This is about being a Free Software license, which is a much more basic set of conditions than the GPL, e.g., BSD/MIT falls under it too, as well as a host of other licenses (look it up). It is harmful in the sense that the success of free software depends on the licenses allowing unfettered distribution, while this license prevents that.

It's a license that's misleadingly almost-free. It's like the old 4-term BSD license with the "obnoxious advertising clause".

> It's a license that's misleadingly almost-free.

Like the GPL is misleadingly almost free? As in it's free but if you release software which uses GPL software, it is under GPL? I am not arguing about the meaning of free, I am arguing against your "it's a license that's misleadingly almost free". Just like GPL's terms doesn't make it non-free, "don't use it for evil" doesn't make it non-free.

EDIT: I see people arguing about "don't use it for evil" violates "can use it for any purpose" clause. Fine. GPL's clause violates "use it to distribute software without making my whole software GPL."

On the terminology side: language is about communication, and there is a broad consensus on the meaning of "free" in terms of free software, exemplarized by the FSF's defnition, the OSI open source definition, and the DFSG. None of these definitions is identical to the others, and no doubt there are edge cases that would fall under one but not another, but neither the GPL nor Crockford's license is one of them; under this consensus meaning of "free" (which you claim not to be arguing about), the GPL is a free license and Crockford's is not.

On the practical side: the GPL is established and popular enough that you should know what you're getting yourself in for with GPL code. In theory I can imagine someone unfamiliar with the free software movement hearing that a piece of GPL code was "free" and assuming that meant they could use it in their closed products, but in practice when people talk about "free" software they tend to either mean the free software definition, or "freeware" (i.e. binaries which can be downloaded without charge) - which generally doesn't permit modification and redistribution. So I don't think this is a practical concern.

By contrast Crockford's license is not popular or established. Worse, it looks exceedingly similar to a popular, established license - the MIT license - which is free (in the free software sense). So it's very possible to accidentally download a Crockford-licensed piece of code, thinking that you can use it in a free software project, when in fact the license does not grant all the rights you would need to do this - as is demonstrated by the fact that the author of this post did exactly that.

Crockford is free to distribute his code under whatever license he likes, but using a license that looks very much like the MIT license but grants you a much smaller set of rights (a set which puts it on the other side of a line that, while in many ways arbitrary, is of great practical relevance when it comes reusing code in the wider internet community) is not nice.

If it were really established that in software free means "Free" the way FSF defines it, then we would not have this 100+ messages discussion here arguing what is free, and in which context.

Maybe if would be easier if instead of "free" we started using more precise terms like Free(c)FSF, $Free, or name actual licenses, since each one goes under some assumptions.

My position depends on "free" being a broadly-applicable concept. If I'd said "Crockford's license is not DFSG-compatible" then the response would (quite rightly) have been "well, who cares, the DFSG is just Debian's set of arbitrary rules". The reason the line between free and non-free matters so much from a pragmatic perspective is that there is this consensus, the criteria for being included in Debian are (more-or-less) the same as those for getting free hosting on sourceforge, or for being able to use the OSI trademark, or...

> but using a license that looks very much like the MIT license but grants you a much smaller set of rights (a set which puts it on the other side of a line that, while in many ways arbitrary, is of great practical relevance when it comes reusing code in the wider internet community) is not nice.

So what is he supposed to do? His options are either use the MIT license unchanged, or invent his own? He liked the MIT license, and he wanted to add one of his own conditions. It's as simple as that.

As another commenter pointed out, the cognitive dissonance is the free software community is mind blowing. We are all for freedom and choice, as long as you are playing by our rules. We are against regulation, but only bad regulations. Ours are meant to do good and you are supposed to follow them.

While the right to fork is fundamental to the free software philosophy, on a social level we emphasise that it should be a last resort (and that it's often worth putting up with a less than ideal upstream in the interests of compatibility). E.g. the FSF recommends using the perl license for perl packages, even though it recommends against that license in general. This is nothing new.

What Crockford's done is the equivalent of changing a few bytes in the PNG format and making his software use this "new" format. Making a completely new file format/license is fine, but you should only do it when your improvements are big enough to justify the incompatibility. A new one that's just a trivial tweak to an existing one, but incompatible, is bad.

You can use GPL software to distribute proprietary software. Google does it all day long.

Don't write "use" when you mean something else entirely.

"Free" for the FSF does not mean "compatible with GPL".

Lots of GPL in-compatible licences are "Free" according to the FSF. Here's a whole list of "GPL-Incompatible Free Software Licenses": http://www.gnu.org/licenses/license-list.html#GPLIncompatibl... Some notable ones: Apache Licence, BSD licence, etc.

The BSD licence is not GPL-incompatible. It is the original BSD licence which is GPL-incompatible, because it has an advertising clause requiring all advertising material for a product to include a note for each contributor under that license.

He complains about the license not being DFSG-compatible, which is different. The DSFG are a set of very simple rules to decide if some piece of software is eligible for inclusion in Debian.