Monday 5 October 2009

The middle ground

The OSM-talk mailing list has descended into farce.

There are people who want OSM to have a leader, some sort of benign dictator, who will arbitrate over issues like how to tag things properly and people who do not want this at all. There are people who respond to every post regardless of how little they know about the subject and people who get irritated by this. There are people who believe that everything can be solved by writing a wiki page and those who largely ignore the wiki. There are people who believe that new-comers should be vetted or guided before they are allowed to edit the real database and other who create tools to make edits by a new-comer even easier. There are people who think there should be a proscriptive list of tags and others who don't. There are people who think every change to the list of tags needs to be agreed with a vote and others who dismiss voting as meaningless. We have seen our first public resignation from the list, but how many more people have just given up on it and walked away? People are beginning to become entrenched and affiliations are forming on one side or another, yet all these issues crossover to some extent.

So, where do I stand?

The calls for a leader are well meaning but naive. SteveC has been held up as such a leader but is someone with vision the right guy to go through the minutia of shield shapes, shop types, road colours, surface type etc? I don't think so. Is it likely that one person can have the knowledge and the foresight to make binding decisions that will apply across all countries and legal jurisdictions? I very much doubt it. Will one person understand the reason that everyone contributes, so what tags really mean to them? Certainly not.

In spite of a few people shouting it down, open tagging works very well. A consensus has emerged about almost everything. Some people want that consensus carved into stone tablets so no tags outside of the 'rules' can be used. So what will they do to the people who don't follow these rules? Throw them out of OSM? The outcasts will then just take the freely available data and start again (and I'd join them). If you don't throw them out then the only other way forward is to change their tags and start an edit war.

The calls for consistent tagging often mention how much harder it is for software using the data to make good use of inconsistent data, yet these people don't seem to make use of the data - the people who do use it occasionally mention that it really is no problem.

it has been said that some people don't map an object because the tags are not clear. I wonder how much real tagging in the real world these people have done. The real work is gathering data from the ground. It is easy to change a set of tags later rather than go out to an otherwise well covered area to gather the data again for a feature you deliberately ignored.

Some people favour Taglist. This shows the number of times a tag is used allowing a consensus to form, but, like any kind of evolution it has no guiding hand. We get convergent evolution where two different sets of tags mean the same thing. We get broken tags where no thought went into them so they don't stand the test of time.

The biggest problem area is the lousy wiki that is used to document things. Creating wiki pages is easy. Creating a useful wiki page is much harder. Creating a wiki page that is useful and stays useful after various people have tinkered with it is very, very difficult. Creating a wiki page for a new tag is easy. Assimilating the comments to make a tag better is tedious but can be very productive if it is done well by someone willing to accept ideas and not just push their own agenda. The final voting system to then adopt a tag is meaningless. So few people vote compared to the number of active mappers that a majority is worthless and sometimes ignored anyway. Once the tag's wiki page is promoted to Map Features it gets changed later without voting anyway.

Open format tags are very useful for our world. Very few things in life are black and white, so extreme views like 'closed tag lists vs anarchy' really don't help. People contribute to OSM for many, many reasons and all are equally valid. Their reason, their culture, their language, their temperament, their environment, their software and their experience all flavour the tags they use and the things they tag. If we close the list of tags much of this will be restricted.

Do I have any answers?

Create a tagging list to get the issues with tagging off the Talk list. Moderators on busy lists should remind people to exercise restraint, both in not just commenting blithely on every message and also not repeating the same argument over and over. Scrap voting on tags - it gives a false air of importance to the process. Scrap the wiki for tag documentation, replace it with some solid tags in a harder-to-edit form and emphasis Taglist and open format tagging.

These are my ideas. They are not binding. In no way do I intend these to revoke anyone's human rights. You are free to ignore them and/or express ideas of your own, but please, please don't start arguments on OSM-Talk.


David Earl said...

You will have seen my proposal for just such a mechanism forming part of the API, posted at previously on the dev list and almost the same time as your post on talk:

maning said...

here, here. The talk list is getting too noisy these days.

Let the tagging evolve, we do not need any intelligent design to muddle with the tags.

Gregory Marler said...

...There are some people that don't have the time to keep up with the lists/blogs/forums/wiki, but would like to know what's going on.

I believe the tagging problems have been in OSM for most of it's life. And it's also messed up due to historical reasons such as highway=footway being an early tag. Is it the legal designation, or the physical description?

I say the answer is in what gets used (rendered/searchable) is what should be tagged. I also say render-stylists/programmers should make use of what's being tagged. Big chicken and egg problem, but it's what makes OSM fun and a rare bottom-up management, it's how things like the Cycle Map evolve into popularity or speciality.

One thing could be to create a table of tagging and what makes use of them. But who spends the time updating (need to have a knowledge of all the renders code) and which uses are listed. Another chicken, another egg.