Conversation

eg: Streamers make mistakes. They often view this as embarrassing, apologizing to viewers, but it's valuable to see them think through sol'ns and talk out loud about how they got into a mess. By contrast "screencasts" are often edited to be perfect, sometimes to their detriment.

Another category they highlight: insightful tangents. When you're working on something with a mentor, they might realize in the middle of a task that it's a great time to stop and explain some underlying idea. I think this context is often more natural than explicit instruction.

The authors also point out an important problem: often streamers really can't verbalize what they're doing and why. It's instinct; it's contingent; it resists routinization; "it comes with practice"; etc. Tacit knowledge is a problem with "real" cognitive apprenticeship too, ofc.

Are the troubles of tacit knowledge exacerbated in the streaming format, relative to real apprenticeship? One obvious difference is hi-fi interactivity. When I worked with more experienced designers, I'd pepper them with questions. Sometimes that'd make their instincts explicit.

It can be hard to create enough context in these streams. A designer may have spent tens of hours interviewing users, and they're now using all that context to guide their search. You can present your "user research synthesis" as context to viewers, but it's gonna be weaker.