Knowing Isn't Enough
Say that you have a problem, and you know that you're doomed to fail.
- Knowing that you have a problem isn't enough.
- Knowing that knowing you have a problem isn't enough.
- Knowing that knowing that...all the way down isn't enough.
- Maybe at some point you'll find enough meta-justification that it spurs you to Go Do The Thing, but I doubt it'll be consistent.
- What does sorta work, is just straight-up doing it. Paying attention to what it feels like when you don't want to do it. Then doing it.
This feels weird because it runs contra to the advice of "listen to what you want and just do that". I want a lot of things. And I'm a self-feeding cycle, where things I do as output get put back as input. So stuff I do has an effect on future stuff I do. Consistency counts a lot because mostly it's a sliding window which controls what I feel excited by. Also, myths and stories. If I convince myself that X is sexy, I'll want to do more X.
When you fail, one correct response is to slowly work your way back up. You can do that with small successes, of doing the thing you want, on a smaller scale. In general, you want to be rewarding the correct behavior (ala shaping), but also just generally rewarding doing anything at all.
Modeling Progress
Progress isn't linear, and every day isn't a save point. But you're learning more stuff, even if you're not retaining all of it. Sliding window shows up again.
There's a disconnect between the most biologically accurate description and the most beneficial ontology for what's going on.
When thinking about growth, there might be some upward trends. But what do you care about? Rate of growth? The integral of all the stuff so far? Relative rate?
Probably some of them are true. But being specific might also mire you, if you choose something that you can't really optimize for.
I think I want to choose to maximize net output over time, or something like it. But I also care about how I feel on a day to day. I care about making sure my median feelings are fairly good.