5 ms·
OK, this is going to be longish. While OSM is similar to Wikipedia in many aspects, it is very different in others. In particular while both movements have a s
by sp8962 7y ago
OK, this is going to be longish.
While OSM is similar to Wikipedia in many aspects, it is very different in others. In particular while both movements have a supporting formal body, in the case of Wikipedia the WMF, for OSM the OSMF, that run (fsvo run) the central infrastructure and formally publish the data from a legal pov, in practice they are very very different.
This starts with the size of the budgets (the OSMF is roughly 1/1000 of the WMF one), to the fact that the OSMF is a membership organisation that determines its board in a democratic fashion (coming up soon), and so on.
But the relevant difference for this discussion is that the OSMF has never been involved with developing software (with one small exception) and that all the bits and pieces that make up what you experience as "OSM" have been written and maintained by the community. This has always included commercial entities and major parts of the editing API and the website have been written by developers on the payrolls of companies with an interest in OSM.
Editors (the programs that allow you to add to and change the data in the OSM database) have kind of always been an exception. The 2 in-browser editors Potlatch 1 & 2 that led to the breakthrough of early OSM were for the major part written by Richard Fairhurst, JOSM the most popular desktop editor is developed by a group of volunteers and similar things are true for most of the less known apps.
Back in 2012 it was clear that the Flash based Potlatch 2 had only a very limited remaining life in front of it and Richard started work on an PoC for a JavaScript based replacement that he named iD.
This gave Mapbox the idea that they could repeat what they had already successfully done once with Tilemill, to apply to the Knight Foundation for a grant to improve the OSM website and create a new JS based editor. Naturally this was not just out of the goodness of their hearts, just as with Tilemill this was also an opportunity to scale up the company. In any case Mapbox was awarded half a million $ for this work which led to the release of iD in May 2013.
Note that the current developers and maintainers of iD were not involved at this time, and I'm not even sure of any of the original devs are still with Mapbox.
In due course iD replaced P2 as the default editor on openstreetmap.org, that is the editor you get when you simply click on the "Edit" tab. This is a rather privileged position as for nearly all new contributors this is the point of first contact for OSM and can be formative for ongoing editing behaviour (while from a number of users iD is by far the most popular tool, JOSM outstrips it in the actual volume of edits).
Obviously the $500k didn't last forever and while there was one larger refactoring of the code that was worked on by a team in Mapbox, the involvement of Mapbox was obviously scaled down after the money ran out.
In 2015 (iirc) Mapbox hired the current lead developer and last year he was joined by a further developer employed by a consultancy. There have never been any formal communications from either of the employers as to scope of their current and future engagement, not even as simple as to where the money to finance this rather expensive undertaking is coming from.
Which brings us to the tensions that led to Frederiks blog post.
We've always had a bit of a dictatorship of the developers when it comes to editors, and all of us (including myself) tend to have "good ideas" (that now and then turn out not to be so good and have to be undone :-)) that are not pre-vetted with the community at large.
What is different with iD is its privileged position, and that because OSM has grown so much the much larger, the impact that changes to the default editors behaviour have, the IMHO in-transparent decision processes (who is -actually- calling the shots?) and the easy to offend devs. Combined this is rather an explosive mix.
The latest episode is the decision to start displaying brand logos in the editor and retrieving these online and mainly from Facebook. There are a lot of reasons why this is problematic, not the least that it violates EU privacy regulation without prior consent of the users (this is true regardless of the source of the logos, but obviously using Facebook makes the whole thing even more controversial). Not unexpected this led to complaints which in turn led to the devs locking down issue creation on the iD repo (this has AFAIK now been undone).
Obviously this has to be defused in one way or the other and running a patched version of iD on openstreetmap.org that is under more community control is one possible solution (though not my favourite).
Simon
- canada_dry 7y agoExcellent recap. Re: displaying brand logos in the editor and retrieving these online and mainly from Facebook. There are a lot of reasons why this is problematic, not the least that it violates EU privacy regulation without prior consent of the users I'd like to understand what the issue is here. I'm assuming it isn't just a matter of grabbing a logo (jpg) from Facebook. I assume iD is sending (meta) data about the person who is simply browsing the map to Facebook (i.e. without disclosing that info is being sent or what data is sent)?
- tmcw 7y ago> I'm assuming it isn't just a matter of grabbing a logo (jpg) from Facebook. I assume iD is sending (meta) data about the person who is simply browsing the map to Facebook (i.e. without disclosing that info is being sent or what data is sent)? Given the scale of the drama, you'd assume that it was something nefarious, but it isn't. iD loads images for brands from Facebook, and nothing more: here's the relevant code https://github.com/osmlab/name-suggestion-index/blob/66412f05f362f78893490d8c72e56174ac0e4860/build_wikidata.js#L431-L450 https://github.com/osmlab/name-suggestion-index/blob/66412f0... Fun story from back in the day: we added 'social share' buttons to iD. Knowing the fundamentals of web security, they were implemented as simple links, with no JavaScript component on the iD side, no 'trackers', nothing of the sort. But some buzz came out about how other Facebook trackers were up to no good, and the OSM community got in a tizzy about how we had installed nefarious tracking devices in the codebase, and after a few days someone recommended that we should implement the buttons as simple links, which is precisely what they were in the first place. That combination of bad faith and nobody reading the code was constant.
- rmc 7y agoEverything you do on Facebook is tracked, monitor and used. Everyone knows that. If you start requesting images of fast food chains from Facebook, we all know Facebook is going to analyze and utilize your new found interest in fast food.
- maxerickson 7y agoThe feature literally only pulls images from Facebook's graph api. It is of course a privacy leak, but there's not actually pushback on addressing that, just not a time machine to go back and fix it already. So there will be a privacy policy notifying users of potential third party data accesses and an option to turn that stuff off.