SimonPoole has commented on the following diary entries

Post When Comment
OpenStreetMap Community Statistics Revisited 4 days ago

@joost the "rawer" data would be a rather largish database dump, which I can naturally provide, I'm just not so sure that it would be useful. I can naturally run specific queries if a good case can be made for them.

Given that the published souce code is currently just a readme file the 1kB zips are not that surprising (the source code is totally uninteresting throwaway SQL with some perl around it) :-)..

OpenStreetMap Community Statistics Revisited 6 days ago

@Imre that is a bit of an insider joke, those are simply the changsets that couldn't be geocoded because the center wasn't inside a country boundary (see the original blog post), they can simply be ignored.

OpenStreetMap Community Statistics Revisited 6 days ago

To put the "205" in a bit more context, a comparison over the same time period with iD (~78'000 new users using iD over the period from April to December 2016):

Editor      >10     >100   >1000
iD          63%     35%      10%     11%      1%      0.2%

I don't think this is negative as such, it just shows that users starting off with are unlikely to be a source of lager mappers that in the end keep OSM alive.

OpenStreetMap Community Statistics Revisited 7 days ago

@BushmanK of the ~89'000 contributors that started off with ~2'200 tried a different editor at least once. In other words a bit over 2%.

The other interesting number is that just 205 of those 89'000 made more than 1000 changes in 2016. Note "changes" not "changesets", 1000 changes is really small, typically a changeset or two with a conventional editor.

OpenStreetMap Community Statistics Revisited 7 days ago

@pitscheplatsch IMHO the most interesting question currently is if we will see any uptick in longer term mappers due to the large numbers of people starting off with, or if the underlying trend continues as is

Active contributors per year

2014: 154'994
2015: 162'059
2016: 171'929 + 89'131 ( 
OpenStreetMap Community Statistics Revisited 7 days ago

Caribbean and Central America is lumped in to South America IIRC

3 years of welcome messages, more than 3400 of them about 1 month ago

@mmd rarely, however I do superficially check contributions and if necessary follow up with a 2nd message or changeset comment, that will typically get some more feedback.

3 years of welcome messages, more than 3400 of them about 1 month ago

@Vincent de Phily the difficulty is naturally that, on a larger scale, we don't actually know who got a welcome message or not (only who we sent one).

The more general problem with measuring the success (or absence of such) is on the one hand that the numbers are very small: we would expect perhaps 10% of contributors to edit more than once and an even smaller number to become regulars. Halving the numbers to do A/B testing (I wouldn't want to stop for a while because sign ups very often depend on external factors which we can't control) is not going to help with that. And on the other hand that effects may be hard to measure, contribution quality particularly. We do know, as already said that there hasn't been any drastic changes, wit: which would appear to be stable (contrary for example to Germany where the number of daily editors has been going down for a while).

Interesting enough one thing has subjectively changed over the last 2-3 years: we have a lot less broken municipality boundaries than we used to have (it used to be a at least once per week, and now that is extremely rare), but that could just as well be due to better tools.

Totalrevision der Verordnung über das eidgenössische Gebäude- und Wohnungsregister (VGWR) about 1 month ago

Well "person in charge" is me, and there hasn't been any changes as of late. Originally we were planning on documenting in OA format, but given the potential changes (as mentioned) it doesn't seem to make sense until we have clarity wrt the GWR.

Totalrevision der Verordnung über das eidgenössische Gebäude- und Wohnungsregister (VGWR) about 1 month ago

@MapMakinMeyers nope, essentially there is some cantonal and city data, but that is it at this point in time (OSM is currently likely the best open dataset), see

Mapping my home town : Gobichettipalayam about 2 months ago

Great! You show the value of somebody surveying and mapping on the ground that knows the place.

BTW I noticed that this road nearby has a rather weird "ref" value, that should probably be replaced with the proper value or removed.

My Area Needed a Lot of Work about 2 months ago

N-S vs E-W is not that uncommon see for example and actually quite easy to explain: the TIGER surveyors at the time drew a paper map and couldn't remember which side was up..... (it should be pointed out that there are places where TIGER actually did have something to do with reality).

My Area Needed a Lot of Work about 2 months ago

Ah yes TIGER as we all love it. DesertTrip +1 for fixing.

My Area Needed a Lot of Work about 2 months ago

DesertTrip care to give us a pointer to the area in question (what you are describing sounds like original TIGER data though :-/)?

Indoors at Wherecamp Berlin 2016 2 months ago

Yes obviously the antenna seriously limits what you can do with a current smart phone, dual frequency reception however will still give you significant gains compared to single frequency (which is why it really would be the more interesting development).

Naturally I'm aware of RTKLib, vespucci has been able to receive data from an on device RTKLib instance since ages and I've used it myself with an external GPS device (USB) and a proper antenna.

Mapping Baltimore Buildings in 3D 2 months ago

I'm slightly surprised that you missed out on F4 and the original first 3d OSM based map both which predate the offerings from MapBox and Mapzen by years.

Totalrevision der Verordnung über das eidgenössische Gebäude- und Wohnungsregister (VGWR) 2 months ago

Wir haben gute Beziehungen zu und im Prinzip dadurch auch Kontakte in Bern.

In diesem Fall ist das Parlament aber nicht involviert (da eine Verordnung und nicht ein Gesetz) und die Zeit vermutlich zu knapp um irgendwas via Politik noch zu bewirken.

Scaling multilingual name tags with Wikidata 2 months ago

The tiny weeny issue with this is naturally that there is the underlying assumption that wikidata is correct and that the data meets our quality criteria (as in actually being in use and not invented).

Updated Contributor Statistics 3 months ago

To use your analogy from your last paragraph, our main "income" is the time and work our contributors donate. So yes, how much is contributed in that sense is clearly an important KPI and not something that should be glossed over. If we only had contributors using to edit, well then we wouldn't actually have a map.

I don't believe anybody is claiming that the edit count is a perfect metric, but it is clearly a lot better than counting changesets which was never particularly good and has now become completely useless. Separating out POI edits would be rather difficult and would require detailed analysis of every edit, and naturally agreement on what a POI actually is.

I haven't seen or heard of anybody taking the position that a user can't be part of the OSM community, but I would expect such a user to want to be part of OSM and not simply be declared as such for marketing reasons. In turn that requires actually knowing what OSM is and the lack of that is what has been widely criticized.

Updated Contributor Statistics 3 months ago

The edits corresponds to the num_changes attribute in the changeset dump, which is roughly the number of objects deleted/modified/created. Creating a 4 node building from scratch should end up with a count of 5.

But that doesn't really matter, because even if you were to "normalize" the number by multiplying by 5 (given the popularity of doodling buildings) the total contribution share would still remain under 1%/month.

As to the edit numbers levelling out (which likely they are not, I suspect in reality it is simply not so rapid growth, all a question of how you scale the diagram :-)): in lots of places where OSM is really good we are clearly seeing saturation effects in that the low hanging fruit has been mapped and adding meta data of all kinds tends to be a slow process.

With 60-70'000'000 objects touched in one way or the other each month I don't really believe that even stagnation would be an issue (one day we will have something near complete coverage of all major features and then undoubtedly the number will start falling, would that be a problem?). .