Cycling infrastructure between towns

[aktiivimallikansalainen]

Sorry for repeat posting, but as Tolstoi21 reminded me of the Finnish discussion, there’s another way to approach the basic concept of cycleway hierarchy if that’ needs further justification.

Helsinki is building a network of cycleways with stated design criteria (Finnish link), freely translated by me:

”Baana” are direct, high quality cycleways that make it possible to ride at a constant speed. When completed, the ”Baana network” will form an approximately 140km regional cycleway network, that connects major suburbs to downtown and commercial districts.

(Parts of this network have existed for years now.)

To me this sounds analogous to how main roads would be described, both the promise of a high and uniform quality level and the intended purpose of connecting separate areas. Is there opposition to hierarchy-for-cycleways tagging in this scenario?

[Richard (Richard Fairhurst)]

@aktiivimallikansalainen

Is there some reason this is not desirable that I’m failing to understand?

It’s not desirable if it doesn’t reflect an on-the-ground designation. We don’t just invent our own classification of “this is the most important route from A to B”, even if it makes our lives easier. We need to evidence that from something observable and verifiable.

Fortunately…

@aktiivimallikansalainen

”Baana” are direct, high quality cycleways that make it possible to ride at a constant speed. When completed, the ”Baana network” will form an approximately 140km regional cycleway network, that connects major suburbs to downtown and commercial districts.

This is exactly what we map. These routes would be mapped as route relations, just as they are in many other countries and cities. They are officially designated and (presumably) signposted routes, observable and verifiable. As such they’re fair game for recording in OSM.

@aktiivimallikansalainen

From a very brief look, it seems that there are areas where the logic doesn’t work as expected (small patches of major cities, or indeed the entire town of Inca, Mallorca). Am I wrong to assume you’ve spent quite a bit of effort fine-tuning this for the areas your map mainly focuses on?

Bugs happen. Perhaps in this example the built-up area polygon might be degenerate and therefore the spatial query is failing. Not sure, I haven’t looked at this particular example.

[balchen]

@Richard

It’s not desirable if it doesn’t reflect an on-the-ground designation. We don’t just invent our own classification of “this is the most important route from A to B”, even if it makes our lives easier. We need to evidence that from something observable and verifiable.

That’s not clear-cut either. The definition of highway=tertiary in Norway leaves a signficant judgement to the mapper. highway=secondary requires you to have access to the government road database to tag it according to the NO wiki, and other roads may even require you to know their numbering history to know for certain how to tag them.

I’m sure this isn’t unique to NO.

Leaving some judgement to the mapper for cycleways doesn’t seem unreasonable.

In NO, all cycleways are owned by someone. Typically the county will own the cycleways that are related to county roads, whose ownership define them as either primary or secondary roads. So we could use the same ownership as a basis to tag these cycleways as primary or secondary cycleways.

Etc.

@Richard

This is exactly what we map. These routes would be mapped as route relations, just as they are in many other countries and cities.

They’re not always part of a route relation.

[Richard (Richard Fairhurst)]

@balchen

That would be awesome!

What would “performant” be in this case?

But: the cycleways are never continuous around here. How would you bridge the inevitable gaps? How long can a gap be? Etc.

This requires a bit of a history lesson.

In 2005, researchers noted that most intercity car journeys follow a similar pattern. Broadly speaking, you take a residential road from your house onto a distributor road; then you follow that distributor road to a main road; then you follow that main road to a motorway; you do most of the miles on that motorway; then you turn off onto a main road, then onto a distributor road, then onto a residential road which takes you to your destination.

Essentially, highway systems are ordered according to a hierarchy, and most long-distance journeys simply navigate up that hierarchy and down again. A router can exploit this to find an A–B route quickly. Rather than exploring every single turn along the way, they trust in the roads higher up the hierarchy to return results more quickly. This algorithm is called Highway Hierarchies.

The signposted highway system is usually the fastest way by car from one place to another, but not always. Not all routing applications have signposted hierarchical highway systems (e.g. walking and sometimes cycling). So what if we could devine our own hierarchy?

That is what the followup algorithm does, Contraction Hierarchies (2008) - still one of the core algorithms used by route-planners today. Essentially, this constructs a hierarchy by looking at the speeds/distances on each road, rather than just trusting in the official classification. It might find, for example, that the best driving route from Gloucester to Ross-on-Wye is A40->B4215->B4221->M50->A40, even though that requires descending the formal hierarchy (A40->B4215) and then ascending it again.

It then records that route in its internal hierarchy as “best route between Gloucester and Monmouth”. As it builds the hierarchy upwards, that route may in turn form part of “best route between Oxford and Newport”. And so on.

In other words, route-planners have already worked out a network of the best continuous cycle routes between any two points. That’s how cycle.travel and other sites are able to find you the best route from Brussels to Rome in under a second - they use their computed hierarchy. It doesn’t matter whether there are gaps in the formal cycling infrastructure: the route-planner will find its way. If you want to show a network but there’s no on-the-ground evidence that you can map in OSM, this, I think, would be the place to start.

[aktiivimallikansalainen]

@Richard

We don’t just invent our own classification of “this is the most important route from A to B”, even if it makes our lives easier. We need to evidence that from something observable and verifiable.

First of all, we’re talking about the only route(s) here, not anything subjective like “most important”. Second, there are any number of up to tertiary and unclassified roads that do not have any on-the-ground-designation to distinguish them from residential roads. (Tertiary roads should have an id number in a database somewhere.) What they do have, as do the cycleways I’m talking about, is an on-the-ground observable fact that they lead to to another road or the next village, and not just to someone’s yard.

@Richard

This is exactly what we map. These routes would be mapped as route relations, just as they are in many other countries and cities. They are officially designated and (presumably) signposted routes, observable and verifiable. As such they’re fair game for recording in OSM.

But these are quite different from route relations. These are purpose-built, uniform ways. You can follow these without any signposts because, well, they look like a duck. (I use one to commute to work, and if there’s a sign somewhere I have not noticed it)

A route relation has no inherent promises of the quality of infrastructure. A route can combine cycleways with parks, forest road, quiet residential roads. Or highways if we go to things like Eurovelo. A route physically exists as signposts, not as ways.

I understand that some countries have solved the same underlying issue (efficient routing for bicycles) using routes and node networks, but the on-the-ground realization is completely different from what this is. Don’t get me wrong, I would happily pay double taxes for a few years if that got us the kind of node network Belgium has, but the reality is that we don’t.

And as we have something different, and I’d like to make the best of it, play to its strengths, instead of trying to forcibly map into something it’s not.

[SomeoneElse (Andy Townsend)]

@Richard

We don’t just invent our own classification of “this is the most important route from A to B”, even if it makes our lives easier. We need to evidence that from something observable and verifiable.

Just to add a bit about this - my recollection of the cycle routes in Helsinki (from a while ago) is that there may well be verifiable differences. For example, my recollection is that the main north-south cycle route through the greenery here is definitely signed as such. The main cycle routes in te development to the northwest are wider, have a dashed line separating traffic in each direction and have winter service (the local cycle connectors to individual places may not). For example, here is a Google Street View link of this OSM way that shows the sort of thing - and that OSM way already has a bunch of extra tags that a mapmaker could use to decide to show that cycleway ahead of some others.

[Tolstoi21]

@SomeoneElse

Just to add a bit about this - my recollection of the cycle routes in Helsinki (from a while ago) is that there may well be verifiable differences.

This is true. In fact, just a little westward, the way you linked to changes into this bicycling superhighway! Two wide bicycle lanes separated from pedestrians by inlaid stones, and excellent winter maintenance. @balchen s proposal of cycleway=expressway fit the bill perfectly for these fabulous ways!

The problem (for @aktiivimallikansalainen’s point) is that these superhighways exist in Finland only inside cities (well, there may be one or two between Helsinki and Vantaa, and perhaps one goes a short distance from Helsinki to Espoo), but generally cycleways between cities are definitely not in this category. There are cycleways between many cities, but they are of the very ordinary variety (i.e. possibly bicycle=designated but almost always also foot=designated and mostly segregated=no in that case. Often smoothness=intermediate, even if surface=asphalt).

[balchen]

@SomeoneElse

For example, here is a Google Street View link of this OSM way that shows the sort of thing - and that OSM way already has a bunch of extra tags that a mapmaker could use to decide to show that cycleway ahead of some others.

It may work in sone cases. In the case of my map, I’ve already exhausted those options. There are still a lot of cycleways I would like to handle differently.