Open Data License/Closed Issues

What sort of access to Derivative Databases is required?

How often do I have to make a new version available? Must this be as often as my produced work or can I do this on a less frequent basis?

ODbL does not require you to publish derivative databases, but the public use of a derived database or work produced from a derived database requires that an offer of the derived database is made available. This offer can point to a publicly accessible dump, diff or explicit instructions for recreating the database, or it can be an (email) address at which the author can be contacted. If someone takes up the offer - makes a request for the database - you must provide it to them within a reasonable time from receiving the request. ODbL places no explicit time limits on the validity of the offer or the response time to the request, so common sense should be used. It is also acceptable to provide a copy, diff or instructions for the latest version of the derivative database - it isn't necessary to retain old versions of the database. You are not required to provide regular dumps - the only requirement is that you provide them on request.

Unmatched rights between the FIL and the ODbL

The rights granted by the contributor to the data using the FIL (assuming that this is the license used) are not the same as the rights granted by the ODbL. The ITO World lawyer suggests that the rights should may made to match.

This appears to be resolved with the DbCL.

The statement that 'Unspecified extra legal rights may apply' will make users nervous

There is a clause that says that unspecified additional legal rights may apply. How can a lawyer at a TV stations agree to a license with that waver. The ITO World lawyer recommends that the license should say that any restrictions should be contained in the notices attached to the database.

Fixed in the most recent ODbL draft.

Anonymous contributions

What will happen to anonymous contributions?

If they will not be allowed then sohuld be not ban them immediately? PeterIto 13:40, 10 March 2009 (UTC)

Anonymous contributions are not anonymous to the OSM Foundation, the user name just isn't public. --abunai 19:56, 10 March 2009 (UTC)

no action required

The "licensing Produced Works" problem

We have assumed that by relinquishing control of Produced Works - which under CC-BY-SA would have had to be licensed CC-BY-SA and nothing else, whereas we now basically say "anything you want" - we increase compatibility with other licenses. For example, with CC-BY-SA it was impossible to combine OpenStreetMap data and CGIAR altitude data into one image because CGIAR was "non-commercial" and the resulting image would have had to be under CC-BY-SA (which would have allowed commercial use).

This problem is mostly solved with ODbL (an OSM-CGIAR combination is now possible) but there is uncertainty about what happens if the "other" license is of the share-alike variant (e.g. CC-BY-SA or GFDL). These licenses do not permit the addition of any restrictions, but ODbL has the 4.7 "Reverse Engineering" clause which restricts the use of Produced Works by saying that if a Database is created from them, ODbL again applies. If OSM data is combined with something else that is CC-BY-SA, then the resulting product would have to be CC-BY-SA, and this does not allow us to add the "but if you reverse-engineer..." restriction.

Possible solution #1: Explicitly allow popular SA licenses

One possibility is to explicitly allow Produced Works to be under these licenses, thus potentially overriding the reverse-engineering clause:

Add to 4.7: "This provision shall not prevent Produced Works from being licensed under any of the following Share-Alike licenses: GNU Free Documentation License 1.2 or later; GNU General Public License 2.0 or later; any Creative Commons license with the "Share-Alike" license element 2.0 or later; or any license deemed compatible to the above by the relevant issuing authority."

This would theoretically enable someone to e.g. make a tile server and serve OSM in CC-BY-SA licensed tiles, then re-trace data from these and claim to be immune from the reverse engineering clause because this was not mandated by CC-BY-SA; but he'd end up with a CC-BY-SA'ed copy of the OSM database which would not hurt us and not do him a lot of good compared to the effort involved.

Possible solution #2: Clarify scope of reverse-engineering clause

It is not clear whether the reverse-engineering clause was ever meant to apply to reverse engineering some levels down (i.e. someone grabs OSM tiles from our server and traces off of them), or whether it was only meant to keep one person or organisation from masterminding a reverse engineering effort. If the latter is our main objective then 4.7 could be reworded to make it very clear that it only applies to one and the same person first creating the Produced Work and then reverse-engineering the database from it:

Change 4.7 to: "For the avoidance of doubt, if you create a Produced Work with the intent of later recreating the whole or a Substantial part of the Data found in this Database, a Derivative Database, or a Database that is part of a Collective Database from the Produced Work, then the resulting Database is still subject to this Licence. Any product of this type of reverse engineering activity (whether done by You or on Your behalf by a third party) is governed by this Licence."

Possible solution #3: Difference between copyright/database right/contract

CC-BY-SA explicitly restricts only the "copyrighted Work". ODbL however works also by contract and database right. It may actually not be against the terms of CC-BY-SA to restrict reverse engineering by contract or by database right, while remaining fully compliant with CC-BY-SA in copyright terms.

Are all solutions proposed above actually possible?

I'm not sure if we have much design flexibility here. While the wording can and should be clarified, I think Solution #2 must be the intended interpretation of the license as currently written, and for a good reason. Any stronger restrictions on reverse-engineering (e.g. making them apply to everyone) would be impractical, at least in jurisdictions that don't have database rights or sweat-of-the-brow copyright on facts. In such jurisdictions, the only way to impose ODbL conditions on produced works would be by contract law, and it would be impractical to ensure that all users of an image would be bound by contract. The stronger interpretation would risk preventing any derived work from being publicly distributed, as the onus would have to be on its author to ensure that all recipients would be bound by contract not to reverse-engineer. (Similar issues apply to the database as well, but since fewer people need to have access to the actual data, it's a little easier to ensure that everyone is bound by a contract. It might still be too hard, but that's another issue.)

The Reverse engineering clause has been removed in ODbL 1.0 RC 2.

The "Share-Alike for interim derivative database" problem

Description of problem

With ODbL 1.0 RC 1, Share-alike is triggered if you publicly use a derivative database, but it isn't if you use the derivative database internally to create a produced work that you use publicly.

Possible solution #1

Insert 4.3b: "The Database, Derived Database, or Database as part of a Collected Database on which you based the Produced Work has to be made available pursuant to section 4.6."

Possible solution #2

Drop 4.5b ("Using this Database, a Derivative Database, or this Database as part of a Collective Database to create a Produced Work does not create a Derivative Database") entirely. This would cause other problems with the maps generated from the database.

RC 2: A Derivative Database is Publicly Used and so must comply with Section 4.4. if a Produced Work created from the Derivative Database is Publicly Used.

Whose database is it?

Who is going to be the licensor of the origial complete OSM database (according to the ODbL license the OSM database has to be published by a organisation or person)? Will this be the OSM Foundation? If not then who?

OSMF is the de-facto licensor, as OSMF is hosting and making the database available.

Who is the licensor of derived databases? Who is the licensor when you merge two databases from different licensors that are available under ODbL or a compatible license?

The person or organisation which publishes the derived database is a licensor and may have developed some IP rights in the derived database. However, the original supplier(s) of the database(s) also retain IP rights, so there is no single licensor. It is prudent to contact all the licensors of all ODbL sources (or their proxies) to determine license compatibility with non-ODbL data, as this can avoid problems later in case the licensors or proxies declare that the licenses are not compatible.

Under what terms do people contribute to the project?

Does the individual user license their contributions under ODbL (a database(s) comprising the changesets they have committed), or under the Factual Information License, or under no license at all (just a contract between user and OSMF under which the user gives data to OSMF and allows then to license it under a specific license)?

The current proposal under consideration by the Licensing Working Group is that contributors will contribute data to the OSM database under the Database Contents License (DbCL) and that a contribution agreement (contract) between the OSMF and contributor will control OSMF's rights to sub-license the data.

What happens in places where there is no database directive?

Assuming that the "Factual Information License" forms the many relationship between the contributor and foundation (see above), and given that the factual information license effectively waves any copyright rights in the data then what happens in places where there is no database directive?

The factual information license (assuming this were to be used) waives copyright in the data, but does not affect copyright in the database, which is what the ODbL protects. ODbL uses a combination of database right, copyright and a contract to try and cover as many jurisdictions as possible. However, we should all be aware that it is not possible to write a license which applies equally well in all jurisdictions.

How will the contract be enforced

How with the contract be "signed" before people can use the data? A "click-wrap" license is not considered suitable because it would restrict people's ability to publish data via automated distribution channels.

This is a similar issue with most open source software licenses; there isn't a clickthrough, nothing is signed, and the applicable open source software license is simply posted on the site or in the files. There certainly are questions in the open source world about whether this constitutes a binding agreement for users of the open source software between the licensor and licensee, whether it is a unilateral contract, etc... However, a number of courts have enforced open source software licenses in important jurisdictions, including the US and Germany, as binding agreements. There are obviously best practices here, such as making the license (and its applicability) clear and conspicuous and easily accessible, that may help bolster the license's enforceability.

Assuming that KML/SVG/CVS files will also be considered to be databases (see Use Cases) then how will the contract clause work for these files. For example London Railways KML file and UK railways kml file.

The license, or at the very least a link to it, should be embedded in the file along with a notice that the data is ODbL licensed. As above, although there are questions about whether this is a binding agreement, the consensus in important jurisdictions seems to be that it is.

Should Share-Alike be required for derived works?

If you create beautiful painted map partially based on openstreetmap then under the ODbL license people will not be able to copy/use those images without permission. That seems wrong.

I have tighened the above text up, but don't believe that it belongs on this page. This is not a page to challenge the intiention of the license, only to question how it will actually work. I suggest this section is moved to the talk page. PeterIto 13:31, 10 March 2009 (UTC)

Under the latest draft of the license (1.0-RC2), it is not required that Produced Works be made available under any particular license.

How soon after the Produced Work is published must I make it available

While there is no explicit time limit in the license, you should make the new version available within a reasonable amount of time of the request for it.

In what data format can I provide it in, must it be in OSM format, or can it be a difficult to use proprietary binary format?

Covered by section 4.8a of the license. "This Licence does not allow You to impose ... any technological measures ... [that] have the effect or intent of restricting the ability of any person to exercise [rights granted under the ODbL]." -- In other words; it doesn't have to be in OSM format, but it can't be only available in a proprietary or encumbered format. See section 4.8b about making it available in both.

On what media I make it available on? Must it be a well know one?

The license requires that the copy be "in machine readable form" (section 4.6). Combining this with 4.8a, above, suggests that the format must not be intentionally obscure. For example; CDs, DVDs, Blu-rays and even DAT tapes are probably OK. Encoding it in notches on a bar of platinum, probably not OK.

Must I provide document on what I changed/added?

You may either provide the whole Derived Database, or a machine readable document describing the changes and additions (section 4.6b).

If I host it on the web then how long to I have to make it available for?

There is no explicit time limit in the license.

How much can I charge for providing physical media? Only the cost of the media and postage, or my time as well, if so at what hourly rate.

A "reasonable production cost". If the physical medium takes a significant amount of time to prepare then remuneration may be reasonable. It is suggested that you discuss this with the person who requested it, and come to an agreement about what is reasonable. Note that very time-intensive methods of distribution (such as scratching it into a platinum bar, above) may not be acceptable under the license.

Protected Derivative Databases

What is the situation if a navigation company creates a derivative database from OSM data, but this database is in some proprietary, possibly encrypted format that is not commonly readable? Copyright law protects the proprietary DB against reverse engineering.

does the licence require the database to be opened?

Only if a "parallel distribution" in an equivalent open format is not provided.

does it require a DB in an equivalent, commonly readable format?

If the derivative DB is provided in an open format then it may also be provided in a closed format.

What is the sitation if the data of the derivative database is available to the public in some way, but using it in the same way as the navigation company is not possible because the company holds a software patent on the process. Would it be possible to create derivative databases that are available, fulfilling the letter of the licence, but still unusable for the public, thus violating the idea?

The ODbL doesn't cover other IP rights, such as patents, so this would be possible.

Issues with imported data

Approval from large donated datasets

Are we going to contact the suppliers of large donated datasets to find their opinions on the new license? Or will the person who did the upload of their data just have to tick "I agree" on their behalf when they next log-in after the change?

Good point. That will certainly be on the OSMF Legal Working Group agenda to approach AND and other known suppliers if not done already (I am new in the Group). MikeCollinson 15:56, 28 February 2009 (UTC)

Who does the work? I uploaded some big datasets, but I'm not willing to do legal negotiations. Pavel 01:27, 4 March 2009 (UTC)

While it is better for the original importer to ask the permission of the data provider, as there is an existing connection, the Licensing Working Group and the new Data Import Working Group may be able to shoulder some of the effort of negotiation. Since it is not possible to distinguish normal user edits from imports, every user will have to tick "I agree". If you have imported data then you are implicitly agreeing to the license change. If you do not have the authority to do so, do not click "I agree".

Hmm. openstreetmap is in big part covered by imports, and companies provided data for imports under CC-by-SA. Having to negotiate imports once was bad, but having to do it second time is stupid... and 2 months do not seem like enough time.

So prepare to delete Czech Republic.
So prepare to delete Italian administrative borders, Friuli Venezia Giulia Official donated datas, as well as Merano and some other data we have just collected to merge in OSM with a CC-BY-SA licence! --EdoM (lets talk about it) 11:34, 5 March 2009 (UTC)

There are many areas which may need to be deleted if the ODbL is unacceptable to the data providers. We need to work together as a community to educate data providers about the many benefits of ODbL over CC-BY-SA. It is important to remember that, although CC-BY-SA and ODbL function in different ways, the intent and effect are very similar: It's an open, attributed, share-alike license.

Relicensing content derived from CC-BY-SA works

How will contributors who derived data from CC-BY-SA sources be able to relicense their work?

It is arguable whether CC-BY-SA applies to data, especially factual data. However, these contributors should get the permission of the original data provider to relicense it to ODbL.

Will a prolific contributor be able to identify those things that they derived from other CC-BY-SA licensed works?

Will they be required to delete this data from OSM before they assent to relicensing the rest of their work under ODbL?

Yes. The data should also be brought to the notice of LWG so that it may be removed from the history.

Most major contributors have probably, at some time or another, derived something from another CC-BY-SA work, so this is likely to impact a significant percentage of contributions to OSM.

Depending on the source of the derivation (aerial imagery, NPE, etc...) this may not be a problem. Derivation of large amounts of externally-provided CC-BY-SA data should be found and removed. If some is missed and later discovered, it should be reported to the Data Working Group.

Features touched by multiple contributors, not all of whom sign up to new terms

Consider the case where:

User A goes out and gathers a tracklog
User B traces that to form segments
User C creates a way from those segments and labels it "highway=primary"
User D adds a name for the road
User E goes through and adds more intermediate nodes, so that it fits the tracklog better
User F merges that way with another primary way (created by A', B', C', D', E')
User G splits the resulting way into primary and secondary at a different point.

If only some of the above agree to the new license then what happens? Please note that we are not able to detect merging/splitting of way from the database history. Also, what happens if someone collects a way and attaches it to be node on someone else's way.

OSMF counsel does not believe that CC-BY-SA data within the database is "viral" in this regard. The original data will have to be removed, plus any later versions of the same element, but it is not necessary to remove nearby or adjoining elements.

Problem with switching license?

Is it possible to change the license if one or more contributors do not agree to it? If the whole planet file is one big piece of work, licensed under the terms of the cc-by-sa, then all the data that is being changed or added to the database has to be licensed under cc-by-sa as well because it is a result from the earlier planet file.

Although the planet file is CC-BY-SA licensed, the primary source of data is the database. When contributors agree to the new license and the database becomes fully ODbL-licensed then planet files dumped from that database will also be ODbL licensed. Older planet files will remain CC-BY-SA.