Nice thinking, Noughtpixy. I've been trying to think of ways to encourage a,b,c mutations. Using a secrocyte is a nice easy way to let all of them change without major detriment.
I got a "b" value mutation in a kite swimmer by making one of its flagellos at a different angle. So the swimmer is asymmetrical to begin with, and mutation could make it swim in a more balanced way.
Substrate, after mutation and stabilization to one genome: https://www.dropbox.com/s/eq380z1ikvfac ... trate?dl=0
Left flagello is pointed straight back, while right flagello is at 45-degree angle. Both began with a=1.152 and b=0.171. It took a long time mutating at low radiation (0.002), but the left flagello finally mutated to b=0.06. This gives slightly better forward movement and turning.
A mutation that straightens out the angled flagello could also restore balance to this swimmer. That didn't happen this time, but I see no reason it couldn't. A mutation to the "a" multiplier of one side could also improve balance, but that also didn't happen here.
I did see a temporarily successful mutant with longer adhesin on the left side, allowing it to stretch and turn right more effectively. This mutation remained in vestigial form in many descendants, but the final version with b=0.06 does not have the long adhesin mutation, so either it was unnecessary and mutated away, or the final swimmer is not one of those descendants.
So I'm fairly convinced now that this whole thread is a misnomer; swim speed mutation is certainly possible, but is tricky, hard to predict, and may take longer than expected. Isn't that evolution in a nutshell?
But it has been a good learning experience! Hopefully valuable to others too.