[License-discuss] "Channelized" Open Source Licensing

[Florian Weimer]

* Kyle Mitchell:

> On 2018-10-20 09:10, Florian Weimer wrote:
>> * Peter Corless:
>> 
>> > There seems to be a lot of buzz these days about licenses in the face of
>> > cloud providers.
>> >
>> > I'd like to ask if anyone has considered, in this group, the concept of a
>> > 'channelized' license?
>> >
>> > Party A: An OSS developer.
>> > Party B: A cloud provider who hosting Party A's OSS, and is is charging
>> > Party C for this.
>> > Party C: A  user, using Party A's software, which is hosted on Party B's
>> > cloud.
>> >
>> > Under current licensing, the OSS license is between Party A, and Party B.
>> > Party B really isn't modifying or contributing to Party A's OSS code base.
>> >
>> > Party C, meanwhile, can do whatever they want to the OSS, since they have
>> > no legal license obligation back to Party A. Their access is provided
>> > through Party B. They could, theoretically, violate the license Party A
>> > distributed their software under, since they are just using it.
>>
>> By definition, OSS licenses do not have a field-of-use restriction, so
>> it is impossible to violate the license just by using the software
>> (unless the act of running the software creates some for of derivative
>> work).  Acts other than running the software typically require some
>> sort of license under copyright, and C can only get that on A's terms
>> (potentially as amended by B, but whether that's possible is really up
>> to A).
>
> The language of OSD 6:
>
>   The license must not restrict anyone from making use of
>   the program in a specific field of endeavor.
>
> The OSI has approved licenses that trigger copyleft
> conditions on kinds of use, alone (OSL) and with
> modifications (AGPL).

I think the expecation at the time the AGPL was created was that it
would be used only for software which has a built-in source code
replication mechanism, e.g., change part of the web site URL from
”.php” to ”.phps”, and you've got the source code in your browser.
Such technology did exist at the time.

I would be surprised if the intent was to support ”open core” business
models via the AGPL, where the software may have a networking
component (or be advertised as a tool for creating network services),
but there is no built-in source code distribution mechanism.  The
expectation behind this kind of use of the AGPL seemed to be that this
creates sufficient legal uncertainty that any commercial user would
obtain a non-AGPL license.  (Others have tried the same thing with the
GPL in the past, but creating such uncertainty there is very
difficult, particularly with the GPLv3 and its cloud computing
exception.)  People who have tried to use the AGPL in this way have
been disappointed about the effects, I believe.

I haven't thought much about the OSL.  Was it intended to foster “open
core” business models by its designers?