OGC Seeks Public Comment on Global Entity Reference System (GERS) yes this would entrench the Linux Foundations Overture Maps as the reference dataset for the OGC.
5 Likes
I’m sure this is very important for geospatial thought leaders who are doubtless firing up ChatGPT as we speak to share some sincere, deeply-believed thoughts on LinkedIn.
In the meantime OSM has been merrily ignoring OGC since 2005[1] and I don’t propose we change that any time soon.
I say 2005 because in 2004 I don’t think we were aware OGC even existed. ↩︎
18 Likes
For those who don’t have their ECMA-376-compliant copy of Word handy, here’s an HTML version of the justification document. OGC standards are often named in procurement processes and the like, though fortunately they aren’t the only source of geospatial standards. The comment period is specifically for adoption as a community standard.
And we were just talking about GML the other day…
4 Likes
So are you saying OpenStreetMap could be rendered obsolete???
I wasn’t aware that GML worked only with a specific dataset from GAFAM …
1 Like
Well that is the Linux Foundations goal (not exactly news) and making GERS a standard will provide them with with substantial leverage to force use of their data.
Naturally there’s the other, hilarious, angle that they threw out the initial implementation and redid the whole thing just last year.
It seems weird to me that the GERS will be standard, but the standard doesn’t include any of the data GERS is based on. So I guess it’s a case of “please make everything you do in the future beholden to us”?
1 Like
I don’t have much of an opinion on Overture myself, but I’ll note that the people I know who are excited about Overture are excited about it because of GERS - knowing that a segment will maintain a persistent identifier is useful. So it makes some sense to me that - if they’ve gotten that to a workable spot - it would be a potentially standardizable thing, even if the data itself isn’t standard.
I’ve wondered about this concept a few times but never seen anything definitive for it?
So when the hairdresser in Shop 6 of the shopping centre (mall) moves out to a new shop[ down the road; the tattoo parlour moves from Shop 15 into Shop 6; & a real estate agent moves into Shop 15, what are the “permanent identifiers”?
Does the hairdresser keep her ID for her business, o does it stay with the shop?
Same for the tattoo parlour - do they keep the Shop 15 ID or take over Shop 6’s?
Does the RE agent get a complete new ID as they’ve just opened their first agency, or do they take over 15’s?
8 Likes
I can’t say I know the answer, and the people I know who want GERS are more interested in it from a roads standpoint. They’re smaller shops, so they’re not necessarily going to have their own linear referencing setup to maintain persistent data, but they’d like to be able to attribute information to road segments or refer to it in other ways. Doing it for roads feels more straightforward, though I’m sure oddities like what happens if you add a new intersection to an existing segment must be dealt with somehow. So, that’s still where I’m not sure just how persistent it is or not. But just taking them at their word for a moment, I could see the value in GERS, if they’ve gotten things relatively stable.
I forgot to mention CityGML being one standard in cities, and roads, as I haven’t figured it all out. We might still learn some lessons from their good and bad.
You might want to warn your friends that this is not what they are getting. GERS is nothing but a marketing name for a UID into the Overture database. There is nothing unifying, global or persistent about it. Or rather only to the extend that the Overture database happens to be such.
We have discussed this problem for years now in the context of using OSM IDs as global, unique entity IDs. For good reason we have been warning people to not do that. And those reasons are as valid for GERS as well. You already mentioned the issue that streets might get split. But the issue starts much earlier by asking the simple question: what is a street. OSM splits a street in two ways, when there is a physical barrier between lanes. Sometimes we add two more ways to model the sidewalks. So, what is the right model here? 1, 2 or 4 ways? The answer will always be: it depends. And lets not even get started on data that Overture doesn’t yet have or isn’t willing to maintain (cycleways for example?).
GERS is not usable as a persistent identifier for your own data unless your data has a 1:1 relation with the Overture database. So not only do you still have to maintain your own internal ID system, you now also have to conflate it with GERS if you want to be “industry compatible”. And conflation is a very hard and very costly problem. So, hard and costly that Overture after four years of trying has given up and decided to outsource the problem to everybody else. That’s the beauty of standardization for them.
If you are a data producer, then GERS gives you a very costly lock-in to Overture, while the data consumers organised within Overture will slurp up your data at no cost for themselves and find ways to monetize it.
17 Likes
You might want to warn your friends that this is not what they are
getting. GERS is nothing but a marketing name for a UID into the
Overture database. There is nothing unifying, global or persistent about
it. Or rather only to the extend that the Overture database happens to
be such.
Maybe they could team up with What3Words. Sounds like they are in the
same field of business.
We have discussed this problem for years now in the context of using OSM
IDs as global, unique entity IDs. For good reason we have been warning
people to not do that.
Because we have an engineering mindset and not a marketing one. We know
it cannot be done well so we’d rather not do it. If you’re a business
looking for mind and market share, you don’t have to solve the problem,
you just have to convince enough people that you have solved, or will
solve, it, and paper over the chasms.
18 Likes
Will you be responding along those lines to the call for feedback?
I’m surprised that there are even prospective users that would consider using GERS for streets/roads (if they are not using “Overture data” in the first place) given that you are going to end up using some form of LR in the end anyway with the current exception of if you are using OSM highway data (because there’s a bridge file for that). I would have assumed that the only real use case would be for POIs.
2 Likes
Are you proposing to take some specific action? Either individually, or via OSMF or in some other way?
And I may be a mistaken peasant not capable of predicting future, but is this title quite clickbaity by any chance? And it is actually not going to destroy OpenStreetMap project?
2 Likes
There are multiple efforts in progress to comment on the proposal.
And yes, while the issue is important, naturally the title is clickbaity, how else are you going to get attention given the mass of, lets be very nice, “not really urgent issues”.
2 Likes
so if someone is not already involved in either of them, then it is not useful/needed for them to become involved?
if it would be useful to do something - is there something actionable for random OSM mappers?
I wasn’t aware of GERS and how OSM tried to stay away from it. This clickbaity title made me read about anything that got posted here, and even though it sounds interesting about what Overture will possibly do to make themselves as the wannabe “industry standard” for what they offer, I indeed can’t see what a random OSM mapper can do about it. Very few people in OSM have great interest and even passion with such stuff as most of you in this topic.
Anybody can comment on the proposal, that why it is the very first link in this thread.