At the same time I'm a big critic of RMS and his work (from GCC to the GPL, passing by the Free Software Song ;) but he is right that software patents are a very harmful monster.

    1) The Patent Office gave proper attention to prior art
    2) The Patent Office implementation of "obvious" was sane?

For a more recent example, consider NP completeness, proved completely independently by Stephen Cook and Leonid Levin.

Or even reinventions made disparate by time: the Cooley-Tukey FFT was first formulated by Gauss in 1805.

If these inventions are "obvious," then all inventions are obvious. The truth, not widely acknowledged, is that most inventions are as much a product of the environment of the inventor as the inventor themselves. They're not smarter and they're not unique. Patents should not prohibit coincidental parallel invention, only reverse engineering.

Well put. That's exactly the intuition I was trying to express.

... how would that work?

So to win in your case, you'd have to tread a tightrope between "not enough evidence to show independent origin" and "the thing is obviously substantive enough to warrant a patent."

Most likely, the thing actually is obvious and there would be other prior art, which the Patent Office currently doesn't pay attention to.

You've chosen difficult wording here. When you ask if things would be so bad if we lessened their badness the answer is obviously no -- things would be incrementally better if made incrementally better. But this is no more than a tautology. Software patents are bad, in any form. They can be made not-quite-so-bad, but they will always be harmful to the intellectual development of society.

    3) Software patents expire after three years instead of twenty

I wonder what his opinion is now.

Whatever it is, its going to be considered. But it doesn't mean it hasn't evolved.

This is the only part of that post written in 2006:

"I can't find this on the web, so here follows a note I wrote in 1991 of an odd event in my computing career."

Everything after that is from 1991. You have no basis for interpreting the 2006 posting of this 1991 note as anything, save preservation of history.

You could imagine that if he felt differently in 2006 he'd have mentioned it in the preamble?

He posts one or two entries a year in that blog; this note was not some noise he dumped, it was a deliberate posting.

If you actually had any interest in truth, you'd ask him. Instead, you've decided you know what his views and motives are based on no evidence whatsoever, and are asserting your opinion publicly.