I actually did 1, 2 and 4. But definitely not 6
Ooh, what did you do for 4?
I've got 1 covered, as my data dumps in the recently-finished RMM43 show. I've scraped a few games, but I haven't automated the construction of the initial data format I'm using, or all the possibilities for how a given mode handles start/end of day/game stuff, so I still end up needing long enough per game that I haven't gone very far back through the history yet.
For number 3 on sudgy's list (comparing a current game to stats on previous games), this is somewhere I wanted to go, but didn't feel like I had the data to make a convincing case. In RMM43, I wanted to know the probability for a given game size that an all-town-on-town wagon reaches 5 town votes... my hunch (which is just sort of my own internal stats counter having learned from looking at wagons by hand on the games I've played in) was that it was very improbable. Turns out in that game I was at least right about some of the people involved not being town (hi Haddock and faust!), but I didn't get around to building up enough of a database of previous games that I'd have felt confident in the stats. Plus, a given individual player's behaviour won't necessarily adhere to the average play. I'd still count this as "automated note-taking", since the inference of who might be scum when I notice a pattern like that is still all on me. Do others feel differently?