Daniel Mayer wrote:

And the Linux volunteer free software coders are harmed by GNU/Linux distributors?

Depends on who you ask. :-)

The Linux kernel has its own stable/unstable configuration and so do the distributors. Both work together to create a better product.

Ideally yes, in practice there are more than a few ugly disputes. But it makes a good example to learn from. For instance, people have tried to avoid forking in the past, but it doesn't work, so the usual process nowadays is to set up a deliberate fork (aka release branch), make a high bar for patches to the branch, and require programmers to patch both branch and trunk, possibly having to rewrite the branch patch to be applicable to the the trunk if the trunk has changed a lot.

If the analogy is applicable, it suggests that the tagging concept won't work well, because small good edits often follow on the heels of big bad edits in Wikipedia, and so there's no version with only good edits. But if you make small good edits in a separate Nupedia, they won't find their way back to Wikipedia, which might or might not be desirable - if Nupedia is the "release version", then it's likely to be out-of-date relative to Wikipedia, and the patched article(s) may no longer exist. If the Nupedia edits are still valuable, then perhaps they could be expressed as a patch file and the patch applied to Wikipedia a la source code patching.

The process becomes more complicated for Nupedia, but presumably that's not a serious obstacle because the whole theory of Nupedia is to improve quality via additional process.

Stan