I'm building a new vocabulary for myself to describe the experiences I have when reading some work, be it human or LLM, or trying to generate new ideas or resolve questions around attention, taste, and writing. Constructing this vocabulary at the right level of abstraction gives me an anchor to put new experiences on. I think of it like a big velcro wall that I could stick index cards to. They won't fall down randomly (like post-its) but I could put them on a new section of the wall easily, plus it makes a satisfying rip when I rearrange them.

Evaluating your own experience is key in this. There's a lot of squeamishness in doing this because we've followed a lot of dead (and deadly) ends being a little too trusting and attached to our own thoughts. A grand theory seems intractable, but taking a functional approach will be fruitful, and then we can build better stuff.

The first idea is overlay. It's the experience you have when a thought-image flashes against something you're reading. This materially changes the context and thus the meaning of the words on the page. The best conditions for this are being in the zone reading at a moderate to fast pace - just before you start skim-reading. It's the LISS of reading. Maybe these kinds of experiences can be slightly more reliably induced, by thinking along more metaphorical and associative dimensions of a given text.

The second is path. Short paths produce stale outcomes. More concretely, in a new search space, using a strategy which is just one or two actions or turns away from the objective produces content which is too tinged by the strategy. If you are on the producing content side, you need to start further back, the right number of turns away. Too far though and the objective will be ever elusive or the hit-rate super low. On the consuming side, in a super content-saturated environment, you'll pay attention to the content that you have just enough context for but the content's persuasive or other aims remain implicit (not short path), or at least implicit enough. This path and turns framework gives you, as a producer, enough structure to generate a variety of approaches and develop a bigger search surface-area while preventing a scatter-shot approach (too long of a path).

The third is trueness (maybe there's a better word for this). It's just rebranded authenticity but I feel there's squeamishness around the term since it can be boiled down to cliches and has vague self-help undertones, and might be dismissed as a diversion because it's too first-person. The obvious case is that it can be useful to adhere to because it allows you to produce more complex, inimitable work, which people consuming are pleading for while railing against slop. By trueness, I mean a heuristic that you use to filter information based on some kind of 'disciplined subjectivity'. It is complementary to path because this filtering process naturally puts you at the right path length from the objective when done correctly.