Return to Answer

Firstly, Stack Exchange contributions are to share know-how. They are not to share intellectual property. Typically we are talking about small pieces of code - if not, why are you pasting them on Stack Exchange? Whether these are sufficiently large to constitute an independent work protectable by copyright that does not come under a 'fair use' provision is going to turn on the facts, and be a matter that can be debated by lawyers. So if I want to use a little code that's on Stack Exchange, the licence is of no help to me. As Ian Ringrose pointed out, does a call to create a new GUID require attribution or not? Under the existing CC-BY-SA licence, we don't know (because we don't know whether the one line of code is a 'work', and we don't know whether it is within a fair use extension), and under the proposed new licence (MIT with attribution) we also don't know. I'm guessing not for one line. If it's 100 lines, may be yes (but what's 100 lines of code doing there?). Five lines? You are not actually helping your readers.

Firstly, Stack Exchange contributions are to share know-how. They are not to share intellectual property. Typically we are talking about small pieces of code - if not, why are you pasting them on Stack Exchange? Whether these are sufficiently large to constitute an independent work protectable by copyright that does not come under a 'fair use' provision is going to turn on the facts, and be a matter that can be debated by lawyers. So if I want to use a little code that's on Stack Exchange, the licence is of no help to me. As Ian Ringrose pointed out, does a call to create a new GUID require attribution or not? Under the existing CC-BY-SA licence, we don't know (because we don't know whether the one line of code is a 'work', and we don't know whether it is within a fair use extension), and under the proposed new licence (MIT with attribution) we also don't know. I'm guessing not for one line. If it's 100 lines, may be yes (but what's 100 lines of code doing there?). Five lines? You are not actually helping your readers.

If you are going to fix it, fix it properly, not apply a band aid

The original proposal gave (almost) the right answer, but you went about it the wrong way and it was incomplete (for all the reasons set out in almost every answer). The right answer (as far as I'm concerned) is to MIT licence new posts (not MIT with attribution) as you suggested, but with a dual licence for CC-BY-SA, and to leave old posts as they are. Then ensure the licence is actually listed with the code.

Firstly, StackExchange contributions are to share knowhow. They are not to share intellectual property. Typically we are talking about small pieces of code - if not, why are you pasting them on StackExchange? Whether these are sufficiently large to constitute an independent work protectable by copyright that does not come under a 'fair use' provision is going to turn on the facts, and be a matter that can be debated by lawyers. So if I want to use a little code that's on StackExchange, the licence is of no help to me. As Ian Ringrose pointed out, does a call to create a new GUID require attribution or not? Under the existing CC-BY-SA licence, we don't know (because we don't know whether the one line of code is a 'work' and we don't know whether it is within a fair use extension), and under the proposed new licence (MIT with attribution) we also don't know. I'm guessing not for 1 line. If it's 100 lines, may be yes (but what's 100 lines of code doing there?). 5 lines? You are not actually helping your readers.

Let's not assume that this only impacts people writing proprietary software. Even an open source project using an SE contribution has just as many licence compatibility issues as under the previous licence (because that open source project is unlikely to be using your mangled MIT licence). You haven't helped.

And you aren't actually helping your contributors. If someone wants to create a new GUID, and I tell that person how to create a new GUID in the most efficient possible way (with a line of code), I want to do just that. I want to teach them, and to make their life better. I do not expect every time they use that technique to reference an article on SE to acknowledge that I was the guy who taught them how to create a new GUID (NB it was actually someone else). This is a qualitatively different action from publishing software on Github, where (incidentally) I get to choose any licence I like. I don't need the attribution, so don't give it to me. Just like I don't need it to be a GPL type licence (and you didn't give that to me).

The only conceivable result is those using short bits of code will be no more happy with MIT+attribution than they were with CC-BY-SA. IE they will either just ignore it, or be in a position where they are don't know what the right thing to do is, and possibly rewrite in order to disguise a copy. This helps no one.

So, if Alice copies Bob's code, Alice has to copy it with an attribution, but may omit the text of the licence licence. Charles then copies Bob's code, and no longer has to acknowledge the original author (Alice), because the license that requires it has been removed. But had Charles copied it directly from SE, Charles would have had to do something different. This makes very little sense.

Even better, unpleasant SE scraping site B can legitimately copy every post on SE, attributing them properly but not replicating the licence. Unpleasant SE scraping site C can then copy scraping site B without attribution. Yuck.

If you are going to fix it, fix it properly; do not apply a band aid

The original proposal gave (almost) the right answer, but you went about it the wrong way, and it was incomplete (for all the reasons set out in almost every answer). The right answer (as far as I'm concerned) is to MIT licence new posts (not MIT with attribution) as you suggested, but with a dual licence for CC-BY-SA, and to leave old posts as they are. Then ensure the licence is actually listed with the code.

Firstly, Stack Exchange contributions are to share know-how. They are not to share intellectual property. Typically we are talking about small pieces of code - if not, why are you pasting them on Stack Exchange? Whether these are sufficiently large to constitute an independent work protectable by copyright that does not come under a 'fair use' provision is going to turn on the facts, and be a matter that can be debated by lawyers. So if I want to use a little code that's on Stack Exchange, the licence is of no help to me. As Ian Ringrose pointed out, does a call to create a new GUID require attribution or not? Under the existing CC-BY-SA licence, we don't know (because we don't know whether the one line of code is a 'work', and we don't know whether it is within a fair use extension), and under the proposed new licence (MIT with attribution) we also don't know. I'm guessing not for one line. If it's 100 lines, may be yes (but what's 100 lines of code doing there?). Five lines? You are not actually helping your readers.

Let's not assume that this only impacts people writing proprietary software. Even an open source project using an Stack Exchange contribution has just as many licence compatibility issues as under the previous licence (because that open source project is unlikely to be using your mangled MIT licence). You haven't helped.

And you aren't actually helping your contributors. If someone wants to create a new GUID, and I tell that person how to create a new GUID in the most efficient possible way (with a line of code), I want to do just that. I want to teach them, and to make their life better. I do not expect every time they use that technique to reference an article on Stack Exchange to acknowledge that I was the guy who taught them how to create a new GUID (NB it was actually someone else). This is a qualitatively different action from publishing software on GitHub, where (incidentally) I get to choose any licence I like. I don't need the attribution, so don't give it to me. Just like I don't need it to be a GPL type licence (and you didn't give that to me).

The only conceivable result is those using short bits of code will be no more happy with MIT+attribution than they were with CC-BY-SA. I.e. they will either just ignore it, or be in a position where they don't know what the right thing to do is, and possibly rewrite in order to disguise a copy. This helps no one.

So, if Alice copies Bob's code, Alice has to copy it with an attribution, but may omit the text of the licence licence. Charles then copies Bob's code, and no longer has to acknowledge the original author (Alice), because the license that requires it has been removed. But had Charles copied it directly from Stack Exchange, Charles would have had to do something different. This makes very little sense.

Even better, unpleasant Stack Exchange scraping site B can legitimately copy every post on Stack Exchange, attributing them properly, but not replicating the licence. Unpleasant Stack Exchange scraping site C can then copy scraping site B without attribution. Yuck.

What you're now suggesting is MIT with attribution. Really, that doesn't help. MIT without attribution is not an OSI approved licence, and we don't need yet another licence around - licence proliferation is a bad thing. If you are going to require attribution, use an OSI approved licence that has an attribution requirement rather than making another one up.

What you're now suggesting is MIT with attribution. Really, that doesn't help. MIT with an exception that the licence does not have to be reproduced with attribution is not an OSI approved licence, and we don't need yet another licence around - licence proliferation is a bad thing. If you are going to require attribution, use an OSI approved licence that has an attribution requirement rather than making another one up.