Tuesday: I went to several AI panels; they were fun, but mostly not related to what I write about here, except by loose analogy. (When AI goes wrong, it goes really really wrong, though.)
To me, the most compelling moment was the slide of the spreadsheet he used to track word use and figure out which elements needed to be changed. I could really have gone for a 20-30 minute deep dive just into that aspect of the writing. (Hypothesis: because Her Story was successful and looks simple, there will probably be ripoffs, and they will probably be terrible. There is a lot of invisible craft that goes into the word choice to make the game function as it does, and someone careless about that issue could get it very wrong.)
Alexis Kennedy talked about choice, consequence, and complicity, drawing on a lot of lessons from Fallen London; in particular, some observations about how number of branches may affect the player’s perception of a choice. (Two branches are dramatic and elegant; three allows for asymmetry; five allows you enough room to conditionally lock a couple branches and show the player the results of earlier choices.)
Also, here’s Gamasutra on lessons from Tarn Adams and Tanya Short, which I include because I wanted (but failed) to get into their Monday talk on procedural generation of mythology.
Wednesday I attended fewer talks, though I did catch Jane Ng on the art of Firewatch — which drove home how much of the space design is driven by the need to have discrete map sections to load and unload, rather than long vistas across the entire map. At the same time, a lot of that design was fairly story-focused. Landmark locations have narrative payoff, which corresponds to how people think about remembering travel. You remember moving from one interesting spot to another; so the interesting spots are both physically cool and also significant to the story.