Capital Bikeshare is great, and I really love having it. This recent post over at DCist, however, reminds me that unbalanced stations are still one of the system's major flaws. I've experienced both full and empty stations, as I'm sure anyone who uses the system enough has or will.

(from Mr. T in DC on Flickr)

When I first heard about CaBi, my initial reaction was "won't everyone just ride them downtown in the morning and back out in the afternoon?" The person who was telling me about Capital Bikeshare completely brushed off the concern, saying "oh don't worry, there are guys with trucks to move them around".

Of course, guys with trucks can only do so much and can only move so quickly, and there are currently no other mechanisms in place to rebalance stations. Last summer there was an experiment called "Reverse Rider Rewards" that was designed to encourage CaBi members to rebalance stations during the morning rush hour. It ran during June, July and August; but I never really heard much about it after that.

Reverse Rider Rewards was set up so that a trip from a "typically full station" to a "typically empty station" earned the Bikeshare member a point and entry into a raffle. At the end of the month, prizes were given to folks with the most points and to the winners of the raffle.

It's not especially easy to declare the experiment a success or a failure. The chart below shows the daily number of trips from "typically full stations" to "typically empty stations" between 8am - 10am, for all non-holiday weekdays March through November (the three months before and three months after the contest period) and for only registered members (to be consistent with the rules of the contest).

(click to enlarge)

There is a bump in the number of "reverse" trips starting on June 1st. This is exactly what we would hope would happen. The biggest question is whether it's a big enough bump to make any difference in station balance. Unfortunately, the total number of trips from typically empty to typically full stations ("forward" trips I'll call them) clearly still overwhelms the number of "reverse" trips by a ratio of nearly 10-to-1.

(click to enlarge)

Was the experiment a success? It depends on how you define the outcome. There was a noticable jump in "reverse" trips during the contest period, but without a multivariate analysis, it's impossible to know whether Reverse Riders was actually the cause or if it was merely correlated by chance. On the other hand, Reverse Riders clearly didn't generate enough interest to offset the number of "forward" trips with "reverse" trips.

What else can be done? What about better incentives? Cash rewards instead of points and prizes? What about a negative incentive (ie. a "congestion fee" for all trips from typically empty to typically full stations)? Or do we just need to invest in more guys and more trucks? Maybe living with full and empty stations the price we have to pay to keep the system "free" in its current state?

The only evidence we have so far is that Reverse Rider Rewards maybe kind of worked a little, and that's ignoring the fact that there was surely some cost to administer the contest. If we're fine with the occasionally empty or full station, then that's the end of the conversation. If we're not, then more study and experimentation needs to be done.


    On June 08, 2012 Bossi said...

    It looks like the contest did more to spark interest in CaBi in general, as there looks to be a notable jump in reverse trips before the contest... but afterward the levels appeared to be rather sustained. I'd wonder about those couple of spikes... perhaps some media outlet gave it some coverage that day or the day before?


    Given the relative small range of trips (0-18) I wouldn't simply rule out that those two spikes in July are merely noise in the data.

    On June 08, 2012 Rob B. said...

    While one answer is more trucks, I think the better answer is more stations.

    Montreal is my starting point on this. When I've visited, the bikeshare trucks seemed to be all over the place. Maybe I was just looking for them.

    Montreal's density does not only help the issue of full stations. On one of my trips, a dock malfunctioned. Because of the density, there was another station 2 blocks away. Problem solved.

    In DC, the stations in Arlington are at the Metro stops and not near my destinations. If I were CaBi, I would not have implemented in Arlington until I could roll out a lot more stations.

    Yonah Freemark wrote a great blog post about this:

    On June 08, 2012 Rob B. said...

    Argh. Just wrote a long comment, and didn't make it here. Here's the short version:

    More trucks are great. In Montreal, you see the bikeshare trucks all over the place constantly rebalancing. But the bigger piece is station density.

    Montreal's station density is fantastic. So whether your destination station is full, or your the station won't accept your bike, there is almost always another one a few blocks away.

    CaBi in Arlington is too sparse. There are stations at the Metro, but no place to dock near my destinations.

    Here's Yonah Freemark's great piece on density:

    On June 08, 2012 Anonymous said...

    Seems the fundamental problem is that bikeshares are not intended to supply bikes for everyone to do a to-center-of-town in the morning commute, and a from-center-of-town in the evening return commute at rush hour.

    Living in NYC, I look forward to upcoming bikeshare system, but am not going to count on it for my usual work commute. That's what my trusty bike is for. However, on days where I'm not on the bike, and riding the subway or the bus, I certainly look forward to using it for connecting odd bits or trips, going crosstown, or short distances. And I think that's the point. The system works best within an area of density where there are reasons for people to makes trips in all directions, not so much for helping people get from less dense areas to the city center.