Learning from research papers in a developing field

Question

(Note: related question here, not duplicate.)

I'm reading quite a bit about quantum computing, especially optical quantum computing. Part of the problem is even the all-encompassing bible of quantum computing (Nielsen and Chuang, Quantum Computing and Quantum Information) has only 10 pages on this particular physical realization. Other textbooks, by the time they're published, are already out of date. Hello, reading scientific papers!

Of course, there's a problem associated with that - results are retracted, new ones encompass and expand upon old, and papers are rather notorious for being difficult to slog through. Having slogged through a few myself, I can tell you that notoriety is well-deserved. Worse (for the self-learner), most papers are focused on only a specific problem.

As in, I've read 10 pages of incomprehensible text before (think half an hour for the 10 pages, if I'm really, really lucky) just to figure out one fact, and then looked at another paper that casts doubt on that fact. Add the whole issue of paywalls to the business, and it only gets worse. (I'm not a university student, though I am lucky enough to be within driving distance of a decently sized university, but not everyone is.)

Papers often lead me down rabbit holes as well - tracing citations (I could tell a few entertaining stories about that alone), looking up terms ("oh, that's what you meant...well, why didn't you say that?"), and going down that oh-so-fun Wikipedia trail. You know, look up word, read third word on page, click on that, then continue until somehow you end up at philosophy (ironically, the previous page doesn't end up at philosophy, if I did it correctly).

How can learning from papers be made a little less painful? Things like:

• What tips do you have to get the most bang for your buck reading a paper and retaining that understanding?
• How do you help restrain the rabbit hole?
• How do you know when a paper's trustworthy?

Answers don't need to cover all of the above, but be along those lines.

Answer 1

Seems like a developing field is a discussion that a community of researchers engages in. To participate, you'd need to:

1. Speak the language, e.g. know about quantum computing terminology.

2. Know the background, e.g. quantum physics and computational theory.

Once you have some basis for understanding the discussion, then you focus on getting your own ideas straight. In the end, it's really all about you; sharing research results really isn't a goal so much as a duty to be performed when/if you become the foremost authority on some particular understanding.

If you're a younger student, you've probably got a lot to learn and develop before you can really enter that arena. But, if you're interested in it anyway, it seems like you could do two things:

1. Focus on the well-agreed-upon concepts. You don't need to understand them or even be able to use them just yet, but have them in the back of your mind as you progress in your more foundational studies.

2. Take note of the skills and tools that seem to be most useful in the current community. For example, if the field seems to be best-attacked through a particular branch of mathematics or using particular computational tools, then you might want to put acquiring those skills on your to-do list.

Getting to Philosophy

Papers often lead me down rabbit holes as well - tracing citations (I could tell a few entertaining stories about that alone), looking up terms ("oh, that's what you meant...well, why didn't you say that?"), and going down that oh-so-fun Wikipedia trail. You know, look up word, read third word on page, click on that, then continue until somehow you end up at philosophy (ironically, the previous page doesn't end up at philosophy, if I did it correctly).

Definitely, Philosophy is exactly where you want to end up! Anyone who can't trace their field back to Philosophy hasn't understood it so much as memorized it.

If you know exactly how to get from basic philosophical principles to the current state-of-the-art, then you can really claim to understand the field.

Questions

What tips do you have to get the most bang for your buck reading a paper and retaining that understanding?

Lots of profiling. For all anyone knows, that random gibberish full of typos and conspiracy theories on Reddit is the deepest, most correct theory of physics ever. But, since it's probably not and there's only so much time in the day, your bets are better hedged by focusing on content that appears to be of higher quality.

To that end, you should develop a bunch of prejudices over what's likely to be better content, then selectively focus on it. Sure, if you're truly out of stuff to read, then maybe you'll end up going through that weird Reddit post; or, if you've got too much good stuff to read, you might not even be able to make time for Einstein's words. It's all relative.

Examples of common prejudices:

• Prefer scholarly sources for widely understood concepts.

• Prefer blogs and communications from top-experts for the-very-latest commentary.

• Well-formatted over poorly-formatted.

• Thoughtful over thoughtless.

• Analytically precise over vague.

How do you help restrain the rabbit hole?

1. Follow the rabbit hole down as far as you can go without losing sight of the light of day.

2. Understand where you've ended up, and build a solid foundation there.

3. Build back up 'til you're out of the rabbit hole, standing on more solid ground.

4. Rinse and repeat, iteratively falling down further and building back up more. The more you do this, the greater your understanding grows.

How do you know when a paper's trustworthy?

They never are. Whenever you read a textbook - even the best-known ones in well-developed fields, e.g. introductory Calculus - never just trust it. Put each word on trial, and absorb only the ones that you can't tear down.

You often need to memorize large amounts of content without first thoroughly scrutinizing it. All of this content should carry a red flag in your head that labels it as unscrutinized claims. If there should come a time where you really need that information, or want to build upon it, you should really scrutinize it first.

Answer 2

The problem you pose is a difficult one. Every graduate student researching for her/his dissertation faces it. Suppose you have a particular paper in hand and want to learn from it. The normal solution can be expressed recursively:

method research(Paper x, Author y){
    if(y == IsaacNewton){
        read(x);
    } else {
        foreach(Paper paper: referencesOf(x)){
            research(paper, paper.author());
        }
        read(x);
    }
}

The search space is humungous, but at least tree-like (well, mostly). Sometimes you can prune the tree.

The problem is that in an area that hasn't yet "shaken out", with secondary sources already existing, you are left with primary sources only. And notice that the recursion is bottom up. You read the older papers first just to learn the terminology used.

Hope I didn't err in the algorithm. It has been years and years since I personally had to execute it. But I feel your pain, still.