SimonPoole has commented on the following diary entries

Post When Comment
Updated contributor numbers 7 months ago

@jaakoh I believe the "200'000" number stems both from the Pascal Neis 1st study (which is simply a coulple of years old) and from last-editors numbers which by the nature of the beast tend to be smaller (currently at ~280'000). On top of that we shouldn't forget that we dropped contributions from roughly 15'000 contributors* during the licence change, most of them very very small, but which distorts the registered user vs last editor numbers even a bit more.

The historical run rate for actual contributors at least since 2007 (when the project started getting off the ground proper) has always been at roughly 30% of the registered accounts, the only period this was different seems to have been in 2012.

  • some of these have since come back and have accepted the CTs
Attribution and all that (a rant) 7 months ago

@Tordanik historically we have given a lot of leeway in what we consider appropriate attribution unluckily that leeway has been misused with an increasing intensity. Simply giving in to the misuse is not likely going to be an option with popular support and as I point out in the blog post, we are not asking more than what is the industry standard.

And using common sense is not prohibited, so providing better, more legible and visible placement is always ok.

Attribution and all that (a rant) 7 months ago

@SunColbalt you are seriously misrepresenting the discussion in question, which didn't even remotely concern attribution, the person in question was asking for us to vet his business model wrt the share-alike conditions in the ODbL.

Attribution and all that (a rant) 7 months ago


Attribution is not just important for legal reason, it provides postive feedback to our contributors that are interested in seeing that their work is being used and further it is the only way we can assure that users of the end product know from were the data is coming. See the recent threads on hackernews and reddit etc. were there are still a large number of people that have never heard of OpenStreetMap. Google knows very well why it insists on attribution.

Further while the poor developer left alone without any support might make a nice sympathetic pricture, it is -very- far from the truth. The worst offenders are not small mom and pop operators and are fully aware of what they are doing.

Attribution and all that (a rant) 7 months ago

@ThomasZander the LWG has not given a wholesale carte blanche to mobile applications because we've seen people move the attribution off screen based on space arguments when it was clearly not necessary. The current suggestions for appropriate attribution allow enough leeway when it is really (emphasis is on that) difficult to provide a clickable link/button or whatever.

What is the OpenStreetMap convention? Do we tag addresses on buildings or on separate nodes? 9 months ago

The stats seem to be quite compatible with the numbers I generate now and then here

While I typically tend to add the address to the building outline if there is no further information available, there is no doubt in my mind that the best solution is to add it to an entrance node on the building outline (forgetting about indoor mapping and multiple addresses in the same building for now).

Adding the address information to the building itself can and does lead to routing artifacts (being routed to the nearest street to the building and not to the nearest to the entrance) that can be avoided if the entrances are mapped and hold the address info.

Vandalism in Sydney ? 10 months ago

Youo need to report this on the talk-au mailing list, it is rather unlikely that posting it here will have any effect.

PushPin als App für alle 10 months ago

Für "crowd sourced" Übersetzungen gibt es schon 2 Platforme: translatewiki (wird von der zentralen OSM infrastruktur gebraucht) und transifex (wird von iD, vsepucci und anderem verwendet).

Some more Android hackery for efficent mobile surveys 11 months ago

The main issue is that for the most mainstream mobile device category (Android with roughly 80% global maket share) none, well at least none that I know of, of the surveying apps add the information. If I can get the code I added to osmtracker to work stable enough for general use I will make it available.

Some more Android hackery for efficent mobile surveys 11 months ago

Well in reality these typically do measure the mangetic field on three axis, and are a bit more complicated than a conventional compass. But the net effect is that you can determine which direction you were pointing the phone camera. There are a couple of "ifs" like for example that you should naturally calibrate the sensors before you set off on a mapping tour and other things you may have to take in to account to get reasonable results, but in principle it works.

Stadt Vechta und Landkreis 12 months ago

Die Fehler in Vechta sind mit hoher Wahrscheinlickeit von der Lizenzumstellung im letzen Jahr.

Ich hoffe du bist fleissig dabei die Schäden zu beheben :-).

More Addresses about 1 year ago

The US is interesting because it shows that obviously (at least currently) imports are slower than simply collecting the addresses. But I suspect that large scale imports in the US are unavoidable.

More Addresses about 1 year ago

Naturally I did the numbers too, but ~2 million addresses a year implies that we just started collecting them two years ago :-).

So yes, the current rate is too low for the Web 2.0 generation, but in all likelihood it is still increasing and we will just have to see where it levels out to the typical OSM linear growth mode.

Using OpenStreetMap on a daily basis about 1 year ago

There is ongoing work on your questions 2, 3 and 5.

As to question 1) searching for Café, Montreal will give you the desired results, and nominatim does AFAIK take the current viewbox in to account, however it is primarily an address search and not a POI locator.

A more general remark: while we are not very clear about it, is mainly about collecting and editing our data and not about providing services to end-users. On the other hand it is naturally the primary place our contributors and other end users go for such services.

Service development on the main site, adding a POI display and routing, is for the above reasons mainly thought as adding further tools for mappers and not as a google maps look-a-like. Deplyoment of such additional serives is not just dependent on more funds for hardware, we need volunteers to actually look after whatever we choose to add to the main site.

Updated "no address" Layer on over 1 year ago

IMHO it is actually not such a common practice (legally naturally countries differ greatly in what actually has a number, the lot, the building or an entrance) and doesn't seem to be documented (not that that means very much).

In any case it would be rather expensive to change the queries so that such a node would get included simply by expanding the building polygon or similar (not to mention that it would generate tons of false negatives). It might be possible to support it if the node is actually on a fence polygon enclosing the house, which however doesn't seem to be the case in your example.

noaddress layer on over 1 year ago

Hi DaCor

You are correct it is a new feature that hasn't been "announced" yet. There will be a further "has_address" layer too. Right now, i'm pre-rendering some of the tiles and am working on getting a coupel of technical issues resolved, as soon as that has happened I'll write a blog post.


OSM History Viewer working again over 1 year ago


PS: could you change the data licence to ODbL?

Adding Addresses over 1 year ago

What works reasonably well if you have building outlines already available in the OSM data (or have added them yourself before going out on a survey is to use vespucci to add house numbers directly to the buildings. I would suspect that this works just as well with the iOS based mobile editor.

1 million is a big number over 1 year ago


we actually have something over 300'000 contributors that have made edits (the numbers on alltogetherlost only count the last editors of objects). As to the 1'000'000 accounts, I believe reporting the number is quite valid (well at least in marketing terms). Particularly since we know that we have a substantial number of contributors that register an account, go away and come back a long time later and start editing.

I am myself such a case with a good year between creating the account and my first edit.

noaddress layer on over 1 year ago


First question: not easy to answer and I don't have a strong opionion on if it makes sense to duplicate the information.

Wrt site relations: currently that would seem to be very expensive to implement, I will have a look at it asap.