Being point data rather than line data, it’s even harder now to imagine the background topographies that underlie these patterns of dispersion. But it does illustrate some of the key qualities that different systems can exhibit:
Size: Paris leads here, unsurprisingly. As Matt Yglesias recently emphasized, this is key to attractiveness of the system, hence the big starting scale for New York.
Density: Mexico City seems to have highest concentration of bike stations. It’s not clear to me that density of stations matters that much. Density of stalls certainly does, since that’s the essential unique commodity for share-cyclists who are trying to leave off the bike. But does it matter whether there’s one rack with 20 spaces, or two racks of 10 spaces only a block away from each other? When making that decision, street environment issues probably dominate.
Fragmentation: Seoul and several Chinese cities lead here. Without knowing the topographical constrains, this certainly seem as it would threaten the vitality of a bikeshare system. Isolated pockets of bike stations can probably only be justified if they’re near an extremely popular destination that is compatible with bicycling over a distance where no stations are available; and worth taking the risk that there are no mechanical problems along the way, while crossing the sea of no stations.
What I’d really like to see visualized for bikeshare systems is elevation. This, in my own experience, seems to explain many of the problems of bicycle drift that leaves some stands completely full and others completely empty. Different systems have different ways of addressing (or ignoring) this, such as Vélib’s bonus time for those who are willing to pedal up Montmartre to leave the bike. It would be interesting to compare different cities in terms of what elevation challenges they face.