Monday, 2 May 2011


There are parts of the OSM database where it is important that contributors take a disciplined approach to the way that subjects are tagged. Data that is used to support subsequent processing will often only be usable if contributors stick to a relatively closed set of keys and values, and clearly understand their meaning.

Those responsible for the major renders and routing engines rely on sharing a common understanding with contributors in order to use the data they contribute. Contributors rely on sharing an understanding with the renderer if they are to see the results of their work. Hence both data contributors and those who use the data rely heavily on certain tags being well-documented with a closed set of options, that are applied reasonably consistently.

However, it seems to me that there are also some areas where a more open and expressive approach is appropriate. The database contains a mass of information contributed by different groups with different interests, and it needs to provide some space for them to explore and share ideas. In some areas a more flexible approach should encourage more creative and expressive contributions. On the other hand there are areas where there is too much diversity, and a greater degree of consistency would allow data to be used more widely. Neither is easy to achieve, and maintaining a balance between allowing complete anarchy and enforcing a highly structured approach is not easy. There is no single answer, but despite the difficulties, the OSM community  has (so far) proved remarkably resilient and effective in keeping enough of a balance to maintain forward momentum.

I suspect that by their nature the most committed contributors to OSM have a leaning towards more structured processes, and higher levels of standardisation. Unfortunately these tendencies can encourage a style in the wiki that is sometimes close to impenetrable, and can lead to authoritarian processes for policing the way some of these things are documented.

At the moment I am particularly exercised by the "historic" tag, which I've been trying to use in my rendering experiments. Although this tag is barely used in the major renders, it has been widely applied in the database. So presumably it is seen by a lot of contributors as having value. On the whole  it is used quite consistently. However, the range of values in the database is wider than the closed set that is documented in the wiki. There are also several areas where related topics, such as a change of use, or the current status of an abandoned structure are recorded inconsistently.

Although the "historic" tag is of particular interest to me at the moment, it is just one example of similar issues that occur elsewhere. As the scope of OSM content expands, and the mix of contributors widens, it seems to me that there are likely to be a growing number of areas where consistency might better be maintained by communicating general principles, rather than trying to police usage of a limited number of options.

To my eye the current documentation seems to be heavily reliant on saying "you can only this tag with these values, and they have to mean this". That's a sensible model for "highway", but it's not so good for "historic". There are discussions on tagging all over the place, but only the most determined will be prepared to plough through it all, and the core documentation on the wiki is a bit thin on "here is a good way to express what you want to say", "most people have resolved this kind of problem in this way", "there are several different ways of doing this, but no agreement yet on which is best", "there has been a proposal on how to define this, but it has not yet been widely adopted". That kind of guidance isn't terribly sensible for the  "highway" tag, but I think it sits comfortably with the way that "historic" (and several other tags) are being used at the present time.

I know I'm not the only one who finds some of the current approaches frustrating, but I'm not sure how best to help work towards a solution. Too much of the discussion of contentious issues is at the level of the playground, and some of the platforms where discussion is supposed to take place have become a joke. So what I've done instead is to start with a specific suggestion by proposing some changes to the way the "historic" tag is documented on the wiki - here.

Basically what this tries to suggest is a move towards providing guidelines on how to document historic subjects rather than (or in addition to..)  providing a catalogue of values with a limited number of options. It's as much about the style of approach as the specific content.

Maybe I'm trying to address a problem that only exists in my imagination. On the other hand, it may be a widely recognised problem, but this solution is seen as inappropriate. Or maybe there's a better alternative. Or (who knows) perhaps this is a reasonable start that just needs some improving. I would welcome comments. If I am somewhere in the right area, then it's important not only to get feedback from those determined souls who try to police the wiki, but also from the more occasional OSM users who might just want to use it (such as those who sometimes arrive here). If you don't have, and don't want, a wiki account, please feel free to leave a comment below.


Harry Wood said...

Yes. It's a strange kind of tug-of-war between people that are having fun mapping (and inventing new tags) and people that are trying to *use* the map data (and generally wishing OSM was more rigid with tags) Perhaps also in the case of 'historic' as well as people inventing new tags, people often want to be widen/narrow tag definitions.

Your points about "communicating general principles" is an excellent one. We can do a lot more of this kind of thing, and it's very helpful and productive actually, because debates about a specific tags become that little bit less endlessly re-occurring if we can point people at a general principle. This worked well in the case of Verifiability and I've been collecting a few other ideas of pages linked from Good Practice (mapping in general, not just tag principles)

On the flip-side however, we might aim to avoid a situation as on wikipedia where people often complain about all of the bureaucratic rules and policies (personally I don't find this a very fair criticism, but nonetheless it is a frequent criticism of wikipedia)

Doug said...

Hello - your blog has been recommended as I was looking for good cycling blogs to read. First impressions - really interesting!

gom1 said...

Thanks Doug - you must know some wise and perceptive people. You've arrived when it is all a bit heavy on OSM activity. Sorry about that. Normal service will be resumed as soon as possible.

gom1 said...

Thanks Harry, for your thoughtful comments.