Questions to LWG on ODbL

From OpenStreetMap Wiki
Jump to: navigation, search

I have added comments to the responses so far. I don't intend my comments to be considered as part of the final document, if there is such a thing, as they are more my personal views. I am interested in getting the LWG response and circulating it as widely as possible for information purposes. --TimSC 19:52, 15 February 2010 (UTC)

Contents

Original Preamble to Questions

To the License Working Group,

I have attempted to summarise the issues raised on the “Why You Should Vote No” wiki page. I hope that the LWG can consider and give a rebuttal on these points to assist contributors in making an informed decision on OBdL re-licensing. Apologies to the wiki contributors, if I have distorted your points beyond recognition.

It might be most productive to pitch the responses without assuming in-depth legal knowledge but provide references to legal opinion where it is available.

Regards,

Tim Sheerman-Chase (OSM Username: TimSC)

This document is released under Creative Commons Attribution-ShareAlike 2.0

Creative Commons Comments on ODbL

Creative Commons provided comments on ODbL. Can each of these be individually addressed?

  • ODbL Fails to Promote Legal Predictability and Certainty Over Use of Databases
  • The ODbL Is Complex and Difficult for Non-Lawyers to Understand and Apply
  • The ODbL Can Result in High Transaction Costs on the Data Sharing Community
  • The ODbL Imposes Contractual Obligations Even in the Absence of Copyright
  • The ODbL Can Cause Legal Non-Interoperability for Data Sharing

Full text at [1]

Response

  • Legal predictability: There is no absolute certainty, but more than with CC-BY-SA for geodata. Creative Commons suggests Public Domain as a solution.
    • According to surveys by OSMF and from the OSM community, the majority of OSM contributors are against PD. However, when signing up to the new license terms, users will be able to indicate a preference for PD.
  • Complexity: see Proposal document, p. 5
  • Transaction cost: refers to ODbL for research databases, and is a legitimate concern in that context.
  • Contractual obligations: It is true that this is a new approach and there are uncertainties. Again the only solution without uncertainties would be PD.
  • Non-interoperability: True for all share-alike type licenses, but with ODbL Produced Works data sharing is much easier than with CC-BY-SA.

[Added clarification: PD = Public Domain per see or a PD-like license, such as CC0]

Futher comments from TimSC

I think you will need to state why the predictability of ODbL is satisfactory rather than bash CC-BY-SA. The view of the contributors aren't relevant to addressing Creative Commons comments.

The human readable summary helps with the complexity. But proposal document states: "Complexity in law is not something that the LWG can address directly [...]." This does not make sense to me as the comment blames the law for complexity, when the concern is with the ODbL. As I see it, the trade off in writing the license is between "features", complexity and predictability. You have sensibly tried to cover the main legal eventualities, but it is conceivable the license has simply become too complex to justify the benefits. You might chose to simplify the license at the cost of predictability or to remove some of the features of the license. It would be worth stating why these options are not the best.

The point of transaction costs applies to all users of the database, not just research AFAIK. The mentioning of scientists in the CC document was just an illustration.

"Again the only solution without uncertainties would be PD." To be pedantic, not all jurisdictions recognise PD, AFAIK so there is non-zero uncertainty with all options. Also to only see the certainty issue is ODbL or PD is oversimplifying. A simpler license could, in principle, be written (but it would probably be detrimental to legal certainty or license "features")

"Non-interoperability: True for all share-alike type licenses" I am not sure this is generally true. Can't software licenses interoperate such as importing LGPL into GPL? If this does not apply to databases, please cite reasons.

Response from Mike Collinson

Predictability: The ODbL can thought of as having two components. 1) What we want. 2) How we intend to enforce that.

What we want is free use of our data, Attribution and Share-Alike.

The first two of want we want are obvious, and therefore I believe, predictable. Share-Alike of data is a different beast and here CC has a point. Data does not have reasonably clear distinction between making it (e.g. writing software) and using it (e.g. writing a book with a word-processor). If you prepare your data source to make a map, are you making data or using it? Data is like a cooking ingredient, it is there to take a little of this, a little of that and mix it all together. CC suggest the solution is simply not to have Share-Alike and suggest [CC0]. Fair enough. But if folks do want Share-Alike, then ODbL does have predictability. It IS written for databases. It does define Derivative Databases (Share-Alike needed) and Collective Databases (Share-Alike not needed). It does to a large degree split making (data/database) from use (Produced Work, maps) and allows other licenses to be used for the latter. The degree of predictability will be tested by use. Since OSMF is the licensor rather than 220,000+ contributors, there is someone to ask. OSMF can ask the community to make guidelines which it can endorse. For example, CC do not quote examples but two areas of unpredictability are 1) "Trivial Transformations" - if you transform the data with some algorythmn to make access faster for a game but do not add data, is this a Derivative that you have to make available?, 2) Artificial Collective Databases, keeping OSM data in one, your own additions in another, simply to get around sharing data.

We intend to enforce by using copyright, contract and database rights.

CC point out that these vary wildly throughout the world when applied to data. Fair enough, this is why we are using three rather than one. They go on to argue, their main predictability argument, that this leads to unpredictability and that is bad. Frankly, I just don't buy this - if the whole intent is to somehow enforce what we want above, why is unpredictability a bad thing? I would certainly like to see any real or hypothetical examples of "good guys" having difficulty using our license because of this.

Non-interoperability: Exceptions can certainly be explicitly made. I just cannot see how ODbL is special here. CC has a valid point, and conclude that "Open Data Sharing Should Converge on the Public Domain". Fair enough, ... if, as much of our community is not currently, prepared to drop Share-Alike.

The OSM Foundation will have unlimited rights to all your data/Change of license terms using voting but against the wishes of a minority of individual contributors

Each contributor holds the copyright to their contributions and currently, the license may not be changes without their permission. The new license permits a license change after a vote of OSMF members and active contributors. This violates the principle that the contributor may exclusively determine what terms apply to use of their data.

This is in context with an expansion of OSMF influence. On the OSMF wiki, on the page “What is the foundation”, it states “The Foundation ... has no desire to own the data” which is presumably the previous OSMF position. A revised statement of what the OSMF is and how it relates to the OSM contributors should be produced.

Response

Even with CC-BY-SA 2.0, contributors accept that Creative Commons can issue new versions of the license without their permission. The OSMF does not have unlimited rights: License changes can only be to free and open licenses and require a 2/3 vote from active contributors.

Further comments by TimSC

It is not true that CC-BY-SA automatically re-licenses work to the latest version. Their website: "No works are automatically put under the new license, however."

If the uses of contributions are so well defined (point 3 of the contributor terms), OSMF doesn't need a "worldwide, royalty-free, non-exclusive, perpetual, irrevocable license"?

The "tyranny by majority" issue has not yet been addressed.

'Response from Mike Collinson

Yes, there is a 'Tyranny of the Majority'. If you are dead, are no longer actively editting or have not bothered to keep your contact details up to date, then the license can be changed without your, or your estate's, agreement. The question how you balance the effect of this potential oppression of the individual contributor compared with the public good of keeping a license up to date. Given my age, I shall probably be dead or not wanting to be bothered when the dataset from this project is still being actively used. I may not agree to a future change to another free and open license, but I do not lose monetarilly, my daily life remains unchanged and my general aspiration of contributing data that can be used by everyone is still met. I would much prefer it if the license were just changed rather than having to contact my daughter and ask her permission. I argue that we have got it about right. If I care about the project and still mapping (i.e. my data has not been effectively replaced by later edits), then I get a say. If I do not care, no say.

Why the "worldwide, royalty-free, non-exclusive, perpetual, irrevocable license" bit?

Why not? It is world map, none of us are contributing with the expectation of being paid, a contributor should be free to do what they like with data they themselves have created, and why should there be a point when the database becomes unusable because licenses on parts of the data have expired? Or taking the liberal nature of the whole of section 2:

  • It is easy for end users to read one license, the ODbL or current successor, controlling the whole database rather than having to check that some of the individual parts of the data are under a more restrictive license due to Contributor Terms that may change over time.
  • Flexibility for unknown circumstances. Future active contributors will be more governed by broadly deciding what a "free and open" license should be than having made certain options made impossible by, possibly unintended, restrictions here.
  • Non-substantial extracts. European database law kicks in when extracts from the database become ["Substantial"]. We'd like small users, such as school children, village communities, to very easily use our data in conjunction with their projects.

The OSM Foundation will have unlimited rights to all your data/Change of license by a hostile party

A power grab might be attempted by a hostile 3rd party using bots to create voting user accounts or a large group join as OSMF members specifically to influence voting. The ODbL license terms might be changed to remove the share alike term, etc. Also, there are no safeguards in place that would require the OSMF to call any kind of vote before changing the Contributor Terms. Finally, has “active contributors” been defined?

Response

According to the Contributor Terms, bots cannot vote - a contributor with multiple accounts has only one vote. If OSMF changes the contributor terms, it will have to ask each user to accept the new terms. "Active contributors" definition is in the Contributor Terms.

Further comments by TimSC

How is the no bots rule enforceable?

Response from Mike Collinson

Best effort, investigating reports, looking for suspicious patterns, blocks of similar email addresses, same IP addresses. In political elections, whole sections of cemetaries can arise to vote despite no dead person rules. The key is to ensure that the number of votes from legitimate contributors is far greater than those from crazies.

Even if the current licence is less strong than it could be, there is no evidence it causes any problem in practice/Principle of least damage

There are a few instances of users finding CC BY-SA too dangerous to use, but this is far out-weighted by users that are happy with the existing license. There is no evidence that people are abusing the intent of the existing license. So, even if we assume CC BY-SA is not legally sound, the practical effect is the intent of the current license has not been compromised.

Response

In our opinion it is better to be proactive than wait until abuse occurs. OSMF vote and community survey show there is strong support for the license change.

Further comments by TimSC

Why be proactive? The extent of future abuse is probably unknown. However the potential damage is very possible and significant with 14% voting no on the doodle poll. I suggest the response focus on how much more use we would achieve with a more certain license?

Even if the new licence is better, the cost of switching is unknown or too high

No serious study on how many users would re-license their data has been conducted. It may be that a few significant contributors withhold their data and cause a large loss of data. The plans to mitigate this are sketchy and not presented in the proposal document.

Also, if a large proportion do not relicense under OBbL, OSMF has not proposed a “Plan B” to revise ODbL or adopt a different license. This apparently gives contributors a choice between agreeing with ODbL or the project crashing and burning, i.e. not a real choice.

Response

The plan is flexible because we cannot foresee all possibilites. Following a fixed plan might cause damage.

For a successful license change, we need a very large proportion of users to relicense. Otherwise there will be a Plan B. OSMF will make sure there is no large loss of data (and even if we didn't, the CC-BY-SA data would still be there for others to use).

Futher comments from TimSC

A comment, none of this is publically visible on the main OSM pages. It would be good to publicise your current work as far as possible. I suggest you add information to the OSM wiki front page and the OSMF blog.

I don't recall reading further revisions of the license being publically raised as a possibility before the OSM re-licensing date?

ODbL licensed data can be reverse engineered

The proposal document states OSMF “concluded that it [the risk] was minimal”. Can the background to this conclusion be provided?

Response

We had discussions with our legal counsel. Works made with ODbL data needs the datasource to be attributed (incl. license). This way it is clear that there is an underlying datasource and under which license it has been made public. Creating a database by drawing over a map is actually the recreation of the origional database. This newly created database will legally be the same as the source database and will therefore still have the same license.

Protests Against the Method of OBdL Adoption

Although OSMF has invited comments from many interested parties, they have not polled the contributors. Claims as to what the “silent majority” want are mere speculation and it would only carry weight if a poll were conducted. The OSMF is meant to have a supporting role, while the contributors are the heart of the project. Specifically, OSMF did not seek from all or majority of contributors the OK to change the licence away from BY-SA and they did seek a majority consensus as to which license we should change to become. “The OSMF has considered existing and emerging licences and has come to the conclusion that maintaining a Share-Alike/Attribution licence is in the project's best interests[...]” - meaning the OSMF committees and working groups took the decision, not the majority of contributors.

The poll of OSMF members is a positive step but further active consultation is needed. But the poll of OSMF members was probably prefaced by the OBdL proposal document. If the poll had been accompanied by a proposal for a different license, we may have seen a different result. Is it possible to regularly poll the contributors as a guide for future policy decisions, in line with OSMF's supporting role?

Response

OSMF not polling the contributors was a mistake (says Ulf), but someone else did a survey of their own. The Doodle survey shows clear support for the license change.

The license draft was discussed and significantly improved based on input from the community over a long period of time on email lists and on the OpenStreetMap wiki before the voting period. We accompanied the OSMF vote with a very neutral survey, the results of which can be seen at http://www.osmfoundation.org/wiki/License/OSMF_member_vote_results. Certainly, continuing that kind process in the future with the broad contributor base sounds a good idea.

Futher comments from TimSC

"We accompanied the OSMF vote with a very neutral survey" If you attached Creative Common's comments on ODbL, I might agree it was neutral. The OSMF might need to do some research on how to conduct fair and objective polls. The doodle poll was not officially sanctioned and not randomised, so we can't draw any firm conclusions from it. Also, not all options were presented (for example PD and not ODbL). The fact is OSMF still have not formally asked the contributors if the project should adopt ODbL or officially polled them on opinion.

Response from Mike Collinson on ODbL Adoption

The approach we have taken is to engage anyone who is interested in the license and is prepared to take a little time to read up and understand the issues involved. We've gradually broadened from the license mailing list, to OSMF members and now that the key documents are all finalised so that there are no ifs and buts, to the entire contributor community. We openly publish as much as we can ... minutes, drafts, FAQs and have wiki pages where anyone can edit. From feedback, I now believe we could have done a better, earlier job of publicising disagreeing/alternative positions but believe we are coming up to speed on that. Contributors will be asked shortly whether they want to accept or decline the new license and we are gradually organising and summarising our published material to help them if they want more info.

Incomplete History for Split Ways

The editors split ways in a manner that make it difficult or impossible to reconstruct the entire history of a way. Without knowing who is the original contributor of each way, data may be transitioned into the ODbL database that violates someone's IP rights. Even if only a tiny fraction of ways are effected, the legal ambiguity would be harmful to the project.

Response

We've consulted legal counsel on the broad issue: shared nodes and so on. (xxxx more here needed xxx) Yes, split ways is incredibly difficult and we are unlikely to achieve it. The ultimate response is, any where there is questionable data in the database and the owner protests or it is otherwise reported, we, or rather the Data Working Group, will remove it.

Futher comments from TimSC

I am inclined to agree on this one, since we don't have much choice! (in the long run)

Not all questions answered on the Use Case page

Some use cases on wiki are not addressed (although some or all are probably recent additions).

Response

As far as we can see, they are all addressed but we are always happy to look at specific issues communicated to us.

Futher comments from TimSC

There are still yellow boxes around issues TBC. Surely we can agree on the intention of ODbL at this stage?

Response from Mike Collinson

We have removed those now and added clarificatory text. Basically, there are a small number of issues which we feel should continue to be looked at in the long term but have been 'covered' as far as the release of ODbL 1.0 is concerned.

Not really Share Alike/Making Copies of your own map may not be allowed

The intention of the current CC license is that all derivative work is share alike (the extent that it achieves this is disputed). The ODbL intentionally changes this to allow raster maps (produced works) to be of any license. This might not be desirable as some people prefer use of their work to be continued to be kept share alike to promote reuse of publications.

Despite OSMF claiming that OBbL is in the spirit of CC BY-SA, this is a departure in the intent of the current license. What is the justification to move away from the existing uniformly applied share alike terms? Is OSMF in a position to judge, on behalf of the community, that use of OSM data by commercial companies out weights the benefits of having share alike produced works?

Response

We are a data project not a map tile project.

ODbL means share-alike on the original data (cf "your own map"). With CC-BY-SA you might get access to the raster maps only, but not to the data used to create them.

The additional flexibility on produced works (for example maps made from OSM data) doesn't just help commercial companies. For example it also means you can use NC licensed data with produced works. Please remember that the principal objective of the project is "The project was started because most maps you think of as free actually have legal or technical restrictions on their use, holding back people from using them in creative, productive, or unexpected ways." ... not Share-Alike.

We believe we have achieved a reasonable consensus between a large sector of the OSM community who aren't interested in Share Alike provisions at all and those who believe they are important.

Futher comments from TimSC

You seem to support SA in one argument and be anti-SA the next. The fact is produced works under ODbL are not share alike. SA is an encouragement for other parties to also release their data as share alike. Stallman makes this point in the context of software quite well:

If we amass a collection of powerful GPL-covered libraries that have no parallel available to proprietary software, they will provide a range of useful modules to serve as building blocks in new free programs. This will be a significant advantage for further free software development, and some projects will decide to make software free in order to use these libraries. University projects can easily be influenced; nowadays, as companies begin to consider making software free, even some commercial projects can be influenced in this way.

So this actually futhers OSM's goal to increase use of maps in "creative, productive, or unexpected ways". So getting the raster maps of the final product from multiple sources may be better than just the updated OSM derived data. Another alternative is to insist both the data and the produced work are share alike.

And if all restrictions are bad (as implied in the OSM mission statement), why not just use PD?

"We believe we have achieved a reasonable consensus between a large sector of the OSM community who aren't interested in Share Alike provisions at all and those who believe they are important." I would be interested in hearing who is pushing for this form of share alike? And can their arguments for their case be presented?

"We are a data project not a map tile project." In LWG's view perhaps, but the current situation is the raster map is share alike. And therefore the cartography colours and label placement as well? (Not sure on this) So currently we are BOTH a data and raster project. ODbL changes this AFAIK.

My personal view is coalescing to be the ODbL version of share alike is the worst of both worlds: complexity and little protection (for produced works specifically). When I share my data with you under SA, I expect you to share back to me any produced/derived works.

Can single contributors sue license violators?

In the case of abuse of OSM data, is the contributor and/or the OSMF able to sue? What grounds would be most appropriate? (Obviously, answer for only the main jurisdictions.)

Response

We understand that OSMF can sue in case of abuse, and contributors can sue if the license on their own data is violated.

Further Comments by TimSC

One what grounds can a contributor sue? (Original point made here [2])

Response from Mike Collinson

Ownership. We have since had specific legal advice to confirm this. Our understanding is as follows. Contributors do not hand-over ownership of their contributions to OSMF, they license the OSMF. If they see their data being used in a way that they feel is not what they have allowed, then they can sue. That would include any third party, since the data is still 'theirs'.

OSM's Contributor Terms are not compatible with ODbL

Taken from the wiki directly, consider the following scenario:

  • 1.X takes a copy of the OSM data
  • 2.X adds new data to it
  • 3.X publishes it under ODbL as a derived database ... fine so far ...
  • 4.Y receives a copy of the database from X
  • 5.Y cannot add the new data to OSM because of the Contributor Terms

Only X who is the copyright owner of the derived database can agree to the contributor terms. Y is not the owner and cannot grant "a worldwide, royalty-free, non-exclusive, perpetual, irrevocable license to do any act that is restricted by copyright" to the data.

Response

There is a trade-off between flexibility with future license changes and importing ODbL licensed data. We feel it very important that there is a mechanism to review the license over the years and that it should be left there. Yes, it creates an issue with third-party ODbL imports but our data may well have a life of 50+ years and we cannot predict what a future generation want or need. If Y cannot simply ask the specific permission of X then, for the time being, ODbL imports can be accepted on a case by case basis. There is always the option of slightly modified Contributor Terms for specific donations provided that it does not violate the general spirit of the project. Non-substantial parts can be imported under the DbCL, which is compatible with the Contributor Terms. (We are making some adjustments to the Contributor Terms to incorporate the "DbCL" Database Contents License mentioned here).

OSMF Has Not Made The Case Against Public Domain like Licenses

Despite Creative Commons recommending public domain for databases, OSMF has similar stated it is “not in the project's best interests”. Can further background leading up to this conclusion be made available?

Response

This is not a statement that the License Working Group has based its work on; it is new to us.

We have tried to create a coherent workable license that achieves a broad consensus across the OSM community. Some of us favour PD, some do not, but our current license is Share-Alike and a significant part of the contributors want to keep that. As has been pointed out, a project split is the worst option for OSM, and we aim to avoid that. In the interests of gathering more data about what contributors want, new users will be able to indicate a preference for PD when signing up to the new license terms.

Futher comments from TimSC

It sounds like the PD option was ignored because it was too controversial or just assumed from the beginning of the working group (ie. not decided with any accountability). I seem to recall that Steve Coast, OSM's esteemed founder, is anti-PD which might explain things. And if project fragmentation was to be avoided, more effort should have been made in polling the contributor base at a very early stage - before beginning to draft anything. (Granted, the LWG invited comments from a wide variety of sources - but this would achieve the best option rather than the consensus option.) Just assuming a particular license will be the long term best option doesn't fill me with confidence.

I therefore have a supplementary question: how will LWG or OSMF in future better represent the interests of contributors? Specifically, how will the future direction of licensing be determined?

Response from Mike Collinson

Yes, the LWG is not recommending the PD option now because it is controversial and it was tasked with introducing a workable Share-Alike license. We are not though assuming a particular license will be THE long term best option.

The mechanism for determing future direction of licensing is laid out in the [Contributor Terms] clause 3. Our thought behind it is as follows:

  • Data from our project will probably still have some value in 100 years time. None of us know what is the best long term option nor should we impose on future generations of mappers.
  • We can however agree and impose that data should be released under a "free and open" license. We should leave it to each generation in turn to decide what "free and open" means and how it can best be practically implemented, i.e. slightly refining or radically changing the end user license.
  • By empowering "active contributors" to change the license, we have defined a specific practical mechanism. The interests of contributors are now represented directly. If anyone can persuade 2/3 of active contributors to change the license, then they can. LWG and OSMF are not required. [I understand that you will have concerns about Tyranny of the Majority and I have tried to address those above.]