How to pick what to read ?

The problem is aptly stated by Scott Aaronson :

Instead I want to explore the following metaquestion: suppose someone sends you a complicated solution to a famous decades-old math problem, like P vs. NP. How can you decide, in ten minutes or less, whether the solution is worth reading?

For a blogger like me — whose opinions are both expected immediately and googlable indefinitely — this question actually matters. Err in one direction, and I’ll forever be known as the hidebound reactionary who failed to recognize some 21st-century Ramanujan. Err in the other direction, and I’ll spend my whole life proofreading the work of crackpots.

I have a solution to this problem, one with which I am extremely satisfied. It goes like this :

A. If you are someone I personally respect, or if you come recommendedⁱ by someone I personally respect, I'll hear you out.

B. If you are my student, which means you are paying me to teach you the subject matter, then that comes with a certain allocation of "I'll hear you out".ⁱⁱ

C. If you are neither, you will be regaled with some repackaging of the simple idea that "if indeed you can't obtain my respect, and you can't obtain the recommendation of anyone I respect, and you either can't get the money to pay me to teach you the thing or else don't have the patience to be taught then you have much more pressing problems to attend to than whatever it is you think you are doing and want to talk about", in such a form as I'll judge you're best able to digestⁱⁱⁱ.

That is all, without exception. I do not consider anonymous items on their "merits", because anonymous items do not have merits.^iv

This solution works admirably in all the science fields, which is what the original question considers, but it also works just as well in all the other fields. It's a perfectly fine approach to humanities as well as the correct way to do business, and anything else.^v

Sure, I stray from time to time, especially when involved in some new and turbulent field^vi, but I stray in my own time and in full knowledge and acceptance of my sinful state.

There is no "issue" of elitism at work whatsoever. Elitism is the entire point of existence in the first place, outside of it there can be no happiness nor human culture nor human emotion nor anything of any value, interest or import. Just as long as one keeps the filtering principles properly attuned to reality^vii, elitism is the best way to live.

1. This is where Web of Trust systems become all important : they not only allow efficient scaling of far reaching, multi-layered trust trees (where I try to establish what value my trust of A and A's trust of B and B's recommendation of C has if any) but also allow the work of translators, those people who are trusted in disjunct sets thus allowing the sets to globally and broadly cooperate (while perhaps on occasion also helping out by smoothing out trouble created by difficulty in translation from one set of mental representations to another, hence their name). [↩]
2. This is what teaching means, anyway : reviewing the braindamaged propositions of the teachee and pointing out the braindamage. Everything else they can do on their own, and they should do on their own. Getting them from the latter to the former is their mother's job. [↩]
3. Which, in most cases, does include a good helping of public humiliation. This is also to your own benefit : you are finding yourself in a situation where your estimation of yourself is so badly divorced from the reality of yourself that you find yourself wasting time working on problems that don't exist instead of using that same time working on the problems that do exist. Public humiliation is the cure for this mental issue. [↩]
4. Ad absurdum should some third grader produce under my very eyes the proof of a superlinear circuit lower bound for SAT, it will have been my eyes that produced that proof, not his hand. His hand was simply moving about randomly on a piece of paper, much like a fly would be moving on the wall. [↩]
5. Yes, sex and relationships included. [↩]
6. Bitcoin is a prime example, there are plenty of coder folk involved that you can talk to whenever you feel like, but as far as I know I'm the only non-programmer entity accessible in the same way. [↩]
7. Which does involve only respecting people worthy of respect, rather than going for the easy or the convenient. Which does involve spending a lot of time to keep the WoT in correct working order. Which does involve making an earnest effort to teach, whether it brings the object to tears or not. Which does include a number of other "brutal" and certainly inconvenient nooks and crannies. [↩]