O.O'Brien

Having travelled to both Milan and Singapore in the last few weeks, it’s worth a note on the bikeshare provisions there.

Milan – BikeMi, Mobike, Ofo

Milan has BikeMi, a long-serving dock-based bikeshare system, which is one of the nearly 400 city systems that I have mapped in Bike Share map. It covers a big part of the city. There was a docking station close to both my hotel and to the location of the the conference I was attending, however far more noticeable were the numbers of Mobikes around the streets. Ofo also operates in Milan, although their bikes are not as prevalent.

I ended up using Mobike – mainly because I already had the app installed on my smartphone, for my occasional usage in London – and Mobike, like Uber, works nearly seamlessly across the various cities it operates in. This simplicity, combined with the new EU free roaming laws meaning that I can use the app without incurring roaming data charges, and the fact there was one to hire just around the corner, means that Mobike for me won over BikeMi. It’s a problem that dock-based systems are going to struggle with, unless they can somehow collaborate with each other globally. Perhaps a start would be JCDecaux “Cyclocity” systems allowing use across their cities if you are already signed up for another one, and a similar approach for Motivate and Nextbike systems. Although, as each city typically has a monopoly dock-based operator, this approach still has a more limited approach.

(I say nearly seamlessly – both apps struggle a little initially when moving to a new city. In Mobike’s case, the system boundaries generally don’t appear until the app has been restarted a number of times, after moving to a new city. For Ofo, the pricing indication is initially wrong – I’m pretty sure it’s not free!

Mobike and Ofo have not suffered in Milan like their have in London. They both operate throughout the historic centre of the city, and much of the suburbs, rather than being artificially constrained on a borough-by-borough basis like in London, or only operating in small sections of the city by operator choice/resource limitations, like in Oxford and Cambridge.

Milan does not have as much in the way of dedicated cycling infrastructure as London, however I generally found a good level of respect from drivers and felt comfortable moving around the city. It was exceptionally hot when I was visiting (35°C) and although I noticed a few other cyclists and also bikeshare users, there weren’t huge nunbers in London. I noticed that most of the Mobikes were single ones which had been left by the previous user, rather than stacked by the operator, and also more than once they had been left in a position entirely blocking the pavement and meaning the pedestrians had to get onto the street to get past. This is a big problem in many cities where dockless operators run systems – how to get users to park responsibilty. The solution – a combination of user attitudes, rules and facilities, is one that does seem to have been solved for the next city I visited…

Singapore – Mobike, Ofo, oBike/GrabCycle, SGBike, GBikes

In Singapore, there is no dock-based system but a wide variety of dockless operators have bikes here – as well as Ofo and Mobike, there are the home-grown SGBike and GBikes (the latter based on the SharingOS system), and OBike (cobranded GrabCycle) – which started out in Singapore. Both GBikes and OBike have ceased in Singapore, presumably their apps no longer unlock their bikes – but they are both scattered everywhere – or rather, this being Singapore, they are neatly stacked up against trees and fences. Will anyone remove the bikes for companies that have ceased to be?

Singaporians being extremely considerate people, I didn’t notice any bikes blocking pavements, or indeed showing any sign of lock tampering or vandalism. A small number of bikes had damaged wheels. Again, I have both the Ofo and Mobike app on my smartphone, but with no free data, and no particular need to cycle, I didn’t hire any bikes this time.

One thing I noticed over and over again is multiple operators bikes, being stacked neatly together. Sometimes, they were located within a painted rectangle indicating an “official” cross-operator permitted parking location:

Other times they were in “dead” pavement areas, such as traffic islands, where no one would otherwise need to be walking or driving:

There is clearly some city council direct involvement with the operators – in Ofo’s app, there appears a message from the Land Transport Authority about parking neatly:

(Incidentally, GrabCycle is I believe a multi-operator marketplace/broker. I haven’t seen this before for bikeshare, and Ofo and Mobike are not currently part of it – they are presumably “too big” – but I think ultimately cities with multiple competitive bikeshare operators will likely have to eventually enforce/insist on allowing the bikes to be used through marketplaces. No one wants to have six bikeshare apps installed on their phone, and no city wants six incompatible bikeshare bikes sitting on every street corner, each only useable by a separate dedicated app.)

The central part of the city, at least, is not as bikeable (or walkable) as Milan – with large roads and the lack of much in the way of dedicated cycle lanes, the only cyclists generally seen were on riverside paths and in parks. Pedestrians also can come to junctions without any crossings – except perhaps by a nearby overpass. It’s all very well having a marked parking bay, but if it’s beside a huge road with no cycle lane, is anyone going to use it?

The high temperature (30°C) and humidity, characteristic of the city state throughout the year, doesn’t make it the greatest place for comfortable or leisure cycling. I saw almost no-one using the dockless bikeshare bikes, even through are are loads waiting to be used, but I did see plenty of people on eScooters – both for-hire ones and private ones. It is only a matter of time before these become huge in London.

Intriguingly, there may note be a dock-based bikeshare system in Singapore – but I did notice these bike docks. They are “semi-smart” docks – they have an electronic lock that just locks the front wheel, but they are small and unobtrusive, presumably much easier and cheaper to put in than London’s hard-wired, “street furniture” type docks. They were, here, being used as spaces to keep Ofos, rather than by the bikes they were designed for:

I did notice a number of interesting varients of both Mobikes and Ofos. They have been in the city longer than they have been in the UK, and so a greater number of models are on the streets – but as well as older versions, there are newer ones that have not made it yet to Europe. These newer ones generally look chunkier, with bigger locks.

Here’s a Mobike with a rather nice frame painting detail, an unusually chunky looking lock, and solar panels in the front basket:

Here’s an Ofo with a very chunky appearance, particularly the tires, which look like they could withstand anything. They are solid, but with lots of holes drilled through the rubber to presumably reduce weight and possibly generate a warning whistle (not really):

Notice also the SGBike and Mobike parked neatly beside by other users. Off the pavement, off the road, not knocked down. London, you have a lot to learn.

I did get the feeling that bikeshare might be a spent force in at least the central part of Singapore (I did not visit the suburbs) and eScooters may already have taken their place here. Coupled with Singapore not being a particularly cycling friendly city – certainly compared to London – I wonder if the only thing remaining is for the city council to sweep up the unused bikes for the bust operators, and tell the others ones it’s time to move on? I understand that the council has recently tightened up the regulations for bikeshare in the city, making it harder for the companies to operate, but a necessity in a city with the layout and streetscape like Singapore’s.

There’s lots of bikeshare systems in the UK now. As well as the third generation dock-based bikeshare systems, fourth generation dockless (and hybrid) systems are starting to appear on various streets around the country, led by Mobike, Ofo and Urbo, three dockless providers and operators.

I’ve put together this simple model to try and understand where systems are most likely to be successful, for which I’m defining as a lot of (legal) journeys made with each bike placed by the operator. To do this, I split the country into its local authorities, apply three scores, and then multiply them together to produce an overall “propensity for bikeshare”, or PFB score (the name is a nod to PCT.bike from Lovelace et al) which can then be ranked.

Many mid-sized cities in the UK have their own local authority, approximately covering the urban area, while Manchester and London are split into multiple LAs. Conversely, some LAs individually cover multiple smaller towns/cities and large rural areas too. Hence, this is a very simply model which is not going to be completely fair to every urban area (Stirling, in particular, gets pushed well down the rankings as its LA includes a huge rural area). Still, for most of Great Britain, it produces results I would anticipate and is a good start towards potentially developing a more sophisticated model. The local authority geography is also appropriate, as local authorities act as the gatekeepers for which access is negotiated (for systems which pay themselves) or as authors of bids for subsidised systems.

Model Inputs

For this first, simple model, my three compounding factors are:

1. Residential and workplace population density – on the assumption that bikeshare systems need a critical mass of people passing by their bikes/hubs/docks, in order to be seen and used sufficiently frequently to justify the costs of equipment/maintenance, on the basis that a major source of income is per-use fees. Both residential and workplaces populations are used, as people have journey opportunities that can be facilited with bikeshare, both from their office (e.g. lunchtime errands, commute to evening socialising or back home) as well as their home location.
2. Proportion of people who already commute by bicycle, again looking at both workplace location and residential location (bearing in mind that many commute journeys, particularly in London, cross local authority boundaries). While such people are less likely to convert to bikeshare, as they already have a means of cycling, their presence on the streets and associated culture and facilities (e.g. bike coffee shops, marked cycle routes or existing cycle-friendly infrastructure) help normalise the idea that cycling is a possible option for a journey need.
3. Vandalism rate – theft and criminal damage to bikeshare bikes happens and it is an expensive one for the operators. It can be what causes systems to fail – particularly if they run out of bikes, or broken bikes litter the landscape and turn public opinion against the concept. While the dock-based systems are less vulnerable to this, as the bike is either securely locked to a clamp or in the hands of a paying user, dockless bikes are particularly vulnerable to vandalism, as they are not secured to immoveable objects, and their locks are, unfortunately, relatively easy to break by a determined offender looking for a free bike. The tendencies for vandalism of property that is not yours does vary significantly around Great Britain, while stereotypically it is likely to be more urban areas and areas with a younger and less educated population that is more likely to vandalise, using actual crime data on vandalism allows a more nuanced approach to be taken. Other crime classes (e.g. theft) were also considered for this model, but I think that vandalism rates act as a good proxy for how the local population around a bikeshare system will “care” for it or abuse it.

I have not included the absolute populations of local authorities in the calculation, as in general, with one exception (Isle of Scilly), LAs all have a significant night and/or day population, so they are all large enough to have a self-contained system. Another obvious factor, hilliness, is likely already correlated with proportion of cycling commuters and so is not included. N.I. is excluded from the model for now. Data sources include the latest police crime statistics (with populated-weighted averaging when across multiple LAs), and census data.

Results: Propensity for Bikeshare by Local Authority

Here are the results of the model run. Clicking an underlined title takes you to the main bikeshare for that area – forthcoming systems in brackets.

Some notes:

• London boroughs score consistently highly, and even London as a whole, which is interesting as outer London is anecdotally not known as a particularly cycling-friendly place. Considering the size of the city, and the intimidating conditions cyclists often have to put up with in much of the capital, it is great to see it scoring so highly here.
• Bikeshare operators are doing their homework and generally, the top end of the list is already well populated with bikeshare systems, in some cases multiple systems are competing.
• The top local authorities without a bikeshare system (operating, announced or consulting) are Haringey, a north-London inner-city borough which has been surprisingly quiet until now, and Merton, an affluent outer London borough to the south. Oustide of London, the highest ranking areas without a bikeshare are Portsmouth (small system launching this summer), Gosport (neighbouring), Gloucester, Poole, Worcester and Hull. With the exception of Hull, these are all southern English urban areas, with generally affluent populations and some established cycling culture.

See also my London borough bikeshare scorecard.

My research lab, the Consumer Data Research Centre, is making available much of the docking station empty/full observation data that I have been collecting, at frequent intervals (up to every 2 minutes) for generally the last 5 to 8 years, for over 50 cities around the world, to facilitate and encourage further quantitative research in the field. I already get numerous requests for this data – it is very interesting for all kinds of research projects, particularly because of it spans multiple years, so introducing this new mechanism is a good way to manage these requests. You can see the cities listed at the CDRC Data product page.

For three cities (Cork, Limerick and Galway in Ireland) we have directly attached the data to the record page. It is the data that has been collected up to the point that the record was published a few days ago, and on request we will reload the data, capturing more recent observations.

For the other cities, the data record is “Safeguarded”. This is because the logistics and technical limitations of attaching the very large amount of data to the records. Namely, it takes quite a lot of time to prepare each dataset for download, and the platform we are using is not designed for hosting extremely large files – plus, it is likely that researchers would want the most up-to-date data, meaning that we would need to build a mechanism to update the record regularly. Using an application process also minimises spurious requests – we have invested time both in collecting the data and processing it, so we need to be confident that it will be used. Additionally, some systems (particularly those in the USA) come with their own data licence restrictions from the operating companies meaning that we cannot freely distribute the “raw” data. In Europe, most of the datasets are explicitly open, meaning use of the data is unrestricted (although normally requiring attribution). the Irish cities listed above have a slightly more restrictive licence, requiring us to distribute it on the same terms, which we have done.

The data is available on application to interested institutional researchers. In practice, this means academic, public sector and non-profit organisations who which to carry out public/publishable research with the data. Application details are attached to each record.

Above and below are simple graphs produced from the data for various cities. I have looked at the number of bikes available every day at around midnight and plotted it on a simple graph against time.

Access the data here.

Stay tuned because I am planning on releasing two further “data portals” of bikeshare data, soon. These are slightly more manageable in terms of file size and administration, so I am aiming for these to be directly downloadable.