Web service ToS follow-up

[ Rather than respond to comments to the previous post, I will create a new post — I think this is important enough to clarify ]

Declin said:

> I don’t see how putting a particular license on the *data* solves the problem. Terms of *service* are something entirely different.

We’re not putting a new license on our data — the data remains under the existing licenses.

I am suggesting that the ToS for our web service state that:

  1. Anyone can access the web service.
  2. Data returned by the web service is covered under the CC license, since we may return data where portions are PD and portions are CC. We must use the greater common denominator license if we want to be consistent with our data licenses.

This has the effect that anyone can connect to our web service, but using the data returned from the service has to be use in accordance with the CC license. In order to not be in copyright violation, the user needs to qualify as non-commercial set out by the CC license. The two clarifications on non-profit I give serve to make it clear that these uses fall into the non-commercial realm, since the CC license errs on the side of commercial use if a use falls into a grey zone.

Technorati Tags: ,

> Software being free or not has nothing to do with it being commercial or not!

We’re not talking about software — we’re not even talking about data (which you can download for FREE (as in speech)). We’re talking about providing a service. Each call to our service costs a finite amount of money (power, bandwidth, server cost amortized over its life) — granted its small, given that we’re doling out 30M web service calls a month. By applying the CC license to the data returned by our web service, we’re ensuring that people who use the service are doing so in a non-commercial manner. Otherwise, your donations to the MetaBrainz Foundation go towards supporting other people making money. I don’t think that is right.

> If you want to block other hosts from abusing the web services, or charge a fee to heavy users so you can add the required capacity, just do so. I don’t see what this has to do with the database, or whether these third parties are making money.

This doesn’t have anything to do with the database, as stated above. Rod explained this quite succinctly in his recent post to the musicbrainz-experts list.

However, charging a fee for heavy users is far from trivial. Have you thought about:

  1. Purchasing the machines to do this
  2. Finding hosting for the machines
  3. Administering the machines
  4. Tracking and billing heavy users

Until we have more money to hire (more) staff, this is not plausible. Unless you wish me to spend the next year building this system and neglecting the overall features of MusicBrainz. By going with a CC license as part of the ToS we can off-load this burden to our commercial customers.

Jamie wrote:

> I’m not sure that saying “Applications that charge a registration or shareware fee and that donate at least 10% of their fees to the MetaBrainz Foundation are considered non-commercial” makes any sense.

It does if you consider the exact text of the CC license. As I mentioned above, the CC license errs on the side of considering things in the grey zone to be commercial. The Creative Commons (and this is according to our Director Emeritus Larry Lessig) says that the licensor should refine this grey zone and show which cases are non-commercial and which are commercial. The two clauses I gave do just that.

> Also, you miss the case of a non-osi “free as in beer” product. Do they have to give 10% of nothing to qualify?

Ohhh — that is an excellent point. I don’t have a good answer for this — any suggestions?

Rod suggested:

> The one thing I would suggest is a requirement that a UserAgent field be set by calling applications so that usage can be monitored. It’s by no means a fool-proof method, but it’s a lightweight change, and at least gives you a fighting chance of a) understanding the impact on the server by different software, and b) blocking out bad neighbours if there’s a problem.

Ohh. Another good suggestion, though that will require some libmusicbrainz changes to enable this. I think also limiting the number of calls per second per client the way Amazon does it is also good. I’ll add these two clauses.

Alex said:

> You might want to consider a specific licensing fee (not just as a threshold percentage of registration) for commercial applications that use the web service (as I understand it, currently the only commercial licenses are for the replication database feed).

Yes, but this can happen outside of the ToS for general use. If a small commercial entity is going to be an insignificant amount of load on our servers, we can grant them a separate license to use our web service. This ToS does not preclude that from happening — we could in effect dual license the web service.

So, the revamped proposed ToS then reads:

  • The data returned by the web service is covered under the Creative Commons Attribution-NonCommercial 2.5 License.
  • Applications that make their source code available under an OSI approved open source license are considered non-commercial.
  • Applications that charge a registration or shareware fee and that donate at least 10% of their fees to the MetaBrainz Foundation are considered non-commercial. (e.g. Jaikoz and Magic-Tagger)
  • Applications must set the User-Agent: header with the name and version of the application making the web service call.
  • Each application instance may not make more than one web service call per second.

The “free as in beer” case still needs to be addressed.

4 thoughts on “Web service ToS follow-up

  1. Decklin Foster

    Could you at least spell my name right?

    I still don’t see what this solves, but whatever. It certainly doesn’t make me feel particularly happy about adding anything MB-related to Debian, or about what’s already there. We are not, if you haven’t noticed, a distribution *for* non-profit use exclusively.

  2. Joe Wreschnig

    Really, this doesn’t make sense. The ToS you propose makes proprietary (even freeware) applications for personal use not okay, but someone building a for-profit service on GPLd software okay (as long as they abide by CC-nc for the data). That really doesn’t seem to be what you want.

    That the data is CC-nc has nothing to do with the program accessing the web server, but what the program then does with that data.

  3. Anonymous

    I have several issues with the proposed ToS:

    1) Why is it that you can’t just put the burden on the user of the service to determine the usage of their data. I would suggest something more like the following: “The service gives access both to the public domain data and to the CC-by-nc data, so the safest assumption is that the data you fetch via the service is CC-by-nc; if you want public domain data only, you’ll have to do the work yourself to ensure you’re only fetching public domain data.”

    2) I feel that you’ve allowed two entirely different reasons to dictate the license on the data retrieved by the service. One is the perfectly reasonable statement that you don’t distinguish between types of data on the service; for that, see point 1: that’s the job of anyone using your service who cares what license the data is under, and nothing says you have to make their job easy, but don’t make it impossible. Two is the less reasonable voice of the people who don’t like the public domain status of the data as it is, want everything non-commercial, and see this as another shot at doing so; I’ve observed such sentiments frequently in the MusicBrainz community before, and I suspect this served as a place where they could vent and find an alternate battleground.

    3) I was under the impression that you *already* distinguished between commercial and non-commercial sources, by making the live data feed (for people wanting all the data, for running an alternate server) is CC-by-nc. That seems quite sufficient to ensure that nobody makes money off of an alternate *server*.

    4) It seems better, to me, to distinguish solely between the for-profit status of applications, rather than the data, just as you’re doing with the remainder of the ToS apart from the first item. That way, you’re ensuring that anyone who wants to sell a proprietary program needs to either pay you to use your server or set up their own server. The only other case in which someone might try to profit off of your service is if they set up a service which uses your web service to make money off of people; however, your prohibition on more than one call per second makes that all but impossible, and you could therefore charge for the ability to make calls more often.

    5) Open Source in no way means “non-commercial”; regardless of anything else, please *don’t* phrase the ToS in such a way as to conflate the two.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s