If we can’t offer a serious quality product, we never leave the margin. OSM, nice background map for freetime and the pennyless.
Pardon me but that sentence is silly on several levels. Firstly, OSM data (and sometimes even OSM provided tiles) are already the preferred map the world over by many governments, nonprofits, and businesses alike. So apparently we are offering a serious-enough quality product, as these people are neither hobbyists nor pennyless.
Secondly, even if OSM were only used by hobbyists and poor people, would that somehow be bad? Because things that help poor people achieve something are “ugh”, or what?
Thirdly, if you say we are “on the margin” then why would we want to leave the margin? I have a feeling you might naively apply standard business thinking to a non-profit project. This is a dangerous approach - just look at all the companies in the OSM space that have entered the scene with big fanfare, burnt a ton of investor money, and then pivoted away from OSM or just closed shop altogether, all because things couldn’t go fast enough for them. You want to turn OSM into something like that?
(edit: formatting)
9 Likes
True, but most are just using OSM as a nicely detailed and free optional background, on which they overlay their own data. Cyber attacks like the current case can quickly shift this preference.
For reliable use of OSM data for applications and functionality that people depend on, I would not recommend OSM, unless the organisation itself does the QA in OSM for what they need, and I would recommend to control refreshment of their map/data, as a safeguard against OSM’s vulnerabilities.
Of course, in other parts of the world than I know personally, this might all be very different.
Which provider would you choose for bicycle routing, that is not based on OSM data? 
1 Like
More relevant to this issue: Which percentage of cyclists would choose a provider who routes based on OSM data?
As for me: I mainly use waymarked long distance routes, which require no routing at all, because the routes are fixed.
For on the fly routing from a to b I myself use OsmAnd, but my wife and all other people I know use Google Maps for all transport modes.
For Node Network trip planning I myself use Knooppuntnet Planner, but most people use various apps, most of which use OSM only as a background and non-OSM sources for the Node Network data.
In Nederland, cycling maps and routes had become a stronghold for OSM, mainly because big G sucked at it. But they have catched up, and it’s free. So OSM is back in its niche.
PS I think OSM is a great project, great potential, but I am also a realist. Its inherent strength is also its weakness.
I absolutely would recommend OSM data - but I might be tempted to wait a day or so for people in OSM to spot problems and not update my tiles based on OSM data if there is a known outstanding issue with OSM data.
I would not recommend OSM’s current tiles, because we know from the most recent case that people were still reporting corrupted tiles many days after an initial incident that was resolved in the data in around 15 minutes.
I’d estimate that the majority of DWG reports have been from people who are either third-party consumers of OSM tiles from the CDN, or are customers of those third parties. The next biggest group is from people using osm.org directly. Smaller than both those is the number of people spotting OSM data problems in downstream systems.
As an aside, a fair number of these seem to be using “a.tile.openstreetmap.org” or similar, suggesting that they haven’t got that particular memo.
Edit: For more info, the DWG has just had a report come in about problem data in tiles just 5 minutes ago.
4 Likes
Here and now, with texts in 2D and animated cars along the vandals ways in 3D at F4Map and they render any change within 24 hours. Ctrl+F5 has no effect…cache cleaned daily too.
Can you link to an object in OSM that appears incorrectly in whatever map it is that you have shown a picture of? Because if it was fixed more than 24 hours ago the statement “and they render any change within 24 hours” must be incorrect. However, with only a picture to look at, it is impossible to comment on anything specific.
2 Likes
Yes, their 2D basemap rendering is still showing the vandalism in some areas/zoom levels (I won’t link it here), the update information is probably about the 3D overlay or maybe it is an overloaded dirty queue for the 2D basemap rendering. I could also show you a third party layer that is still displaying the vandalism in Tel Aviv and Petah Tikva as it only updates once or twice a year and updated into an already outdated vandalized dataset back in late December.
But most third parties that do tile rendering or other OSM data based systems or services (Geocoding, Routing, etc.) take a lot of work and effort into not displaying those vandalism acts via their systems.
It’s mostly the tile.osm.org OSMF tilelayer (ab)users* that complain about their customers complaining while they don’t understand that the OSMF tilelayer is for visual mapper feedback and by that will display vandalism if it is in the data (also to catch that vandalism).
*) I’m not talking about projects like OpenRailwayMap or Brouter-Web etc. as those are heavily involved in the OSM community and know what is going on.
For on the fly routing from a to b I myself use OsmAnd, but my wife and
all other people I know use Google Maps for all transport modes.
For automobile traffic, Google has the huge advantage of ingesting not
only current congestion situations, but also past average speeds. This
is not something we are likely to be able to match, ever, because we
don’t collect telemetrics.
PS I think OSM is a great project, great potential, but I am also a
realist. Its inherent strength is also its weakness.
Yes, OSM is “not reliable by design” because anything could be broken at
any time. But it seems to be good enough for an astounding breadth of
use cases.
7 Likes
Here, no objects to grab. Untick the 3D box right top to see the flat view with texts such as Springfield Close.
Yes, it’s definitely not a new issue and can already cause problems here and now. The server starting to block uploads has the potential to make it much worse unfortunately.
Imagine someone uploading a whole bunch of data where one of the objects gets flagged in some way and is rejected. It doesn’t even have to be intentional vandalism, it could just be a benign upload with a false positive match. With all the nodes getting uploaded first, then the ways and finally the relations as per current editor behavior, chances are relatively high that the first few uploads containing only nodes goes through and only one of the later uploads will trigger the upload filter. That leaves a ton of stranded nodes floating around that have already hit the database.
Potential courses of action are either:
- Transactional changesets over multiple uploads (this would be an actual solution, but we all know that’s not going to happen)
- No limit on number of changes in upload (bad idea and brings a whole other boatload of issues with it)
- Raising the object limit per upload (doesn’t really solve the problem, just makes it possibly a bit less likely to occur depending on how much you raise the limit)
- Changing the upload order in editors (also doesn’t completely solve the problem but mitigates it, at least the partially uploaded data would make some sense but can still be broken in some ways)
- Putting a limit on the number of changes in the editors and blocking upload if it would have to be split, forcing users to break up the changes manually and upload them in smaller chunks (sucks for users)
- Automatically running a revert on the already uploaded changes after a part of it gets blocked (could be a potential solution but still leaves junk in the database)
There is no perfect solution here unfortunately
Amen.
I never hear anybody say:
"We should abandon the horrible public transport and raze all railway tracks because trains have delays all the time!" (which isn’t even close to being remotely true in most cases and most countries where people utter such words), so why do we if OSM doesn’t achieve “100% service level”?
Maybe its time to focus on the 99,9% of the time that we do have a great map good enough for most use cases!
2 Likes
I think so too, but is it the map of choice for most people/apps/organisationsoffering the end user functionality for those use cases? Then why do I mostly see only OSM backgrounds behind other software that handles the use cases?
Looking from the inside you see a wonderful universe with many shiny objects. One that looks from the outside, not so much.
1 Like
OSM is not a map but data and this data fuels about every application,map or use case out there that does not use e.g. government data or GM data (or Here data). TomTom practically switched back to OSM data as one of the data foundations of their products.
OSM is also a brilliant citizen project, and therefore loved. So foremost it is a community, second gis data democratized - and a map is just one visual kind of representation of the data. The OSM ecosystem has tens of thousands of different use cases applied through (probably tens of) thousands of different apps, software systems, maps, etc. pp.
4 Likes
Actually, OSM-Carto is not a good background map style. Perhaps I am spoiled, here in Austria the administration provides a service “basemap” CC-BY that is a much better background a.k.a. basemap.
Imagine the municipal utilities used OSM-tiles as a backdrop for their oBike rental stations. They show in OSM Carto. They must update OSM too, when they change locations. That happens, more than mappers seem to be able to handle timely.
Openstreetmap though certainly is good to push administrative efforts by putting the bar high.
Unfortunately not everybody is concerned with advancing the quality of openly available geo-data. That is what this topic is about.
2 Likes
OSM is one of their sources of road data, yes. But I wouldn’t say TomTom is an OSM application.
Given the regular Map Roulette announcements by them seeking to fix roads and now a series of ‘spiky buildings’, they certainly have made a choice of what data types are OSM based. Frequently my cycle & car nav progs ask me if I want to stay on free or switch to TomTom map base, for money, so they sure have a vested interest in helping and pushing ‘to right’ the mapped data here, which is fine because all OSM data consumers and map users benefit from that.
1 Like
This (and the similar observation by @Woazboat that you are responding to) are not correct, it is trivially possible to create world-crisscrossing ways without any of those operations (and btw have the change count as one edit).
Your statement should actually be the other way around: for validation purposes the bounding box calculation should include all current and previous positions of the parent object, way or relation, and for the changeset the union of all the individual object boxes.
For -non- validation purposes (aka finding past changes) the last point should be done differently, but that is not the topic here.
2 Likes
Thank you, acknowledged. I figured out some counterexamples myself.