[Osmf-talk] The curse of Good Enough

Ilya Zverev ilya at zverev.info
Sat Oct 18 09:02:07 UTC 2014

Hi. Sorry to bother you with another post on OSM stagnation. This one 
will be the last, I promise.

I "remember" good times of OpenStreetMap, before 2010. In quotes, 
because I was not there. OSM was basically run by technical people, not 
afraid to break things, to set rules, to advance. Most 
mappers-programmers were adjusting the project to themselves: so 
uploading data is easier, structures are easy to understand and work 
with, mapping is straightforward. Steve Coast was still with us, and, 
feeling this was his own project, he constantly improved it. At some 
point software and tech stack were deemed good enough, and OSM has turned.

Now the OSM is facing the users. There is amazing progress with vector 
tiles. There are lots of geocoding engines, implementing virtually any 
searching mechanism. Tile stack can be built on any system, using 
multitude of tools, following many tutorials in many languages. We have 
two popular routing engines and 8-10 other online services. New 
programmers build impressive sites using Leaflet and OpenLayers. 
Extracts are plenty: regions, cities, administrative borders, 
coastlines. License terms are clearer then ever, and multiple 
conferences are held, at which hundreds of attendees learn how our data 
can be used in new and exciting ways.

Of course, novices are not forgotten. We have iD, which is fascinating 
in comparison with Google's/Yandex's public map editors. We have a 
tutorial and a constant stream of tasks for beginners from HOT. And a 
wiki which explains everything. It's almost like OSM is a grown-up 
project. Except it's not.

Why, while OpenStreetMap is used by major businesses, it feels like a 
constantly broken and geek-only mapping sandbox? Is it me, or OSM has 
not really grown in past 3-4 years? Wikipedia now feels like a huge 
organization, why OSM is still a dozen of active people, a hundred of 
programmers and some thousands mappers who spend their time fixing data, 
rather then updating it? You can argue and point to rare articles on how 
one person travelled in mountains and collected some GPS traces, but 
such active mappers do not seem to be common in OSM, but more of an 
exception. There are hundreds of thousands new members registering, 
editors are shiny, and users are aplenty, but the project itself is not 
advancing. We still have old API, editors to struggle with, tools from 
00's with no better alternatives.

That's because OpenStreetMap is not for experienced mappers anymore. Of 
course, many will argue: it's a volunteer project, you can do what you 
can: improve tools, go outside and map, submit pull-requests. 
Participate in working groups and make publicity / software / community 
better. Yay.

That was a great thing some years ago, when the project was small. Like 
in computing, you could write a great renderer or editor or other tool 
single-handedly in a spare time. Now, OSM has grown. The planet file is 
26G in PBF. Just to process it, you need to rent a server for hundred 
dollars/month. To set up a tile server, you would need to know Linux, 
OSM tile stack, spend a lot of money monthly, and until recently there 
was no way to have minutely updated part of a planet: you'd get 600 MB 
of updates daily whether you render the whole planet or just a small 
town. But that's a user problem, and such problems are easily solved by 
multiple businesses facing users, like MapBox and Geofabrik.

But experienced mappers, they have nobody to help. Especially if they 
don't know programming. Every step of their way is painful, "as if they 
trod on sharp knives." We managed to make adding new data easy. But 
moving objects is hard. Fixing data offset is hard. Finding deleted 
objects is hard. Looking for changes around is hard. Undoing changes is 
hard. Printing walking papers is hard. Collecting data without pen and 
paper is hard. Aquiring data in a sane format is hard. Validating 
relations — even editing relations is still hard. Discussing issues in 
mailing lists is hard. It is hard right now for me, because I know that 
Board members and programmers will be here in hours and shun me, because 
OSM "works for them", and if I don't like something, it would be easy 
for me to learn a dozen of new languages and libraries and submit pull 
requests to all projects and libraries and processes I find non-perfect. 
API is there, documentation is there, you know what to do and how to do 
it — quick, get programming, don't bother people.

No, I cannot do that. Mostly because I don't have time to fix 
everything, and as years pass, I find more and more broken things in 
OSM. Like that diversity and priviledge issues discussed here recently. 
There are women and autistic people unknowingly oppressed by us, I see 
that and I started seeing that in other communities, and it overwhelmed 
me for a while. But the priviledge issue doesn't end there. People 
reading this are mostly white, male, living in Western Europe or North 
America, on their own laptops. People on the Board and in working groups 
have stable jobs, or businesses, and a lot of spare time to dedicate to 
OpenStreetMap. Well, not everyone, as Frederik's updated manifesto 
reveals. But you know how to run a business, you know that "fast" rarely 
means "good", that you should do only things you can budget. And having 
no programmers means the Board and working groups can only write texts 
(thanks for AoA, LC agreement, community guidelines and other stuff 
which is undoubtedly important) and ban people.

Still, in such technology-heavy community like ours, evolution is not 
possible without programmers. Who now have three options:

1. Write small tools in their spare time, which were written a dozen 
times before.
2. Join an OSM company and write tools for users: navigation software, 
rendering stacks, routing, geocoding.
3. Find a grant and work on important stuff alone.

Nobody takes the third route for some reason. MapBox was our lucky 
chance, and we've got iD editor out of it. But once in 5 years is not 
going to advance OSM. And you would not seriuosly state that first two 
options cover everything we need.

We need specialized tools for experienced OSM users — not explicitly 
mappers — to maintain, to update and to gather data. No sane business 
will create those for us. No mythical "MapCorp" will spare programmers 
to improve JOSM UI. And a lot of tasks need months of programmers 
working full days. Not any programmers — or else GSoC events would 
produce useful and great tools every year, except we've yet to see any. 
Great programmers don't work in OSM, they leave to their own businesses 
or to companies like MapBox. We are losing people, who can improve OSM, 
every week. With our "stability", meaning there is nothing to discuss in 
OSMF Board meetings, nothing to improve except adding and updating 
servers. You can propose in wiki or in mailing lists all you want, but 
our motto is "who wants it, creates it". That worked 4-5 years ago, but 
now OSM is just too big, and tasks have grown accordingly. No single 
programmer, especially volunteer programmer, could give us a new API. 
How do you see a process which ends with updating it?

The new OWL was left by its programmer, in pursuit of a stable job. 
Pull-requests for routing, click-to-see-data (with overpass) and 
changeset discussion are getting older and less relevant on OSM's 
github, while everybody have seen them work and excel at their task. For 
me, an experienced mapper who likes to go outsite, there has been no 
exciting new tools since probably 2011. "Field papers" is just a 
redesign of older walking papers site. Reverting changesets is done with 
an ancient JOSM plugin, and conflict resolution is as hard as it was ten 
years ago. That's stability: we don't evolve, but we are not turning 
worse. And new users provide us with new data every day, to power our 
raising graphs. Disillusioned board members are freeing their seats for 
a couple of fresh people, who will be stripped of their hopes after a 
year of stability and reluctance to change anything.

Please enlist to OSMF Board. Be a candidate. Be elected and see what you 
can do to pull us out of this loop, to improve OSMF as an organization. 
If your job, skills and resources allow that, of course.



More information about the osmf-talk mailing list