joost schouppe has commented on the following diary entries

Post When Comment
Open road data for map improvement in Flanders, Belgium 15 days ago

Hi Erick, the recent layer is weird. I tried it in iD and it worked, but direct tile access (e.g. in a leaflet map or manipulating the url you get from the images in ) does not seem to work.

I believe there is some sort of access restriction in place. Maybe they don't want us to use it, I don't know.

Nothing personal, just GPS tracks 16 days ago

@marc but you can! For Strava, there's this map. It looks like simply adding ?year=2015 should give you filtered data, though I can't seem to be able to get it to work.

"Welcome-to-new-mappers" program in the Netherlands comes to an end. 26 days ago

Some thoughts from a fellow welcomer in Belgium.

  • No first changeset should be left unchecked. In Belgium, we have someone checking the clearly wrong changes, but often things are just "slightly wrong" or "could be better". Pointing out how the new mapper could expand on something that was interesting enough for them to do a first edit might have more impact than just welcoming them

  • Regional messages upon joining should definitely be re-introduced. It really is a burden to send these messages manually (I don't manage to keep it up myself). It isn't so much about trying to teach people stuff, rather an opportunity to include them in the community. For many new people, OSM is a database (or worse, a map). It isn't. It's a community.

  • There is little point in improving user retention with this message. We only need a few dozen addicted mappers more for a lot of improvement. I think a welcome message might help there, but it will never be possible to prove this statistically as the numbers are so small. At the other end of the spectrum, I do believe we can expand the many slightly-interested people. is an excellent tool for that, but there are some real issues I hope get fixed soon. On the one hand, some measures are necessary to reduce nonsense edits (better explaining of what they are doing, moving more stuff to Notes in stead of POIs, etc). On the other hand, I'm convinced the majority does not get messages sent through the OSM system. Considering the volume of edits, I think that warrants integrating our messaging system into the app.

The map is a fractal 30 days ago

Yes, yes, keep up the pseudophilosophy!

Exactly this simplification of reality is what makes a map useful. So maybe there's a point where adding more detail to the map decreases its usability. We're not quite there yet I think :)

Also, I do like your fractal analogy (even if it's technically wrong). What is missing in it, is that the fractalisation is asymmetrical: in some places the detail level keeps going deeper, in other places it's still a rough outline.

The Challenge of Creating the Big Map of Sibiu 30 days ago

You could consider using Mapillary when out collecting data. That way you generate a cc streetview of you city, and others without local knowledge might be able to help you a bit more. Also, it allows you to collect data (often by chance) that you will only be interested in later. So you avoid having to go back to that same street because you didn't collect the fire hydrants.

That's in theory. In practice, you need a very good and focused set of pictures to be able to really do that.

Deriving centerlines from riverbanks without. 30 days ago

I recently tried the inverse of this same problem: trying to measure the width of a polygon. Problem is that it only works decently if your polygons are long (as opposed to square). Same would apply here. If for some reason, the mapper has cut the riverbank up into short and wide bits, then I would guess your process breaks down.

Mapping with Strava about 1 month ago

Thanks for this elaboration, it does make your point much clearer. I especially like your point of not adding detail you're not sure about. It's something we often say in statistics: "it's better to be about right than exactly wrong".

Mapping with Strava about 1 month ago

Hi Bushman, I don't really understand the problem you're trying to point out. I believe you're trying to say that GPS have systematic error, not random error systematic error, not random error That does have an impact in some cases, but mostly in deep valleys or urban areas, where reflections confuse units. In forests, I would say error tends to be random. Except of course for the tendency to adjust the measurements to expected outcomes. But generally speaking, many tracks are better than just one. In the use case I'm describing, that is the only choice you have, as they are not visible on sat pics.

Mapping every turn restriction in Toronto about 2 months ago

Interesting stuff anyway. A toolchain is getting born :) I was just made aware of this project, that creates a simple report about the comparison between OSM and reference road data. We couldn't get it to work, though.

Mapping every turn restriction in Toronto about 2 months ago

This is extremely interesting!

We have a lot of open geodata these days, and a tool like this might make it so much easier to conflate those with OSM. I've been experimenting with this recently, for example taking an open data feed and a live Overpass query to compare mapping quality and completeness (case: dog parks).

The list of "invalid" items is something which might be quite valuable to the official data sources. In Flanders (the northern half if Belgium), we now have very extensive road and path open data. It is more complete than OSM, but it is also often outdated. So a matching is going to be a lot of work. Do you think the data-mapper could be adapted to something as big as this? See this thread for info and a comparison map.

Does Open Street Map have an input API? or Upload feature? 2 months ago

Many new contributors like you have wanted to jump in by importing. The most important thing to realize is that most OSM data is handcrafted by volunteers who do this because they like doing it. Most external data is made by people payed to do it. You will find that OSM data might be sparse and incomplete, but the quality of what is there is phenomenal. Most of the time, external data will be either of bad quality or evaluated to really different quality measures. In practice, that means you will either need a complicated technical solution, or will just be able to use external data as a starting point. Either way, it is a bit more complicated than feeding an API.

OSM is not a collection of geographic features but an integrated database of things with a spatial component. For well defined things, it might be more efficient to look elsewhere. For example, for address databases, there's this project. is a new evil (instead of Potlatch)? 2 months ago

And the same goes for iD. The first time you hit that Edit button on, you know nothing at all about the project. Hell, three years after that, I still feel like I know next to nothing. We even have a whole team needed to keep us up to date of only just the most important stuff!

That said, of course we need to make the tools as good as possible. The editor is obviously a very good idea, even if it has many faults at this time. Let's not waste time complaining and demotivating the makers of this great tool. Let's just give them constructive feedback on how they can make it better.

Why I am supporting Brussels 2 months ago

It will be a pleasure to welcome you :)

Oh, and location-wise, here's the building where most activities will take place.

Entrevista a la comunidad OSM en PerĂº 3 months ago

Gracias, compartiendo con @osm_be

Mapping My Hometown 3 months ago

By Indian standards a small town indeed, but I believe it should be tagged as something larger than a hamlet :) Check this site for some guidance on the appropriate tag.

If you want to work on it a bit more, it's always strange to see so many houses without streets to get to them.

Anyway thanks for contributing, keep it up!

National Missing Maps mapathon 4 months ago

201 mappers completed a square. I'm not kidding.

We completed 1641 squares in total.

Biggest completer of squares was tastrax, who only worked on Kivu. Then fpriem, mitruc and lbataille all marked over 30 squares as finished. tastrax was actually helping from somewhere deep in the south of Tasmania if I'm correct. The other three are all new mappers!

National Missing Maps mapathon 4 months ago

I had a look at Swaziland. We did ANOTHER 25.000 buildings!

The mappers were also looking for more things to do; they improved about 800 roads too.

National Missing Maps mapathon 4 months ago

Registrations kept coming in as we got on national radio in the morning. It got to 230.

I'll try and make more statistics later, but yes, on Idjwi we mapped over 23.000 buildings. I suspect a slightly smaller amount in Swaziland.

Not only did we make the national public television (VRT) , also the largest national commercial channel VTM. Both of the Flemish region, looks like the French speaking media didn't pick up on it in the same way.

African Roads and a Western Bias in Mapping 5 months ago

Only just taking into account the size of the connected place is not enough. I believe that when there are two roads connecting two places, and one is the preferred route, that one should have the highest classification. Road classification should help pick routes.

Of course, all extra effort on tagging roads with things like surface and or tracktype are welcome. To me, the western bias in OSM is quite clear in the fact that no main style uses this info to show which main roads are better or worse. The HOT style does an effort, but only at high zoom levels. Check my user diary for an example of what I'd like to see.

A MapRoulette Update: V2, Pedestrian Safety, GeoJSON, and Floating Ways in China 5 months ago

I saw quite a few of these floating roads in Peru and Bolivia too. The problem is the high res pic suddenly ends (or gets cloudy) and it is sometimes hard to even approximate its course. These days they are often fixable with Mapbox Sat - or an educated guess.