Exactly; I suspect one of the reasons they used to have a lot of turbos fail was due to overspeeding them, as without any way to accurately control things or feedback on speed it seems easy to do. I have a little experience making microcontrollers do things, so with some forms of ECU, it might not be impractical to have a black box telling it how rich to go and controlling the antilag specific parts.
I'm seriously thinking I need to get a testbed engine to experiment with. I mean, why risk destroying good parts first? I don't really have the funding this year though... something about getting married.