Balancing Bike-Share Stations Has Become a Scientific Endeavor | Hacker News Reader

Balancing Bike-Share Stations Has Become a Scientific Endeavor

citylab.com

35 pointswallflower20 comments

Most of the current systems appear to use manual rebalancing - if bikes aren't evenly distributed someone from the company operation the bike share program manually distributes bikes from one station to another station.

I wonder how well an "economic" approach to this problem would work: Dynamically price the cost of a bike based on the station where the bike is taken and on the station where the bike is returned too. Then, if you want to distribute bikes from one station to another station you could assign a negative price for these routes, and thereby increase the traffic in that opposite direction.

Assuming that the traffic behavior is relatively stable you don't even need dynamic pricing, but instead might be able to announce the pricing for a particular route weeks in advance.

I came here to point out this exact same thing. Why use vans and trucks to re-balance stations when you can have people re-balance them?

Not only can you can have dynamic pricing based upon stations and times, imagine having a feature where you can "call" for a bike at an empty station. There can be an app that notifies people near other close non-empty stations to ride a bike over for you. Or, you could even display it on the screen of the kiosk.

How would you get people to ride these bikes for re-balancing or on demand?

Offer something like a rewards plan (re-balance a bike X number of times and receive a discount on your next bike rental) or pay them in micro-payments (Think Tom Hanks returning luggage trolleys in The Terminal).

How could you ensure that people riding bikes on demand don't take advantage of the process to just get a free ride to wherever?

First, make sure that the bike being taken out is checked into the station it's being delivered to. Second, impose some sort of time limit in which that needs to be done. Base the time limit on an acceptable pace and take into account the distance between the stations (with a time buffer). If it exceeds that limit, it becomes a regular rental and the deliverer is charged regularly. Also, it then becomes possible to add bonus incentives for getting the bike to the station faster.

One particular bonus of this system is that it can leverage a city's homeless population as a workforce. (If you disagree due to cleanliness issues, I'm sorry to tell you that the homeless already use citibikes as cycling machines). Of course, there would be a need to hedge against the possibility they may just take out a bike and never return it, since they don't really have any sort of credit. One possibility would be to sponsor a homeless person by covering the cost of a bike and then having them operate on a "one strike" policy.

In Paris, the stations offer you an economic incentive. Unlike in Manhattan which is comparatively flat, riders in Paris enjoy coasting down from Sacré-Cœur and not so much peddling back up again.

So vélib encourages you to ride uphill:

> Return your bike to one of 100 stations perched over 60 meters above the rest of the city, and 15 minutes of free riding is added to your account. Riding uphill is now a power-up. The free minutes are a bonafide reward, too, saving money on destinations that are often outside of the free 30-minute radius. More bikes make it back up the hill. Since the program was launched in June, over nine years of bonus minutes have been awarded.

-- http://globalmoxie.com/blog/velib-bonus~print.shtml

The problem is people don't pick their routes based on cost, they pick their routes based on where they're trying to go to.

Exactly. If this were implemented, it would affect me one of two ways*

1) The cost variation would be small enough that I'd ignore it.

2) The cost would sometimes be high enough to dissuade me from taking a bike where I wanted to go, in which case I would quit the program completely. I don't want to have to change my travel plans at the last minute. Even if the extra cost were non-trivial but acceptable, the mental overhead of having to decide whether it's worth it would make the program much less appealing.

* If I were actually participating in a bike share program. Currently I own my own bike, but I could see using a bike share program in the future.

I think what he means by "negative cost" is that you actually pay people (in credit), to take the bike to the desired station.

Yes that's what I understood as well, although that's just one component of "dynamically price". And even that isn't really going to work. That's equivalent to just paying someone outright to rebalance bikes, except it's ad-hoc, only moves one bike at a time, and significantly complicates the process.

I agree that making money through riding a bike from a low to high demand station would be an edge case, but I don't see how it adds significant complication. You just list what the price would be at different stations if they return the bike there, with some being refunds.

Which is what they do already, except they pay employees/vehicles/fuel.

The providers have some philosophical problem with paying regular people to do the very same.

Once they overcome this, the system will work a whole lot better and more profitably.

See the paper linked in the post below. It deals with exactly that 'economic approach' in addition to traditional repositioning with trucks. But with positive customer incentives instead of added costs.

There's a freely accessible preprint on arxiv under the same title if you cannot cross the paywall.

By the way, incentives do not need to be monetary. In Paris, they give people free minutes for driving bikes to stations that are uphill...

In Chicago the Divvy bike redistribution is horrible. It's not even a hard problem. Need to move bikes into the Loop at 5PM? Give people a free day pass in the future for moving a bike to a high-demand area.

I recently published on the same topic. [1]

In many bike sharing systems (such as in London), there is a daily repeating cycle, where too many people want to get to the city center in the morning and back out in the early evening. You know that will happen. So countermeasures can be taken early on.

[1] Dynamic Vehicle Redistribution and Online Price Incentives in Shared Mobility Systems, http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumb...

Why is this a problem? If X people take bikes into the city center in the morning, and then the same X people take bikes back out in the evening, that sounds balanced to me.

That only works if there are enough bikes for each bike to be used only once during the morning or evening commute. If you need to reuse a bike multiple times in one morning (say at 8am, 9am, and 10am) to serve everyone, it has to somehow go where someone will use it.

How does London handle that rebalancing then? Physically moving bikes with a lorry?

At one point they had nearly 30 repositioning trucks in London (2-3 years ago). At Waterloo Station they basically hand them off trucks directly to the arriving customers during rush hour.

Maybe the real solution is social rather than technical?

How many move around to different places for the sake of internet points? Seemingly lots of them. http://en.wikipedia.org/wiki/Ingress_(game)

Can we just convince them to balance our bike shares instead?

I wonder how this problem is solved by car2go. There are no stations to drop the vehicle, and cars can not be transported in bulk. Would it help to get rid of the stations for the bikes?

In Germany, Call-a-bike (Deutsche Bahn) [1] operates exactly like this. Bikes can be rented everywhere inside the city, and can be left on every junction within a city. There are no drop-off/rental stations.

So far, I did not have any problems to find a bike when I needed it.

[1] http://www.bahn.de/p/view/service/fahrrad/call_a_bike.shtml

How do these systems work if there are not enough stations to return your bike?

Reply on news.ycombinator.com