[Osmf-talk] Death and evolution

Tom MacWright macwright at gmail.com
Fri Sep 26 14:47:33 UTC 2014


> You will suddenly find out your shiny and easy-to-use tools for
processing OSM data, which everyone, including working groups, have
embraced, are written in a way only a company's programmers can
understand and update.

This is absolutely absurd and entirely incorrect. John and I didn't spend
hours writing documentation and blog posts about the internals of iD for
these to be ignored and written off. Here, take a look:

* https://www.mapbox.com/osmdev/2013/02/26/id-architecture-part-1/
* https://www.mapbox.com/osmdev/2013/02/27/id-architecture-part-2/
* https://www.mapbox.com/osmdev/2013/02/28/id-architecture-part-3/
* https://www.mapbox.com/osmdev/2013/01/15/oauth-in-javascript/
* https://github.com/openstreetmap/iD/blob/master/ARCHITECTURE.md
* https://github.com/openstreetmap/iD/blob/master/API.md
* https://github.com/openstreetmap/iD/blob/master/NOTES.md

If you have other specific questions about iD's architecture, I'm happy to
answer them and explain it better, as long as you listen.

Otherwise, do not confuse your difficulty with d3 or JavaScript with a lack
of effort on the iD team's side of documentation. iD is heavily documented
inside and out, for open source contributors like ourselves. And we
constantly accept contributions, handhold pull requests, and listen to
thousands of messages worth of feedback.

Here's my opinion on why open source people don't like working on OSM stuff.

OpenStreetMap antagonizes programmers. The diaries, the mailing list, the
github repos, the IRC channel: it antagonizes people who work on code.

If you write an editor, people will accuse you of destroying the map if you
ever write a bug. If you work on design, you'll get tens of angry folks who
like it better before. If you work on task management, you're reinventing
the wheel and muddying the waters. If I were to work on, as you mention, a
"reverter", the amount of blowback over its potential misuse would be on a
whole new level. If you work on import tools, you're enabling evil imports.
Every mistake you make is a terrible mistake and nobody has patience or a
sliver of tolerance for human fragility.

The other issue is that OSM is slow: if you want to land a PR on osm.org,
you'll need to wade through a hundred github comments before it lands, and
then you post a diary entry and a mailing list entry, and when it finally
lands, somehow "nobody was consulted". The amount of overhead to procure
the absolutely impossible "sign off and approval of thousands of people" is
insane.

But, the second issue is fine. Programmers are okay with process want to
work in communities. Nobody is okay with meanness.

On Fri, Sep 26, 2014 at 9:44 AM, Paweł Paprota <ppawel at fastmail.fm> wrote:

> Ilya,
>
> I 100% agree with you on every single line of your e-mail. I can only
> add that some 1.5-2 years ago I tried really hard to make OSM my job - I
> literally resigned from my day job and moved into OSM consultancy +
> working on some major (read: actually HARD, not flashy) new stuff for
> OSM (OWL and "social" layer for osm.org).
>
> Unfortunately this turned out as one would expect - I could not make it
> work from financial perspective and in March 2013 I just had to get a
> regular job. Later in 2013 I started my own IT company working on a
> product unrelated to OSM and since then I just don't have any free time
> anymore...
>
> From my experience I can tell you that single most heartbreaking thing
> is that there is just a handful of people who are able (yes, I do mean
> in terms of skills) and/or willing to work on complex issues. OSM data
> is extremely hard to work with - if you want to add some feature (like
> OWL for example) that takes into account worldwide data and perhaps also
> full history you are in a world of trouble with literally terabytes of
> data in Postgres, problems with indexes, data structures, algorithms -
> everything matters because otherwise your tool/feature will just be VERY
> slow.
>
> Things like iD or map rendering style are low-hanging fruit and it's
> good that they are progress but who will take care (of course for free
> out of goodness of their heart!) of better history, activity stream,
> routing, search that does not suck and myriad of other things that OSM
> should have.
>
> I said this multiple times in various forums (IRC, EWG, mailing lists) -
> without good developers who are PAID to work on OSM, this stuff will
> NEVER happen. It is very frustrating to hear that some people who are
> high-ranking OSMF people who theoretically (?!) should be on the
> forefront of helping OSM grow say that it will work itself out, organic
> growth FTW, someone will pick up these tasks etc. This will not happen,
> I can promise you that as I did almost 2 years ago - and since then to
> this day none of the tasks from top 10 are completed.
>
> Paweł
>
> On Fri, Sep 26, 2014, at 11:29, Ilya Zverev wrote:
> > Hi!
> >
> > Imagine Wikipedia without a strong Wikimedia Foundation. "Just the
> > servers", they say. Imagine WMF stating that the only resource they
> > possess and plan to is a gigabyte of textual articles, and their only
> > task is to support its openness and accessibility. No Wikidata, no
> > wiktionary, no wikimedia commons. Imagine Jimmy Wales employed in some
> > third-rate encyclopaedia manufacturer, travelling around the world and
> > speaking of wikipedia like it's 1999. Because nothing has changed, and
> > he is still relevant, even having left the project.
> >
> > Hilarious, isn't it?
> >
> > We had quite a good time mapping things and publishing planet files.
> > When I was young and optimistic, I laughed at promises of OSM demise:
> > our data is open and community is big, nothing can beat that. But with
> > time, I see that neither data, not community gives us anything. We do
> > not evolve. When was the last time OSM project had big news? Ah,
> > spring of 2013, when iD editor was published. Was it made by OSM
> > community? No, it was created thanks to a grant by Knight Foundation
> > and MapBox programmers, which are not actually interested in OSM
> > growing, if it doesn't affect their profits.
> >
> > I've always been wary of MapBox, because they are the first, very
> > faint sign of problems in OpenStreetMap governance. Do-ocracy is good
> > for small projects, but now it literally stopped OSM from evolving.
> > Nobody does anything, and if someone submits a pull-request, well...
> > You know, our github is full of wonderful PRs, which won't be merged
> > in near future because of bikeshedding on an epic scale. Hence, the
> > power of change in OSM is not in community's hands, but belongs to big
> > companies outside the project. If a company with ten to twenty
> > programmers appears and wants to change OSM to their liking, it would
> > be very easy for them to do. In do-ocracy, a team with more programmers
> > wins. You will suddenly find out your shiny and easy-to-use tools for
> > processing OSM data, which everyone, including working groups, have
> > embraced, are written in a way only a company's programmers can
> > understand and update. Your data model is set in stone, or else you
> > would have no tools. Or your tagging or some rules, like for imports,
> > were slightly altered, and you notice that only after a year or two,
> > when thousands of novice mappers turn out to see schemes and models a
> > bit differently than you. All that because you, the OSMF, have lost
> > the control, telling yourselves you're responsible only for data and
> > licensing, and for anything else you don't have the means.
> >
> > When data consumers start improving the project, they focus on things
> > they need. Rendering and tile stack have significantly improved in
> > past year, especially when vector tiles became the trend. We have a
> > lot of routing engines, geocoders and map styles. Startups are happy
> > to use our data offline. But — what do mappers have? Oh, JOSM editor,
> > which is still considered extra hard, and which hasn't really turned
> > smarter or easier than it was in 2006. If you want to edit relations —
> > the ultimate challenge — that's the only editor you have. Ten years
> > have passed, but we still, even in Europe, print screenshots and go
> > outside to draw on them. We collect low-quality GPS data and pretend
> > it's the best source. We have zero tools for mapping from behind
> > a wheel. We have no classification of tags. No reverter that can be
> > used by more than 0.001% of users. Nobody had time to make any of
> > those.
> >
> > Every year with the Board elections we hope for change. We wish to
> > elect somebody who will be active, who will take the OSM governing
> > structure and evolve it into something greater, so we won't have such
> > ridiculous problems every mapper faces daily. We hope for better
> > tools, for more funds, for wider publicity. We secretly with for new
> > API. But you won't get it. Nothing will change. Well, maybe we would
> > get some explanations for licenses or working groups processes, but
> > you, ordinary mappers, won't get a thing from OSMF. I see some members
> > writing code and publishing editors, but would they do it not being
> > OSMF members? I'm sure they would.
> >
> > Here I planned to dive into tons of promises candidates to OSMF Board
> > have written over the years in their manifestos, like growing the
> > community, promoting the diversity, favouring volunteer code
> > contributions and so on. Nothing has changed substantially. Of course,
> > I see Simon working on a lot of texts. I see Henk helping with a lot
> > of OSM activities. Matt keeps some working groups running, Frederik
> > enforces his POV on mailing lists, as usual, and Kate continues to
> > push HOT to greatness. But OpenStreetMap does not change. The reason
> > everybody goes to our Charman Emeritus for comment every time
> > something happens in OSM, is that he is still relevant, even not
> > participating in the community. Because nothing. has. changed.
> >
> > And it starts to bother me.
> >
> > Recently I have been shown the exact way OpenStreetMap is going to die.
> > It's terrifying, actually, but I'm under NDA and cannot say more.
> > The only way we can evade that fate is to evolve. Not like before,
> > but aggressively, fast and visibly. We have 3 to 6 years, in which we
> > either build a massive, large-scale and well-funded structure like
> > Wikimedia Foundation, or we will become but a shadow of better — but
> > no so open, not ours — things to come.
> >
> > I understand every one of you has their day jobs and family and
> > holidays. You cannot make OSM excel when working only an hour a day on
> > it. To take control, we have to create our own day jobs, to be our own
> > family, and to make holidays enjoyable with OpenStreetMap. We either
> > start working on an OSM future now, or the future will be lost.
> > The deadline was faint some years ago, but now it has moved closer.
> > Please evolve.
> >
> >
> > IZ
> >
> >
> > _______________________________________________
> > osmf-talk mailing list
> > osmf-talk at openstreetmap.org
> > https://lists.openstreetmap.org/listinfo/osmf-talk
>
> _______________________________________________
> osmf-talk mailing list
> osmf-talk at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/osmf-talk
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/osmf-talk/attachments/20140926/052c3786/attachment.html>


More information about the osmf-talk mailing list