> He believes just three are necessary, AGPLv3, the LGPLv3, and Apache v2.
He being Bruce Perens.
I respectfully disagree, those licenses all have some negative component to them which make them not applicable in all cases. MIT/BSD do not have those same negatives (different of their own) and definitely are needed as well imho. Not saying those three aren’t important, just they shouldn’t be the only three.
Edit: clarified MIT / BSD are not without their own issues, just not the same as the other three.
MIT/BSD do have some legal drawbacks for certain use cases; see the Boost license as an example of a reaction to perceived negatives of MIT/BSD. In particular Boost does not require a copy of the license to be distributed with a binary. Have you ever shipped a binary containing MIT-licensed code, but didn't also explicitly ship a copy of the MIT license along with it? If so, you're technically in violation of the license.
> MIT/BSD do have some legal drawbacks for certain use cases
Sure I think the OP you're replying to isn't implying that MIT/BSD is insufficient. He/she is saying that the three "AGPLv3, the LGPLv3, and Apache v2" are insufficient but these five "AGPLv3 + LGPLv3 + Apache2 + MIT + BSD" would be more sufficient.
He being Bruce Perens.
I respectfully disagree, those licenses all have some negative component to them which make them not applicable in all cases. MIT/BSD do not have those same negatives (different of their own) and definitely are needed as well imho. Not saying those three aren’t important, just they shouldn’t be the only three.
Edit: clarified MIT / BSD are not without their own issues, just not the same as the other three.