I concede some of your points, but hope I can get some across as well.

I agree that when producing free code, one fervently desires its most extreme adoption possible. Alas, opportunity costs rule any material reality. The ‘materials’ in question being the effort of fellow programmers and the mindshare of the user base. (This is more generally true of any intellectual pursuit.)

IP regimes thus compete in a Red Queen’s race which is particularly intense in the software ‘industry’. The most permissive licences, which intended to maximise user mindshare, were largely superseded by hereditarily free ones (such as the GPL family) because they were so open to defection. Now, a lack of users tends to a lack of coders, so both extremes (in the users vs. coders ratio) lead to the same death. Success should be found somewhere around Gotama’s highly vaunted middle ground.

Is this middle ground nearest the GPL family, or BSDward of it (as your comment would imply). Statistics seem to place it much closer to the GPL than the Affero family, but this could be misleading, as the Afferos might be younger or less publicised.

Of course, the licence space is not linear. All the free licences seek to cooperate amongst themselves by sharing code, by presenting a united front against proprietary ones, while ultimetely competing against each other. There might be other unexplored options, but momentarily disregarding the unknown, I suspect that the sweet spot might lie somewhere between the best free licence and the best proprietary one. This magical licence of my intuition would combine the proven advantage of enforced reciprocity with the obvious benefits conferred by a steady flow of profits.

This ‘perfect’ licence I speak of would be best only in an evolutionary sense. My bleeding heart would obviate the advantage of profits, or at least their inherence in any degree of proprietariness, or indeed deny the legitimacy of ‘proprietas’ in general. Selective pressures, of course, could not care less for our sensibilities, and I suspect that certain large and small software firms owe their continued success to something other than mere inertia. Witness the remarkable success of the closed iphone silo.

Please do write your further thoughts on this matter.

The whole notion of counting the number of contributions to the Linux kernel is dead wrong. Open Source code is written to be *used*.

And everybody’s favorite punching bag on this is Google, yet those same people aren’t stopping to look at the massive contributions of Android and Chrome. Or the hundreds of smaller projects Google has published and hosted on Google Code.

Writing Open Source code should not oblige your users to do anything. If you want that, then go back to a proprietary licensing model.

(I wrote about "Giving Back" a while ago: http://prng.blogspot.com/2009/02/giving-back.html)

The open source movement, on the other hand, does care if a few players are able to build a lucrative business without contributing in the proportion you would expect a traditional software vendor to.

So, I would hold that people that care about the latter problem should not necessarily look to the Free Software philosophy for guidance.

Secondly, I think that your argument might be circular, just after “let’s assume”. Lack of AGPL adoption is not a problem because of the lack of AGPL adoption? The GPL remains the most widely known, understood and respected license at least in part because this debate hasn’t played out yet. I am sure many projects will choose the GPL out of familiarity and that the “market” for licenses may be in a process of correction. It is opinion-formers like yourself who are the most likely to accelerate such a change, I suspect.