On OSMF Contributor Terms and ODbL interaction

General talk
,
1

I’m quite unclear what you are trying to say here. The copyright holders with ultimate power to make any copyright-based decisions are each mapper individually, yes? OSMF is not a copyright holder of OSM data, yes? It is only that I (and you etc.) as individual copyright holders have allowed OSMF (via CT) to distribute and sublicense our work under ODbL or similar license, would you agree?

IOW I can equally license my data not just to OSM, but also to say fosm.org to publish under CC-BY-SA 2.0, or to Google to publish under their proprietary license or whatever, should I choose to do so[1].

They are the licensor for the majority of the data

That they are, I agree here at least. :sweat_smile:

I think you’re either confused or being very ambiguous/misleading (likely unintentionally) here.

To help clarify, could you cite exactly what you mean by “relevant clauses”, and also clarify when you say “the data” do you mean “original (pre-existing) data licensed under ODbL[2]”, or “potential new data (i.e. starting with an empty set)” in that sentence?

If CT didn’t have “relevant clauses” allowing OSMF to relicense the data differently (i.e. I’ll assume we’re talking about sections 2 and 3 of CT[3]), they absolutely would not have ANY right to start publishing existing OSM data under some other[4] license. I can’t fathom you’d disagree here; (but if you do, on what basis do you think so)?!

In fact, lack of such “relevant CT clauses” was exactly the reason for CC-BY-SA/ODbL schism back in the day – surely, if OSMF could’ve just relicensed the existing CC-BY-SA data without any fuss as ODbL by simple proclamation, they would have done so, wouldn’t you agree?

I agree with first part, but disagree with second. OSMF would not be able to keep e.g. my old ODbL-licensed amenity=restaurant+name=r1 and mix that with some new evil-licensed other amenity=restaurant+name=r2 in the same database and distribute the result as a single planet.osm.bz2 as they do currently.[5]

That is governed by section 4.4 of ODbL 1.0 (and prior definition of Derivative database in section 1.0) which requires resulting derivative work to be licensed exclusively either under ODbL or compatible license.

I am unsure what you wanted to say here, and additionally find it unclear what entities you meant. Perhaps a clear explicit example might help to explain what you wanted to convey (if it is relevant for the topic the two of use are mainly discussing, and not just a sidenote)?

Especially as in cases of e.g. sublicensing (which is the point there), the same entity (like OSMF) is both the licensor (to its downstream users who download the data) and a licensee (of its upstream mappers who uploaded the data, and are actual copyright holders of the data)[6]

I have not noticed, as I don’t have much interest in AI (buildings or otherwise – except as insofar in how to stop its overly excessive proliferation, or at least personally avoid as much as I can of it[7]), and thus I have not put any amount of research into MS buildings dataset and how it is being distributed/use in combination with OSM.

But at first sight I see at least 6 possible[8] “solutions” that do not require “universe to vanish”[9]:

  • CDLA might be compatible with ODbL, thus satisfying section 4.4 of ODbL.
  • Specific combination[10] of e.g. CDLA buildings + ODbL road/amenities that is being distributed as a database is considered “Collective Database” and not “Derivative Database” (as defined in section 1.0 of ODbL) – e.g. if the layers are kept completely separated in the database, so 4.4 requirements do not apply
  • What is is being distributed is considered “Produced Work” (as defined in section 1.0 of ODbL) and not “Derivative Database”, so 4.4 requirements do not apply (e.g. if only prerendered TMS tiles leave their server, it would not be considered “Conveying of the Database”)
  • The combination of MS buildings + other OSM data is being locally built, and not being redistributed as another database at all, in which case copyright does not apply – as it only covers certain things like distribution, licensing etc, and not mere “use” of a work.
    (e.g. just like how you can link GPLv3 program with completely proprietary library and use it just fine - but you can’t redistribute or sublicense the resulting binary)
  • it might’ve been done in jurisdiction that does not extend copyright protection to databases nor have specialized database protection rights (e.g. EU’s sui-generis database rights)[11]
  • the thing that was done was actually illegal, but nobody cared enough to sue yet (or they did, and the case is still going through the motions, e.g. like AI scraping of libgen/Anna’s archives pirated content did, or Copilot scraping GPL content and producing non-GPL works based on that data[12]).

(There are of course possibly other reasons too, but those 6 came immediately to my mind).

Have you researched all six of them, and ruled out each one?

I would suggest people do so, before jumping to (incorrect, IMHO[13]) conclusion that ODbL does not provide any protections against fully closing database licensed under it.

TL;DR: ODbL itself does contain protections (in section 4.4) that prevent mixing new closed data with existing ODbL-licensed data and distributing the resulting database. CT provisions (e.g. 2 and 3) were not required for that protection.

IOW, it is ODbL which provides protection in case of hostile takeover of OSMF, not CT (in fact, CT slightly reduce that protection – but do provide other benefits unrelated to OSMF takeover, as noted in my previous post)

[*] entirety of this post is wholly human-produced work with meticulous[14] and time-consuming manual effort. No AI was used to generate any part of the text, and no AI was being used at all intentionally by me[15] while writing it. Only neural net that was used intentionally is squishy carbon-based one inside my head.

  1. as the license I’ve granted OSMF is non-exclusive ↩︎

  2. such as i.e. planet-260727.osm.bz2, should the split happen today ↩︎

  3. but if you disagree please clarify what you consider as “relevant clauses”, as I asked for above ↩︎

  4. i.e. the one that they are currently published as, i.e. ODbL 1.0, or compatible with it ↩︎

  5. whether some hostile takeoverrer would be able to find some legal exploit that allowed them the keep data in two separate databases and interlink them somehow in a useful yet license-compliant way is also dubious, but I must concede not impossible. But, if possible at all, it would involve enormous technical work and not simple proclamation of new license ↩︎

  6. as that phrase reduces down to “restricts the use of the data by a OSMF, but not by the OSMF” in that case, which you surely must agree is confusing ↩︎

  7. at least when talking about their unproportionally damaging variants whose externalized costs are several order of magnitude bigger than benefits they bring ↩︎

  8. Note that I have not investigated at all whether any of those are true, I’m just saying that at the very least each of them should be carefully evaluated before drawing conclusions ↩︎

  9. it was quite poetically phrased, though; I like it :blush: ↩︎

  10. As noted before, I have not investigated what that dataset actually does, nor how it is combined if at all; and rather leave that as an exercise to the reader ↩︎

  11. see e.g. discussions on “creativity/work-of-art” vs. “sweat-of-the-brow” doctrines, and (un)copyrightability of facts themselves ↩︎

  12. or have any of those concluded finally already without possible further appeals? pls link if they did, it is hard to follow everything that is happening, so I might have missed them! ↩︎

  13. but hopefully opinion that is provided with enough information of why the claim is considered incorrect, so that it can be verified by sufficiently interested parties. Of course, it might be that it is my reasoning that is incomplete/incorrect, and in such case I’d love to hear your – equally substantiated, if at all possible – counterarguments exactly why you think so ↩︎

  14. yes it is long, because the subject it hard, and attempts to be shorter have failed to reach the agreement/understanding. On the bright side, I have invested significant effort into trying to make it more easily readable and properly referenced, so I’d greatly appreciate if the favor was returned in replies (if any) ↩︎

  15. but I cannot guarantee whether DDG search engine wasn’t using some AI in the background unbeknownst to me, or whether it was used in creation of some of the linked URLs ↩︎

2 Likes
Who Really Owns OpenStreetMap? A Personal Concern About Big Tech in OSM
2

Not quoting the very long text that is based on a clear misunderstanding.

See Licence/Contributor Terms - OpenStreetMap Foundation clause 2: you grant the OSMF all rights necessary to do anything they want with the data. You are NOT licensing your data to the OSMF on ODBL terms.

Clause 3 in turn limits the OSMF to distributing your contributions as part of a database on ODBL 1.0, CC BY SA 2.0, or on terms of other open licences that can be determined by a vote of active contributors.

Yes, the OSMF could totally licence a current planet dump on CC BY SA 2.0 terms*. Nothing in the ODBL requires the licensor to exclusively license on its terms.

And finally, I had already pointed out that the CDLA is not compatible with the ODBL.

* the CTs went in to effect before the licence change, at a point in time when the data was still being distributed on CC BY SA 2.0 terms. Replacing the 1-2 very ambiguous sentences that previously allowed the OSMF to distribute the data.

1 Like
4

Yes, I though as much that we might not be understanding what other is saying.

  • “If the CTs didn’t have the relevant clauses they [OSMF] could simply start distributing the data on any terms they liked” that was sounding clearly incorrect, as by “the data” I understood you meant “existing OSM data”[1], and it seems now that you have meant “(empty set) of new data” (in which case I agree!)

Yes, I fully agree with your understanding of clauses 2 and 3 and CT.

What I disagree with is (what I understood you claim to be):

  • “the ODbL does essentially nothing in that respect [making OSMF hostile takeover much less attractive]”, which is IMHO wrong due to OdBL 4.4 which IMHO would do pretty much to make it unattractive.
    But note that it is not relevant to current situation, as we do have CT, and OSM data is not bound just by ODbL (and ODbL could be removed fully, should OSMF decide so, so any ODbL protections would disappear from that moment on[2] – something that could not have happened without CT2,3 but only pure ODbL - which is the point I was trying to make).

Yes (due to CT2,3), and that was exactly a CT risk I was mentioning before. Some future OSMF which had hostile takeover could likely relicence the OSM data as CC0 or WTFPL, or even some other license which is more closed but still can be smuggled under “Free and open” ambiguous phrase.

Then they could use technical limits to disallow scraping via API, use ToS to restrict how and what you can do with data (like GoogleMaps does), sue people trying to get around those, stop publishing planet dumps and minutely diffs, and only distribute planet data to its buyers under NDA/contracts which prohibits them from sharing the data (i.e. “pull a RedHat move”) and requiring them to only use data under closed source software.

While that legally would not be breaking CT (and ODbL would no longer matter for any new data) the effect would be pretty close to “OSM data is not open anymore”, while they could continue benefiting from remaining OSM userbase (those who were ignorant or didn’t care enough about freedom issue, instead just being interested in usability) while keeping control all the existing data[3] for their own uses as they see fit.

And that risk is why I personally would have preferred if OSM were to either remain under CC-BY-SA only, or gone ODbL-only, instead of choosing current CT solution[4], i.e. “putting all our eggs in single basket”. But, as said, water under the bridge now. We’ll just have to keep a close eye on OSMF and its voting members to prevent such scenario from ever happening, eh? The price of freedom is eternal vigilence…

You did make that claim, yes, that is why I mentioned it in reply. Alas, it didn’t seem to come with any references or any arguments substantiating that claim, so I could only take your word on it[5]. But, good for you if you have verified that claim for yourself.

So only 5 more bullterpoints remain to be eliminated, before we allow the universe to unavoidably collapse, yes? :wink:

Yes, technical piece of history; IIRC CT was considered a vessel to implement ODbL at the time, right? I.e. it was not that OSM was being relicensed to ODbL, it was that the new CT (which you had to accept lest all your years work be deleted[6]) was there as a means for relicense to ODbL (c|w)ould happen in a simple cutoff-point.

Ah, I might have missed issues about those at a time. :man_shrugging: I would probably still preferred only for those problematic terms to be made less ambiguous, then to giving OSMF (or any SPOF entity) that much power as CT did. Copyleft-like licenses work the best when the power is distributed over many licensors[7], IMHO.

  1. and which would clearly been incorrect in alternate history line where CT lacked 2 and 3, and all data was exclusively under ODbL license ↩︎

  2. only remaining if you manage to fetch old dump of database still under ODbL, but any new data that come in would be out of your reach ↩︎

  3. thus making it much more attractive! ↩︎

  4. I am not a big fan of copyright assignments that some FOSS project use either, for pretty much the same risks ↩︎

  5. or do my own research/analyses of course, but as noted earlier I do not care about Microsoft buildings enough to be inclined to do so ↩︎

  6. well, not actually deleted, as there is still a dump available, but deleted from the active OSM database which was relevant ↩︎

  7. In fact, it may be argued it is the only way that they can work reliably.
    If everyone transfers all their powers to someone/something else, then it is that entity, and not you as a contributor, who have the power; including the power to override your wishes of data remaining open. IOW, the similar failure mode as in so-called “representative democracy”, for which clearer term “Elective dictatorship” would usually fit. cough Trump cough :wink: ↩︎

5

not really? How it would make hostile takeover less attractive?

This part, in case of OSMF hostile takeover succeeding, actually makes things a bit worse.

As if people would recreate new organization and use existing data, then they would need to credit OpenStreetMap (now controlled by hostile entity)

While OSMF (now taken over) could relicence data.

disclaimer: not a lawyer, may be misunderstanding something

6

Again you are confusing things here, the OSMF can completely change the licence (with enough votes) and distribute existing data on new licence terms, or simply on CC BY SA 2.0 ones for which it already has “permission”, no need at all to start fresh or anything like that.

Sigh, it isn’t my understanding: this was settled before I had anything to do with it*, @woodpeck has a nice slide on the topic in this talk Die neue OpenStreetMap-Lizenz - media.ccc.de (slide 20 is a bit iffy though as prior to the CTs there was only a very ambiguous text that I have already mentioned).

And no there is no “CT risk” as long as a substantive number of contributors maintain their “active mappers” status, and if they don’t, well then I guess nobody cares any more. The OSMF can’t even pull a Wikipedia (re-licence to a new licence somebody else declared compatible) if you so want the licenses are pinned (on purpose).

The CDLA can’t be compatible with the ODBL 1.0 because, first principles, the ODBL 1.0 is a share-alike licence (I don’t even need to point to Kathleens paper analysing the the LF data licenses).

* the CTs have a small number of issues that I would have loved to fix, but by the time I became involved they were for all practical purposes cast in stone. But those issues have nothing to do with the topic at hand.

7

A PS to this: CC BY SA 2.0 was completely unworkable as a data licence and would have likely been completely ignored once we had reached a size at which we were relevant, while the ODBL is no panacea in that respect, at least there is a sound legal theory underlying it for use as a data licence. And no secret, with the exception of maybe residents of the UK, the sole legislation that still has a sweat of the brow copyright doctrine, as an individual mapper you typically don’t have any IP rights in your contribution to start with (some non-typical exceptions noted), and you can yell CC-BY-SA till you are blue in the face, this is not going to change what was already clear over one and a half decades ago.

1 Like
8

As pointed out at the beginning of this discussion, the ODbL does essentially nothing in that respect, the contributor terms is what binds the OSMF to an open licence and spells out the terms on which it can be changed. Without that there would be nothing stopping the OSMF taking the data private.

Who Really Owns OpenStreetMap? A Personal Concern About Big Tech in OSM
9

This might be getting off-topic (so please do split to new thread if you agree but think it is worthwhile to discuss) but I think I disagree with you here. ODbL requires (to simplify) that if you distribute the data, you must keep it open. If you improve it, you must also open those improvements[1]. So it is in fact OdBL that protects OSM data from being closed (IMHO, of course).

Yes, what CT primarily allows, is changing the license under which the OSM database is published (which is ODbL 1.0 currently) to some other license similar in spirit, without needing to get explicit permission from each mapper again[2]. In a sense, it is our equivalent of “GPLv2 or later”, allowing “upgrading” the license in case current one is found horribly broken in the future.

So, even if CT were (almost) completely missing, AFAICT any new data contributed to api.osm.org would still be covered by ODbL[3].

But the existing CT terms are double-edged sword IMHO:

  • while they allow us to easily switch to say ODbL 2.0 (or something else) if ODbL 1.0 is found to have legal exploit which makes it useless for its intended purpose[4] of keeping the data open, without going through the pains (and data loss) of individual relicensing[5],
  • they also however introduce the problem that if there is hostile takeover of OSMF, efficient army of evil lawyers maybe could argue that something much more closed/problematic still falls under (undefined!) term of “similarly free and open license” and “upgrade” the license to it without us contributors having to say anything about it.

But I don’t see either of those two possible issues on the horizon yet, so I prefer to currently spend my nerves and efforts on DMA abuses, https://chatcontrol.eu/, android becoming fully proprietary ecosystem https://keepandroidopen.org/ etc. other freedom-fighting tasks, until such time that issues about OSM licensing become possible new incoming threat.

I respectfully disagree. Without CT, ODbL itself would be stopping OSMF from redistributing all existing data under new license. Nothing would of course stop them from starting a new empty database from scratch (and try to get users to contribute to that), and license that new database under some closed license but:

  • that is what anybody else (e.g. Overture) could already do today too, and
  • without CT, they couldn’t relicense old existing data to new license[6], and
  • they couldn’t mix that new data with “old” (i.e. all currently existing![7]) ODbL data – ODbL license prevents such combining with its “copyleftness”.

  1. yes, there are subtleties regarding Derivative vs Collective Database and others, but let’s not delve into that right now ↩︎

  2. Which is something that we learned from CC-BY-SA vs. ODbL schism is very painful ↩︎

  3. otherwise, even OSMF itself would not be allowed to distribute it, i.e. publish planets or diffs or allow api.osm.org read access to it ↩︎

  4. e.g. something kinda like GPL v2.0 had with “tivoization” so GPL v3.0 needed to be drafted ↩︎

  5. which was horrible back in CC-BY-SA days, and I can not even imagine how much worse it would be today – kinda like WW1 vs. WW3 ↩︎

  6. not without explicit confirmation from each mapper, e.g. like in CC-BY-SA-vs-ODbL days ↩︎

  7. well, all data except those by users who willingly and explicitly agree to such relicencing. Doh! ↩︎

1 Like
10

You are getting this all very wrong, the entity holding the necessary rights (if any) to distribute and licence the data is the OSMF. They are the licensor for the majority of the data (skipping the other bits as not relevant for this discussion). If the CTs didn’t have the relevant clauses they could simply start distributing the data on any terms they liked, this wouldn’t effect copies of the data that you had already received at that point, but would any that you obtained going forward.

It literally has nothing to do with the concrete distribution licence, in all scenarios that restricts the use of the data by a licensee, but not by the licensor.

PS: you may have noticed that MS has changed the licence of the MS buildings dataset from the ODbL to the CDLA, a licence that is not compatible with the ODBL, the universe didn’t vanish in a puff of paradoxical smoke when they did that.

11

In short, by forcing updated versions of the database to also continue being distributed under ODbL and available to people under such open license (which evil from-profit would dislike for hopefully obvious reasons, as they can’t squeeze as much money out of people then).

I thought I clearly explained it before, but since you asked, here it is in more detail, in (hopefully more easily digestible) step-by-step format .

  • Premise1: we are analyzing hypothetical situation when OSM database would be governed exclusively via ODbL 1.0 (i.e. not current situation with CT, which can drop ODbL)

  • Claim1: database being licensed under copyleft-alike ODbL provides extra protections against hostile takeover, compared to if it were instead licensed under some permissive non-copyleft license, like e.g. CC-BY; thus making it unattractive target.

  • SubClaim2: ODbL is an open license, and if one conveys to others database published under ODbL, they are required to keep it under same or compatible license, giving them full access to it source database, persuant to 4.4 of ODbL.

  • SubClaim3: modifying ODbL database creates Derivative Database (as defined in 1.0 of ODbL), which follows the same “openness” requirements of 4.4 as in SubClaim2

  • Premise2: control of OSMF voting membership is taken over by entity wanting to close down OSM, i.e. convert it to something akin to Google Maps data silo. In short, they, being evil coorp interested only in profits and not social benefits would want to:

    • not lose existing database and continue benefiting from existing data[1]
    • keep updating the database to keep it current / interesting to buyers[2]
    • distribute the database only to parties they want, and those include only those willing to pay[3]
    • prevent the general public from benefiting from continued updates to database unrestricted and free of charge, like they used to

  • Deduction1: in order to satisfy Premise2.1, they cannot start from scratch (i.e. new database under different license), and SubClaim2 is conflicting with Premise2.4

  • Deduction2: in order to satisfy Premise 2.2, they must keep database open due to SubClaim3

  • SubClaim4: in order to give database only to some as per Premise2.3, but preclude others from having same access as per Premise2.4, they would need to exercise different access control to different parties. That access control might be technical, legal, or sociological (i.e. peer pressure, cultural norms).

  • SubClaim5: sociological pressure is unlikely, as most people using OSM do it precisely because it is open, so sociological pressure would work in opposite direction then they want

  • SubClaim6: legal access control is not possible, due to SubClaim3, unless extensive technical and legal effort were combined to try to circumvent its provisions (e.g. separate databases for data pre and post takeover, and mechanism to combine them and resolve conflicts, presented in a contrieved way which would not constitute derived work per copyright)

  • SubClaim7: technical access control is not likely to be effective (as it only requires one person to circumvent it for whole of it to fall apart (see RockyLinux/AlmaLinux vs. RedHat mentioned before), and in any case would require significant extra efforts (thus making it unattractive and confirming Claim1)

  • Deduction3 - for evil-takeover-company to accomplish Premise2, they would need to accomplish either SubClaim5, SubClaim6, or SubClaim7 without expending noticeable extra effort, and that does not seem possible

  • Conclusion1 - Claim1 is confirmed, as only way for evil-takeover-company to accomplish Premise2, they need to expend noticeable extra effort as per Deduction3.

    Thus, ODbL licensed database would be unattractive for hostile takeover, as it would require investing lots money on much work for a (non-guaranteed) chance of much reduced profits, threatening not to even recoup the takeover costs.

All that being said, I’m certainly capable of errors or misunderstanding.
Now, if you still disagree that “ODbL[4] makes target database licensed under it unattractive for potential hostile takeover”, please either:

  • quote (part of) specific conclusion, (sub)claim, deduction, or premise you disagree with, with explanation why you disagree with it[5], or
  • point out things I missed (e.g. where I made mistake of false dichotomy or such), or
  • were I erroneously referenced the wrong element, or made some other confusing typo

Oh, ok, if that is that is the whole of “worse” that you envision? I was thinking more of data itself. In short, not so worrying to me.[6]

E.g. forks could continue to credit "OpenStreetMap contributors" instead[7]. But I don’t find attribution as problematic if data itself remains free/open, as it is similar situation with e.g. Linux (for which you have to acknowledge Linux Foundation which has turned evil since many years) and it does not affect actually important part i.e. actual geo-data in OSM case[8]. Heck, even very FSF-aligned people continue to happily use GNU/Linux-based things despite Linux Foundation. In fact, one could (and should!) use link of that attribution to promote their case for freedom and how to fight back.

So it might likely even be helping fork’s case, depending on how well they handle the situation. (e.g. I’d say CoMaps probably only benefited by fully acknowledging the history of “MAPS.ME (turned evil) → OrganicMaps (turned partly-evil) → CoMaps (good guys)”, and explaining what and why they forked, while continuing to thank MAPS.ME and OrganicMaps for the positive parts that they did.

  1. as taking control of that was major part of hostile takeover, otherwise they could’ve just create a new map from scratch ↩︎

  2. or its value would quickly diminish ↩︎

  3. preferably significant amounts of money, in order to quickly recoup takeover costs and make as big profits as possible ↩︎

  4. or really, basically any other coplyleft-alike license ↩︎

  5. idea is to point straight and explicit to the minimum example of the problem, and not talk generally or ambiguously ↩︎

  6. Sure, there might be some public relations downsides due to “association” with EvilCo, but such PR-damage could happen without any OSMF takeover, e.g. by Elon Musk becoming OSM platinum sponsor, or someone in OSMF board being named in Epstein case, or found being major donor to fascist party or whatever. But it can also be spinned positively, see e.g. below for CoMaps example ↩︎

  7. Or likely even “OpenStreetMap contributors (before June 2026, when OSMF turned evil)” if one prefers malicious compliance ↩︎

  8. or code, in Linux case ↩︎

12

Claim1 is faulty and everything that follows from that which is essentially the whole text.

13

This is where I don’t follow your reasoning. Who are the “others” in this scenario? As I understand it, Evil OSMF is the same entity as Good OSMF, just with different people voting. Evil OSMF doesn’t receive a copy of the database under license from Good OSMF, because they are the same entity.

1 Like
14

OK, I thought that it is about OpenStreetMap Foundation - not some variant not applicable to reality and irrelevant to our situation

15

I think maybe the core of @Matija_Nalis’ misunderstanding might be that he assumes the following: An OSM contributor downloads an ODbL database into their editor, creates a derived work by making an edit, and uploads that derived work (which legally is also ODbL) back to OSM.

Matija seems to assume that this would, in the absence of contributor terms, perpetuate the ODbL.

But since “assume we didn’t have CTs” is only a thought experiment, I don’t see any useful outcome of this discussion other than demonstrating superior intellect?

3 Likes
16

It just doesn’t work that way, it isn’t a creative work like code (there’s a reason CLAs have become popular because irl that’s actually a pain for the owners of a code base), it is just data and there is no separate IP in his changes. Further all the derivative database terms in the ODbL apply to publicly used derivates, but in this scenario it is the OSMF publishing the database, not @Matija_Nalis.

17

How to tell me “I didn’t even bother reading the text” without saying “I didn’t even bother reading the text” :cry:

nothing follows from Claim1. Everything else leads to it. It is mentioned first, not because it is in a chronological order, but because it is customary to state what you are trying to prove before starting to prove it. Like for example “Subject” is put at top of e-mails, and not on their bottom.

Everyone to who the database have been conveyed to. You, me, OsmAnd building their per-country extracts, overpass server admins, facebook, etc.

In the scenario described, there is only “Evil OSMF”, as “Good OSMF” has ceased to exist (i.e. has become “Evil OSMF” via hostile takeover), due to Premise2.

18

Yes, that is pretty much my interpretation. By using word “misunderstanding”, should I take it you don’t agree with that interpretation (and if so, could you clarify why you disagree?)

Nothing more, that was pretty much it :rofl:

In all seriousness though, that discussion of hypothetical situation spawned from this claim:

Which I didn’t agree with, and thus wanted to either:

  • debunk the incorrect claim (because xkcd #386), or
  • find out what is the cause of the misunderstanding (perhaps we are not talking about same things?[1]), or
  • have it explained to me why my understanding of ODbL is wrong[2]

  1. see parts of the discussion asking for clarification are we talking about whole database including original ODbL data, or only new-and-fully-empty database ↩︎

  2. because, I want to learn ↩︎

19

Right, so there is one continuous OSMF that acts as licensor of the database at all times. Other people have copies of the database and are limited by the ODBL terms under which they received it. But how does that affect OSMF?

20

IMHO this is where you are going wrong again, this only commits the licensor (in this case the OSMF) in respect to the instance of the database you just received (to allow you to modify and distribute the database on the terms of the ODbL). It neither creates an obligation to release future instances of the database to you on the same terms, nor to do the same for any third parties except for those that you distribute it to * **

Yes you are correct in that, and I’ve already pointed this out many times, it is not possible to claw back instances of the database that have already been distributed on ODbL terms, but that doesn’t help your argument that somehow the ODbL stops the licensor from changing the licence.

Another place where you get this wrong: “Subclaim 2.1” … and if one conveys … it should be “… and if a licensee conveys ..”, this doesn’t apply to the licensor.

* as a number of licences, including all CC licenses except CC0, the ODbL doesn’t allow sublicencing, so the licensor commits to directly licencing derivates created by you on ODbL terms.

** fwiw I’m not aware of any OSS or open data licenses that would restrict the licensor in the fashion you think the ODbL does.

1 Like