How to Read [Non-fiction]?

Alex Salo
3 min readDec 29, 2020


For the last eight years, while reading anything non-fiction, from books to news, I’ve been using the tool that my professor of macroeconomics taught us, though it really should be taught in school during the language classes.

It’s very simple. The idea is to always split the text into the assumptions, facts hypotheses, and conclusions, just like you would split a sentence into object/action/subject/qualifiers. It’s a sign of a good writer if they clearly separate the facts and assumptions from hypotheses and inferences. It’s a sign of bad writers to hide behind the smoke and mirrors of complicated words, masking assumptions as facts (that’s the most sinister in my view), and decorating the hypotheses (which are by definition up to debate) with moral judgements — I’m looking at you NYT and most of the other mainstream media.

Facts. Ideally the facts are all fact checked so you trust it if you trust the author/publisher. Though this is where it’s helpful to be able to see where the “facts” are actually just assumptions or axioms. A good rule of thumb is that any value statements are just assumptions, and any general historical/statistical data are facts. Though data could be very misleading, and that’s a topic in its own right — consider, for example, a notoriously ignored factor of median age when writers compare different peoples: it’s pretty obvious that young and inexperienced people earn less income than older and more experienced people, so if one group’s median age is 20 years higher than another’s then it would be foolish to expect their incomes to be similar. If you think 20 years is far-fetched, just google it. If you look outside of this country the contrast is even greater — Japan’s median age is 47 years, while Angola’s — 16(!).

Assumptions. Assumptions are the most sinister because they often go unstated and presented as facts. But if the assumptions are wrong, then no matter how accurate the logic of prediction is it will be totally incorrect. I always try to scrutinize the assumptions and check them against my own intuitions and convictions-since assumptions are not facts everyone is entitled to their own treatment of acceptance or rejection. If you find yourself staring at the conclusion/hypothesis that does not feel right but you can’t explain why — try checking the assumptions, usually you’d find them flawed (against your own background).

Hypotheses. Hypotheses are usually easier to see, and of course you should never blindly trust it — I always try to critical-thing about it, debate its merits and check against my own understanding. Don’t consume hypotheses at a face value even if someone smart had wrote them down. Can you come up with the questions that will further test the hypothesis in your own unique background and experience? Only if it passes your personal reality check does it deserve a place in the corpus of your knowledge.

Conclusions. Conclusions are the easiest to test — one just needs to verify the derivation logic is valid, given that you checked the facts, assumptions and hypotheses that the conclusion is based on. What’s cool about conclusions is that typically I can come up with more of them that the text mentions. And that’s useful in testing the underlying hypotheses, but also in deriving new knowledge and insights that are unique to my own background.

As one can notice this process will tend to build an echo-chamber like tree of knowledge. That is fine. You want to be confident in the knowledge you have-not based on some ideology or blind convictions but based on the logic and critical thinking. You do need to keep this fact in mind though, so don’t keep that knowledge sacred — if a good argument comes along and completely unroots a part of you tree- let it, because in place of the old branches you are getting a new and stronger ones.