Lost in the Weeds or RAM Overbuffer

 It's important to look at all the details. Check that none of them are invoking nonlinear power laws while you're not looking. However, you do have limited RAM. Not only do you need to look at all the details even if you can't keep them all in mind at the same time, you need to keep in mind your higher purpose, and not get lost in the weeds. 

 It's very easy to look at all the details that fit into RAM and conclude that's all the important details, or rather, to simply take it for granted that everything you can see is everything there is to see. So, like, remember not to. [Remembering not to] takes up RAM and decreases the number of non-meta details you can keep in mind, but do it anyway.

 In addition to reserving RAM to remember you're not remembering all the concrete details, it's also important to waste even more RAM on remembering why you're looking at the details in the first place. There's already a reserved slot for what you're trying to do, but it doesn't work right. It's very easy to lose sight of your ultimate goal, accidentally replacing it with an intermediate goal. Frankly, don't fight it. Let the reserve slot be broken and largely useless. Reserve some truly arbitrary random-access attention to either remembering to periodically check your ultimate goal, or effortfully keep the goal itself in mind, if it's small enough for that option to be efficient. 

 

For example, imagine a regular business. It's important to individually check every opportunity for profit. Don't intentionally overlook anything. The cost of checking is not high. However, don't get so absorbed and fascinated by checking for new opportunities that you neglect your already existent profitable activities. The point isn't to find all the opportunities, the point is to make money. Don't spend more money looking for opportunities than you can gain by exploiting the findings. Keep the ultimate purpose in mind at all times. This will take intentional effort.
 Owners and entrepreneurs regularly get burned on the [conservative] existing activities while going for [liberal, greedy] new opportunities, and mistakenly respond by vowing never to effortfully look for new opportunities again. Then they go out of business when a competitor eats their lunch. Later, rather than sooner, but inevitably all the same. Also, occasionally those opportunities are huge windfalls that easily pay for the price of casually checking them all.

 Likewise owners will consider a set of opportunities, and if that consumes all their attention, they will assume that's all the possible opportunities. They will stop looking. If you can see even one of the things they're not looking at, they look extremely stupid. This is because they're making themselves extremely stupid. It's childish and naive; it's the totally-untrained habit. 

 

 It's also important to reserve some RAM for remembering to check your search strategy on occasion. What assumptions is it based on? What do these assumptions make you overlook? Is it likely to be worth checking there? Watch a video of someone playing a game you're really good at. Especially something like Darkest Dungeon where the players whine all the time, meaning they tell you what they believe is beyond their control. You can easily see that they could control it; they're not doing it because their search strategy is blind to that possibility. E.g. I just recently saw a leper take a death's door check because the lolplayer had him tank without block tokens, without weak tokens, and without full health. Could have just used withstand... Look for anything in your search strategy which feels blind like that so you can fix it.

 

 When performing a logical proof it's absolutely critical to simultaneously keep in mind every relevant premise. If you hit your RAM limit, there's no way around it: time for set 3. Do stretches, increase your RAM limit. Otherwise the validity of the conclusion will be down to luck. There's no way to know which premises are critically relevant without simply accepting the combinatorial explosion and checking all their relationships with each other and with the final conclusion. Note that doing this nonverbally is almost trivial, but of course you'll have to figure out verbally why the correct conclusion is correct if you intend to build on it. Secondly, to efficiently know the range of validity it's necessary to understand the verbal proof. 

 Once all the relevant premises are processed, it's fine to discard the irrelevant ones and remember only the critical path for the proof. I generally check the conclusion is valid using retrodiction. If it works consistently, I know I haven't mistakenly discarded a premise I used but thought I didn't.
 Note that textbook or university proofs have almost always already gone through the pruning stage, meaning studying them is nearly useless for forming your own proofs. They're only useful if you've already re-invented the wheel and want to use them to check that you re-invented it correctly.