Good points, and KAT seems to be a good example of this. I looked at some of his individual defensive metrics and they have stayed pretty consistent, but by being with bad vs good defensive players the perception of his defense has changed.
I don't think it's impossible to measure defensive impact, just very difficult. I still trust the eye test in combination with consistency across various defensive metrics.
'Eye test' in Lauri's return vs Minnesota told me that he played defense very well. Surprisingly well it seemed to me, so having this thread pop up pretty soon after that game is a surprise also (if it weren't for the turn over record, he played very well overall in that game). Still... he is no elite defender, but with hustle and IQ he can be better than average.
Better last year's stats vs this year's stats may have a lot to do with the switch from a lot of minutes with Kessler to a lot of minutes with J.Collins... Last year the team was better overall though I'd say and defense - if anything - is a team effort.