Again, I fail in my attempts at humor. It's a good thing that I don't try to make my living in comedy! Sorry.
My (successful) purpose was to provoke discussion within the Foresight development community. (It's been a busy day for me.) My concern with it as a front page LWN article is that as a front page article it loses context, which can give misleading impressions. That's hard to avoid.
Perhaps I should have started out by describing it as a thought experiment, but most of the time when I see that done it ends up leading to bikeshedding. I suppose I could have used the subject line "A Modest Proposal" but I'd expect that to have the same result.
Stepping back, though, away from Foresight -- my modest proposal was certainly possible with Conary; in fact, Conary makes it sane to maintain. The question is not whether it's possible (it is), but whether it is a good match for Foresight's goals. Completely independent of Foresight, I'm still interested in seeing more operating systems packaged with Conary, including but not limited to Fedora. Building them completely from source has some advantages (full application of Conary policy and build rules enhancing consistency, for example); a binary import has other advantages (expediency, consistency with upstream).
Hybrids are even possible -- start with a binary import, then package by package move to building from source where it's useful. Frankly, if any existing Linux distro considers moving to Conary, that's how I'd suggest they start -- working in parallel with existing systems at least to start. For example, Conary not only can import RPM and dpkg, but it can also export tarballs, which can then be used to build RPM and dpkg packages, and that could be enhanced to export more directly to make more use of Conary metadata (for example, automatically creating appropriate install/uninstall scripts instead of having to maintain them explicitly in spec files or debian source packages). I suspect that it's hard to understand why you would want to do that without experiencing Conary, so a binary import is probably the lowest-friction way to get to an apples-apples comparison.