Re:Post-Open Source
by JoshTriplett on Wednesday September 09, 2020 @04:19AM
(#24418657)
The article seems to gloss right past Free Software and copyleft licenses based on the claim that some copyleft licenses like GPLv3 aren't widespread because companies don't like and won't use them, then proceeds to opine on what license terms best annoy companies (profanity, ambiguity, lack of licensing, etc). And yet there's no consideration given to actually using copyleft licenses like GPLv3 or GPLv2, just the repeated assertion of the death of "Free Software".
If a copyleft license is sufficiently strong that some people don't want to use it under those terms, some subset of those people may be willing to pay for an alternate license. (The set of people who wouldn't pay for an alternative to a copyleft license seem even less likely to consider any completely non-standard incompatible-with-the-universe license.) If you want to use a license that may potentially lose you a few over-cautious corporate users, but which protects a community of users and the possibility of revenue, a copyleft license like GPLv3 or GPLv2 may make sense.
Many of the people using permissive licenses don't work for companies, or otherwise don't have anything that might block them from using a copyleft license. Many people seem to worry about using a copyleft license because it might mean some people can't or won't use the software. I've even seen companies who would actually benefit from a copyleft license just not-think about the possibility of using one because they have a vague impression that Apache or MIT is what they should use. It wasn't that they considered and dismissed copyleft; they just didn't even think about the possibility and weigh it as a potential option.
(Also, last time I saw a survey of licenses on popular software repositories, GPLv3 was about as popular as GPLv2. LLVM vs GCC is a relevant and memorable problem, but those two projects hardly represent the whole landscape of software licensing. LLVM also gained popularity for reasons beyond just licensing, such as recognizing the value of extensibility and providing compiler framework libraries, much sooner than GCC did.)
I'm not going to say "use copyleft for everything!", and there are good reasons to use permissive licenses (such as propagating standards or releasing tiny projects). But next time you're releasing a project, give it some actual consideration. If you want to release something under a permissive license, and that's a deliberate decision, by all means do so; but don't just do so "by default" without thinking about it, and without thinking whether the permissive-by-default culture is actually in your best interests.
If a copyleft license is sufficiently strong that some people don't want to use it under those terms, some subset of those people may be willing to pay for an alternate license. (The set of people who wouldn't pay for an alternative to a copyleft license seem even less likely to consider any completely non-standard incompatible-with-the-universe license.) If you want to use a license that may potentially lose you a few over-cautious corporate users, but which protects a community of users and the possibility of revenue, a copyleft license like GPLv3 or GPLv2 may make sense.
Many of the people using permissive licenses don't work for companies, or otherwise don't have anything that might block them from using a copyleft license. Many people seem to worry about using a copyleft license because it might mean some people can't or won't use the software. I've even seen companies who would actually benefit from a copyleft license just not-think about the possibility of using one because they have a vague impression that Apache or MIT is what they should use. It wasn't that they considered and dismissed copyleft; they just didn't even think about the possibility and weigh it as a potential option.
(Also, last time I saw a survey of licenses on popular software repositories, GPLv3 was about as popular as GPLv2. LLVM vs GCC is a relevant and memorable problem, but those two projects hardly represent the whole landscape of software licensing. LLVM also gained popularity for reasons beyond just licensing, such as recognizing the value of extensibility and providing compiler framework libraries, much sooner than GCC did.)
I'm not going to say "use copyleft for everything!", and there are good reasons to use permissive licenses (such as propagating standards or releasing tiny projects). But next time you're releasing a project, give it some actual consideration. If you want to release something under a permissive license, and that's a deliberate decision, by all means do so; but don't just do so "by default" without thinking about it, and without thinking whether the permissive-by-default culture is actually in your best interests.