Added a few more pages of book now that I've finished the analysis of last weekend's experiment. If you're working on an IKAWA Pro V3 sample roaster, I can now pretty confidently say that you should draft your control plans based on exhaust temperature which within similar control strategies correlate better with significant roasting events, but then export that data to your favorite spreadsheet and design a simpler inlet temperature control plan.
You'll get less coffee to coffee variation on interior degree of roast and less astringency distracting from what's really available in the cup.
I suspect that even better results would be available if the control algorithm were changed to operate on cubic splines instead of linear splines when matching exhaust temperatures, but since I'm not about to try reverse engineering a firmware upgrade for the machine to create that capability the closest I could do to test that would be using a stupidly large number of control points input through a phone app that's very clearly not designed for plans that detailed.
