I've always thought that the GPLv3 effort was one of the FSF's biggest mistakes. It split the GPL community into GPLv2 and GPLv3, since the two aren't compatible. And the GPLv3 was simply far too scary for companies.
It's clear that the Apache license is more friendly toward corporate concerns. But companies were willing to deal with GPLv2, because there were sufficiently large projects (such as the Linux kernel) that added enough value that it overcame their fears over how the GPLv2 restricted how they could monetize their software engineering investments.
Unfortunately, the GPLv3 and even more so, the AGPLv3, was simply a step too far; put simply, in my opinion, the FSF overplayed their hand. Whether a more scaled back GPLv3, or simply sticking with GPLv2, with all of its admitted ambiguities and admitted legal short-comings, would have made enough of a difference at the end of the day is impossible to know. But I think copyleft as a viable open source licensing regime, and the GPL in particular, would have had a much better chance of success if GPLv3 never happened.
It's clear that the Apache license is more friendly toward corporate concerns. But companies were willing to deal with GPLv2, because there were sufficiently large projects (such as the Linux kernel) that added enough value that it overcame their fears over how the GPLv2 restricted how they could monetize their software engineering investments.
Unfortunately, the GPLv3 and even more so, the AGPLv3, was simply a step too far; put simply, in my opinion, the FSF overplayed their hand. Whether a more scaled back GPLv3, or simply sticking with GPLv2, with all of its admitted ambiguities and admitted legal short-comings, would have made enough of a difference at the end of the day is impossible to know. But I think copyleft as a viable open source licensing regime, and the GPL in particular, would have had a much better chance of success if GPLv3 never happened.