Ignore his 30 watt drop. His power didn’t drop, he just changed testing protocols. So the issue isn’t that he now has a 260 watt FTP. The issue is that he wasn’t good at testing relative to his threshhold and above threshhold performance. Therefore, the “objective” testing may have difficulty pinpointing his level and all I’m saying is the “subjective” surveys will him guide the ML. He’s worried that if he tests and spits out an underperforming FTP, but then he superpasses hard workouts, that the ML will ramp him up in difficulty and he will eventually crack. I’m saying, his response to the survey will help calm the adaptive piece down. If it’s hard, he can say it’s hard, and it should back off.
And also, it will probably be able to tell he was struggling and back off on it’s own.