Try things: for a day / week / fortnight, only use your computer for making things (yes: writing, figuring out a specific problem, programming, composing, animating; no: playing games, tv, reading, pornography). Switch it around: for a day / week / fortnight, never use your computer for making things.
Pre- and post-reduction holism. In thought and conversation, sometimes holism and reductionism/specificity collide: one wants to use concept X as it already exists / in its high level form, the other wants to articulate aspects/examples/structure/implications of X. In the dance of trust, which one to rely on: the one saying they know what's relevant in this context, or the one bidding to pay the opportunity cost to elaborate X? We can tug the rope sideways by distinguishing two things, and conditioning action on perception of those things: one is pre-reduction holism, and the other is post-reduction holism.
Pre-reduction holism takes X as-already-given to perception. Like if X is a ship (as in sailing), pre-reduction X is your intuitive high level picture of ships, if you don't think in too much detail; ships float, you can put stuff and people on them, wind makes them go, more wind makes them go faster, maybe the wind should be blowing in the same direction as the ship is going or something, it creaks. Reduction involves asking questions about details of the ship; how exactly does the rudder make it turn, how steady does the wind have to be, isn't there something about sailing into the wind and how does that work, if you overload it does it sink or break first, why doesn't water get it, etc. Post-reduction holism is like feeling the soul of the ship, having it feel like a part of your body, being dextrous with steering or repairing it, knowing quickly how much of what kind of wind will take you where, etc.
Often there are multiple post-reduction holisms ("essences") to be had about X. Then it gets messy if the one uses an essence of X the other doesn't have. If the other has no essence of X, they'll take strong claims about "X" to be about their pre-reduction essence of X, and those strong claims will be wrong; a hasty reduction of X might follow, after which they say "oh if I knew you were talking about [such and such essence of X] this whole time we wouldn't've disagreed".
Say what you think so you can think something else. Hear what others say as the output of an anytime algorithm.
A ship drifts nestled in the micro-wavy macro-uniform sun-warm surface of the water-ground. Its oars could push it here or there or yonder, v₁-ward or v₂-ward or any-combination-ward, but instead it waits with sails limp now but ready to grab a passing wind, pull the ship into the flow, the voluminous weighty force of the air at the ship's back. The captain knows to wait for the wind.
There's an unfortunate correlation between subtlety in thought and adversarial epistemology. If you want to exploit people while distracting them from what's happening, it often pays to keep claiming (explicitly, or via implicature / keying into social habits of good faith) that there's lots of subtlety in the situation (e.g. "no I was actually acting in good faith from my perspective but I can see how it didn't seem that way but it's complicated [and don't update that my true intent is bad]". On the other hand, reality has a surprising amount of detail, and especially social situations have lots of fast and complicated stuff going on. So subtlety is often the only truthful way out. Cost of cognition is important. Sad.
Gone, but never forgotten.
Depends what your goals are, of course. If your goal is "fundamental understanding", then what I usually go for is: read stuff until there's an interesting question that I can see on my own, and then think about that question without reading further, and keep thinking on my own until I'm mostly out of thought-provoking questions.
Models that really pop. All models are bullshit, but some are useful. Or, all models are true in some regimes but not others, and some are true in a large regime. A model that is true in a large regime is threatening to humans because someone might See Like A State with it. Pointing out regimes where a model isn't true is often taken as an attack (and indeed it might be an attack, i.e. its intended purpose and/or one of its main effects might be to dispel any political will that may have been swirling around to understand the model and make it part of the zeitgeist). It is a commonplace among some people that one should take "attacks" on models not as attacks but as constructive criticism, if it's constructive. This is all well and good. But I think it fails to question a background bucket error that [stating a regime where a model fails] is to be taken as [evidence against the proposition that the model is true]. I often instead prefer to take "attacks" as fleshing out the model. Like how a good contrast between background and foreground in a visual display helps the foreground to "really pop" out, I think of clear demarcations of the regimes of truth and untruth of a model as part of the model (indeed a necessary part of the intended model, if we want to convert the model-fragments usually discussed into full-blown hypotheses for a theory of everything, as "model" is sometimes interpreted), and hence as an assist to the project of making the model really pop.
The blessing you give by asking a question.
For a socially anxious overthinker, it's often hard to just, like, say stuff to people. Am I intruding on your thoughts, am I being an imposition on you, should I go? Am I secretly just trying to get something from you, and even though I really do want that thing I also don't want to trick you about it, so to be safely cooperative I should stay silent (should I go)? Now, even if that's all true, and even if all that should be solved intrinsically in the limit, it's still often good to grease the wheels of social relationships, as a bootstrapping move. So if you ask this type of nerd what they think of X, it bestows a small blessing on them: you've made the cooperation-point be solidly in the regime of yes-talking rather than of not-talking, and they can just answer without worrying so much that their choice of topic is suspicious and/or detrimental (they could still be manipulating you, but of course probably approximately not more than the norm of manipulativeness, which you're already at least in equilibrium about). (I personally have this type of anxiety, and also often ask questions / start threads.)
Also, there's some mysterious(-to-me) way in which being asked a question by another human activates a mental thread / orientation that is much more difficult to access otherwise.
So, by asking questions you are blessing your friends.
Amphicausal, autocausal. Two phenomena A and B are__amphicausal__ if A causes B and also B causes A. Amphi = both sides, as in amphibian (= both kinds of bios (= life)). (We could also call this reciprocausal, as in reciprocal = back and forth, but that seems gauche.) Example: wind blows over the water's surface; a ripple forms, and because it sticks up above the surface, it catches the wind more, and so it grows to stick up a bit further above the surface, and so on. Height catches wind, caught wind increases height.
A phenomenon is autocausal if it causes itself. We can coarsen our view of the wave example, and just say: the wave causes itself.
Obviously there's no amphi- or auto-causality in a plain Bayes net. But the intuitive notion is good, and maybe makes sense in dynamic Bayes nets where we think of the phenomena A and B as being collections of variables that extend over time, so "A causes B and B causes A" cashes out as "A_t causes B_t+1, and B_t causes A_t+1".
(I think there's more clarity to be had here regarding the difference between the mathematical "causes" = "has any effect on", vs the intuitive cause, which has something to do with surprising/specific conditions coming about, or something.)