Paper about "Evolution of the OSM Data Model"

General talk
1

I found this paper written by Jochen Topf. I found it really nice to read and lays out details and what are possible next steps, without talking bad about anything.

I could not find it mentioned anywhere here in the forum. What do you think after reading it?

1 Like
2

Hi Kevin,
there was a blog article about this topic here: Towards an improved data model for OpenStreetMap | OpenStreetMap Blog
And it was listed in the weeklyOSM episode 631 here: weeklyOSM 631 – weekly – semanario – hebdo – 週刊 – týdeník – Wochennotiz – 주간 – tygodnik
And some more information / links in weeklyOSM episode 620 here: weeklyOSM 620 – weekly – semanario – hebdo – 週刊 – týdeník – Wochennotiz – 주간 – tygodnik

All the best
mcliquid

1 Like
3

Based on the osm blog post, the official discussion takes place on GitHub.

1 Like
4

So I’ve read this study a couple times, and I felt that it doesn’t articulate any meaningful problems with the data model that are a real problem for data consumers. It reads like “oh there’s a couple ways we can tweak a few things and make it a little nicer for developers.” Which is all well and good, but if data consumers can cope with the data model as it is, that’s not a real strong impetus for spending effort on it compared to other initiatives.

I was expecting that this study would identify “If a data consumer wants to do X, they cannot because of this deficiency in the data model, and therefore we should change it in this way.” Instead, it basically confirmed that the current data model works fine. Which to be honest, as a data consumer, I have to agree with.

I agree with the fundamental conclusion stated in the introduction, “Despite all the problems OSM might have, its success shows that it is generally working extremely well. So we want to make it fit for continued growth – not fundamentally change it.”

I also agree that there are some niceties we can add, like having the API able to reject invalid polygons or support objects crossing the antimeridian. But when it comes down to what to spend money on, is this more important than, for example, dealing with our growing vandalism problem?

4 Likes
5

Good that it is well known :slight_smile: I did see the blogpost on the OSM blog, but that was before the study and not the conclusion on the study.

@ZeLonewolf Valid points you bring up.

I think that is were blockchain like Ethereum where there is multiple implementations and are encouraged, this to avoid the stalemate. It is pretty cool that there is multiple editors, but having a single backend where everyone has to agree looks to be impossible.

Then OpenStreetMap can stick to the fundamentals like Way, Node, Relation, but that just keeps OpenStreetMap primitive and value being created elsewhere and not in OSM.

It is hard to believe there will be a competitor, there is simply so much data going into OSM. One way could be to build a new editor which does all the smart things like preventing vandalism, but there is no money in creating that.