even more company on the roads each day, more power to 'em. More bikes for me to ogle.
posted by sandettie light vessel automatic at 9:29 PM on May 15, 2014 [1 favorite]

It's been controversial here in Oregon actually. Strava is used by enthusiast cyclists (including me) and rarely used by commuters. In fact, usually I just see bike racers I race on weekends with doing their daily commutes on Strava. It's not a popular tool for casual cyclists that criss cross the city.

The data from enthusiasts is really skewed, it emphasizes hilly training rides along the west side of Portland a bit more than the regular commute paths in the center of town. I could see hotspots for all the big bike shops that deal with road race bikes and I even saw highlights on the Portland heatmap for areas where races take place and everyone racing is using their bike computer.

So while I love the app and I use it a bunch, it's used by people wearing spandex that are way into the sport as a hobby and I would argue it skews upper class to people that can afford bike computers, fancy race bikes, and have the time to do all that riding and training and isn't the representative sample of regular cyclists I'd use to generate urban design policy.
posted by mathowie at 11:18 PM on May 15, 2014 [13 favorites]

I'll agree it looks like the data could be skewed. I live in a sort of bend in the river, and the path taking the long way along the river path is much brighter than the straight cut along a road that I would ordinarily take.

There's also a medium-bright trail of people riding directly on the river. Presumably Strava Metro will allow seasonal filtering because that works better when it's frozen.
posted by RobotHero at 12:13 AM on May 16, 2014

The surprising discoveries include a parking lot behind a Cabela's, the park behind the library and a couple random people's drive-ways.
posted by RobotHero at 12:41 AM on May 16, 2014

Mod note: One comment deleted. Cycling is a big topic, but this particular post is more about data and infrastructure rather than outrageously lenient sentencing for people who have committed vehicular homicide or similar, so maybe it's better to discuss that in a different post.
posted by taz (staff) at 1:00 AM on May 16, 2014

Yeah, I'm a transportation engineer who would use this exact sort of data, and this seems really limited to me.

To begin with, the very nature of the data is problematic -- I would expect the data set to be heavily filled with a small number of enthusiasts, who I would expect to ride similar routes on a very regular basis for training purposes. Not only that, but the focus is on cycling as an exercise, which is great but isn't really all that related to commuting; I think the ideal route for someone training might be a big loop through scenic terrain that matched their conditioning goals, but didn't go anywhere; the ideal route for a commuter might be a pretty direct route that goes right to their work, probably through a bunch of intersections that would really screw up your pace if you were riding for fitness, and on as flat a path as possible so you don't get to work all sweaty. Looking at San Francisco for example, I know that the key cycling commute routes are not in fact through the Presidio, Golden Gate Park and the Embarcadero, although those all sound like really great rides on a sunny afternoon. Heading northeast from the Bay, you see the American River pathway in Sacramento and Folsom lit up much more than you see the roads in Davis, which is probably the most bike-intensive city in America.

Of course, it's not only tracking the wrong kind of travel for what they are selling it as (in terms of cycling as a commute mode rather than as a good exercise), it's also working with a very limited population; fitness-focused enthusiasts who are also very wired. Technologically based data-of-convenience like this tends to emphasize certain demographics; high income, employed, white collar, tech-savvy, male, white, English speaking. This is the demographic that -- after billionaires -- is the most represented in terms of public opinion. I'm troubled by the idea of the digital divide pushing its way further into policy; I know that I'm always surprised whenever I check -- the only person I personally know without a smartphone is my 70 year old mother, and she's talking about getting one, but in fact 40-45% of adults don't have smartphones, and I have a skewed perception because I travel in a tech-savvy circle. (So does almost everybody reading this comment, by definition.)

If it was super-cheap, then I suppose there are potentially trouble spots you could see or something, but this data is the epitome of looking under the streetlight for your car keys; not because that's where you dropped them, but because that's where it's easiest to look.

That being said, there are other apps much more relevant at getting cycling data into the hands of the transportation planning community; a group over in the city of San Francisco have created something called CycleTracks - it's an Android/iPhone app specifically designed to gather the additional information that is useful for planning (like the purpose of travel, and a little bit of demographic information that really adds a lot of context) and is aimed at a more broad section of users, including more casual riders, with a focus on cycling-as-transportation mode more than as health activity.

They've done a couple of cool things with it; not only done some published research [PDF], they've also made it open source. So a number of agencies are working with it, including San Francisco, Seattle, Montréal and Atlanta. If you cycle, running the app and giving the planning community your data would be appreciated.
posted by Homeboy Trouble at 2:06 AM on May 16, 2014 [11 favorites]

The skewed data does seem to be a concern, as with all smartphone-based reporting systems. Boston recently trialled an app called "street bump" to watch a phone's accelerometers and GPS, and automatically detect and report potholes as you drive over them. A brilliant idea on the face of it, but obviously:
The program inadvertently directed repair crews to wealthier neighbourhoods, where people were more likely to carry smartphones and download the app.(source)
Crowd-sourced data like this has fantastic potential, but it's hard to see how you'd correct for the huge sampling biases.
posted by metaBugs at 4:15 AM on May 16, 2014

Of course, the series of interlinking well-paved trails that cover the Eastern Townships are bright blue! As well as the mountain bike trail on the mountain behind my house too.
posted by Kitteh at 5:11 AM on May 16, 2014

I want to add that when my job moved too far away to bicycle to work, I took my bicycle in my car, drove to a safe, good, free parking spot halfway to work and bicycled from there. You don't have to bicycle the full distance.
posted by dances_with_sneetches at 5:23 AM on May 16, 2014

Yeah, the training rides on the few hilly roads we have around here are much brighter than the bike paths everyone's commuting on, and we've got very high rates of bike commuting in this town.

Still though. This is an extremely fun map.
posted by gerstle at 5:37 AM on May 16, 2014

I think peoples' concerns with the thinness and bias of this data are completely valid, but I'm a bike commuter who uses Strava daily and have gotten a few other commuters into using it as well. If a city like Boston said that it would start using Strava to measure cycling infrastructure usage, that would be a huge incentive for riders to start using the (free) app. They could do public education about using it.

Even better would be to get GPS tracking on the CitiBikes so that they have that data as well.
posted by Aizkolari at 5:41 AM on May 16, 2014 [2 favorites]

Hey sparklemotion, I'll be lighting that up further starting on Thursday!
posted by entropone at 5:52 AM on May 16, 2014 [1 favorite]

“So we think Metro is important because it shows just how much commuting is being done by bike and pedestrians. Even before the question of infrastructure, it’s about advocacy and awareness.”

I am reminded of the recent article on New York City's attempt to borrow from Sweden's "Vision Zero" traffic safety approach -- changing infrastructure and speed limits in response to fatalities, with the goal of zero traffic deaths. If we even slightly did that nationally, the intersections and sections of road that are dangerous to bicycles and pedestrians would light up like neon lights on the map and you'd see radically different rules and physical design in place.

Even flawed as this data is (I'd never suggest using it for actual transit planning, say), at a basic level it shows that there is a lot of non-car use, which often literally doesn't count. If that can help drive more inclusive design and regulation, then that's a massive benefit.
posted by Dip Flash at 5:58 AM on May 16, 2014

mathowie: "skews upper class to people that can afford bike computers, fancy race bikes, and have the time to do all that riding and training and isn't the representative sample of regular cyclists I'd use to generate urban design policy."

That should be easy to correct for. Strava has time, date and speed data-- they could just ask their database "show me a result-set for average speed commuters during the boundary of average work hours."
posted by Static Vagabond at 6:09 AM on May 16, 2014

The data also crucially lack evidence of the locations most needing improvement - those are, by definition, completely dark in Strava's map. All those intersections of death, the roads that are just too narrow and fast to ride, the ones that would be a great route if only they wouldn't involve a side trip to the hospital, are not highlighted by this method.
posted by Dashy at 6:16 AM on May 16, 2014 [3 favorites]

Static Vagabond That should be easy to correct for. Strava has time, date and speed data-- they could just ask their database "show me a result-set for average speed commuters during the boundary of average work hours."

Hopefully the non-free dataset can be filtered by date. The "Run" heat map in my city is heavily biassed towards (last year's) 30k run route. Other cities with marathons probably show the same.
posted by Popular Ethics at 6:23 AM on May 16, 2014

I've been using Map My Run for a few years, not for any particular reason other than it popped up on an App Store search and was free. It does a pretty good job, but I really like this innovation and the idea that I'd be contributing to this collective map. Strava users, are there any things that suck about it?
posted by Shepherd at 7:01 AM on May 16, 2014

entropone: Racing/spectating meetup?
posted by sparklemotion at 7:06 AM on May 16, 2014

I'll be in the infield wearing neon green and zebra stripes, riding this. :)
posted by entropone at 7:08 AM on May 16, 2014 [1 favorite]

I'm one of the stereotypical Strava users mentioned here. I do training rides and pretty much use it as a motivation tool. While I do commute by bike my commute happens to be so short that I'm not inclined to track it on Strava, but I'll consider changing this if it helps give a bit of awareness to the rather busy intersection I cross in my commute. I suspect many planners wouldn't fathom that people ride bikes much there.

For all the demographic shortcomings I can't help but think that the data will still create a "holy shit" reaction among policy makers to see just how many roads are regularly ridden by bike. Most people not involved in the sport think of bicycling as something done on a path (or sidewalk, sigh) and it should be rather eye popping for some and may get them a bit more serious about the infrastructure investments needed.

In another discussion forum a person who works on traffic surveys had some measured things to say about this. He noted the demographic concerns, but also pointed out that for the money you get a tremendous volume and breadth of data. No one is claiming that quantity trumps quality, but you can spend $20k on a traffic study employing people with clip boards and produce a data set that is a thousand times smaller in geographic coverage than what Strava offers.

I'm no expert on traffic planning, but I think a plausible approach is to use Strava to see the big picture and where there are doubts follow up with specific traffic studies. It doesn't have to ban either/or choice because the insights of both seem valuable.
posted by dgran at 8:01 AM on May 16, 2014 [1 favorite]

Fairly depressing to see that the most heavily-traveled (actually, heavily-recorded) Stava routes in my area are hiking paths where cycling is explicitly forbidden. This does not help us make the case that cyclists are responsible, rule-following users whose access should be expanded to more parks (neither does the damage that's already been done to off-trail areas in both allowed and excluded parks)

The sampling bias is also obvious and makes the data less than useful for casual and commuter cyclists. A high-speed, five-lane, shoulderless arterial only used only used by experienced and brave (and fast) cyclists shows more traffic than the parallel boulevard that has easily a dozen casual riders per mile on any summer evening.

To make the data more broadly valuable valuable something would have to be done to get an app into the hands of folks riding to work, shops, and school. Gamification, maybe? I'd love to see an tool that allowed schoolkids to compete based on the number of miles walked or ridden each week, and recorded and shared the most-used routes.
posted by CHoldredge at 8:31 AM on May 16, 2014

It's really interesting to look at the run map. I had the impression that hardly anyone was running the trails I have been using in one of the local parks: I could run for over an hour (even on a weekend) and not see anyone. But the Strava users clearly know about my routes: the blue is as bright as the really popular trails in Redwood Park. Huh.
posted by suelac at 8:31 AM on May 16, 2014

I just want to reiterate what Homeboy Trouble said up thread. The data is extremely limited. I do trust the planners and engineers at ODOT to use the Strava data as an interesting supplement to existing data sets, but they can't be so foolish to use it alone to plan bike infrastructure.

I've been involved in numerous conversations since this whole thing errupted on Twitter. There's an OCCUPYSTRAVA group to use it for commuting and not training. They also have a hashtag - #occupystrava.

I work with a lot of travel behavior researchers, and the inherent bias of using smartphone-based data collection is always concern. They tend to forgo that with sampling when their research is more focused on the methodology of how can you use apps to collect travel surveys. Anything that will directly influence policy and infrastructure, that definitely requires more attention to represented demographics.

Though I also think this fits into the larger issue of bicycle facility improvements as a reflection of social class. Just look at how many papers are written about bike lanes and gentrification.
posted by kendrak at 9:08 AM on May 16, 2014 [2 favorites]

Some of the issues people are pointing out with the data relate to why I don't use Strava -- the fact that you have to deliberately activate and deactivate it before and after your run/bike is (at least with the latest phones) an arbitrary barrier to use. Of course you're going to weigh heavily toward the fitness crowd; you're losing the casual users by making them jump through hoops.

I started using Argus a few months ago, and it was a revelation. I run commute, and the last thing I think about when trying to remember to pack all my clothes and accessories is mucking with a tracking app. Now I don't have to do anything, and all my run data just appears automatically. I don't even have an M7 in my phone, so I pay the price by having to charge my phone every night (at least), but that's a trivial price to pay for the benefits.

Argus uses accelerometer data to detect running motion and activates the GPS based on that, so it doesn't work for biking, but a heuristic for detecting biking doesn't seem like an insurmountable challenge.

This is also why I think Strava's lunch is going to be eaten when/if Apple's purported health tracking system comes out. Dedicated fitness tracking will become ever more niche as the capabilities of general-purpose tracking systems evolve. I really hope that Apple and other gatekeepers keep their systems open so they enable rather than exclude groups like Strava, because the whole Strava Metro idea seems like a really cool effort, but I'm afraid I'm being naively optimistic.
posted by bjrubble at 10:05 AM on May 16, 2014

Dashy: "The data also crucially lack evidence of the locations most needing improvement - those are, by definition, completely dark in Strava's map. All those intersections of death, the roads that are just too narrow and fast to ride, the ones that would be a great route if only they wouldn't involve a side trip to the hospital, are not highlighted by this method."

Hmm. Like the RAF planes where you have to armour them where the returning planes don't have bullet holes.

I guess you could compare it to other traffic so you know this isn't just black because nobody ever goes there from here.

And looking through my city, most of the black roads that I know have high car traffic usually have a bright blue road running parallel to it one block over, which is probably a good solution.

What we'd really want to fix are black roads that have high car traffic but don't have any good parallel routes.
posted by RobotHero at 10:14 AM on May 16, 2014

bjrubble: Moves is pretty good at distinguishing between walking and biking automatically, so it's certainly possible. I'd love to know of another app with that functionality that wasn't giving its data to Facebook.
posted by aneel at 11:32 AM on May 16, 2014

Strava - even the free version - allows you to tag your rides as "commute". Should be pretty simple to parse out the heatmap data to show just self-described commuters.
posted by jetsetsc at 2:36 PM on May 16, 2014

The thing is, training or enthusiast routes shouldn't be eliminated from the data, either. Just de-emphasized so that we don't overlook lighter-traffic commuter areas, or new areas of investment designed to encourage more commuting or utility cycling.

Houston's been doing great stuff with rails-to-trails, so we've got some off-road paths that are pretty nice -- but they only occasionally go places I want to go, and the folks on them are generally going much more slowly than I ride, so I mostly ride on the streets.

Trails are nice, but for these reasons I'm kind of opposed to building them instead of emphasizing my right to use regular roads, educating drivers, and making the on-street bike lanes safer.
posted by uberchet at 2:45 PM on May 16, 2014 [1 favorite]

« Older The largest elections in Human history   |   The First Ad On The Moon Newer »

This thread has been archived and is closed to new comments