/ "Evolution is dumb enough to have gone ahead and created its usurper. Evolution really will go ahead and evolve itself to irrelevance. "
Not every possible set of genes or every possible consciousness can be expressed, but (barring annihilation) there will be some subset expressed. And there will be some historical path that got us there, and reasons why certain traits exist while others do not. So I fail to see how evolution can ever be irrelevant. Perhaps nerd/early adopter traits will be selected rather than the historical bigger/stronger/faster, but this is still evolution.
/ "By the time it figures out that the current state of it's cleverest creations isn't one where they optimise their response to future selective pressures it'll be too late"
I'm not sure I understand what you are saying here. Organisms have never optimized their response to future selective pressures, they have merely executed their current adaptations. If you are saying that, as a human, I can wear a condom to enjoy sex without reproduction, I agree. But this is not overcoming evolution. This is merely how I express the traits that evolution has given me (ie sexual desire + desire not to ruin life with infant.) I think one should be careful about anthropomorphizing evolution, as it leads to murky reasoning. Evolution is not clever or stupid, it merely is.
Inferential Distance Out Of Bounds Exception: Fundamental premises and expectations about probable future outcomes incompatible.
So I fail to see how evolution can ever be irrelevant.
Off the top of my head the options include
You are probably familiar with Hanlon’s Razor, the adage that you should never attribute to malice what can adequately be explained by stupidity. In Bayesian terms, stupidity is sufficiently abundant that even fairly strong evidence of harmful intent can’t overcome the base rate. However, there is something of a converse, which to my knowledge doesn’t have an eponymous name. In honor of a recent post by Mark Dominus, I propose Dominus’ Razor: Never attribute to complete stupidity what can adequately be explained by ordinary stupidity and a good reason.
Dominus, well-known as a Perl programmer, found that astonishingly bad code looks better (if still bad) after hearing the reasons for its development. For instance, one program passed data between functions by writing it to a temporary file, only to read it back again. It turns out the programmer did this for debugging purposes, an admirable goal, even if done in non-standard ways.
The Razor is one more explanation for the frequent failure of other-optimization. People and institutions usually have some reason for doing what they do, even if they’ve since forgotten or never knew in the first place. “Evolution is cleverer than you are” (Orgel’s Second Rule) and “Free markets are cleverer than you are” are two related rules of thumb. Something that looks obviously stupid was probably implemented to meet some non-obvious need or constraint.
In the end, this is another way of saying to not expect short inferential distances. Based on personal observation, this community does a good job anticipating inferential jumps when playing the role of the sender, but not quite as well when acting as the receiver. Even if someone is wrong, be careful not to dismiss them entirely.