I’d like to ask better questions about biology, quickly. This is hard, particularly in fields I have no experience with. There’s no general solution to this problem, but there are things that seem to help.
I’d submit that reading papers is actually, in isolation, a potentially dangerous activity. Sure, it’s better than doing nothing. But papers are generated in a rich social context that you don’t have, with important subtext you might completely miss. A paper is to a reproducible result as a mouse is to a human - many things might work in one but not the other.
So, what on earth can you do?
The closest I’ve come to a satisfying answer is to form mental models of the underlying concepts that are robust, manipulatable, and with which you can run thought experiments. This isn’t a necessary or sufficient condition for deep biological understanding, but it’s one of the few things you can do outside of a lab and check. If your model results in an exploded cell or a single bit genome, you can make it better.
What does a thought experiment look like in biology?
People whose thoughts I admire often say that they ‘don’t understand something’, despite a perfect ability to recite the textbook examples of it. So, I’ve always been interested in what it would mean to understand something, particularly in biology.
My current hypotheses:
- Taking a lot of time (at least 20-30% of the hours you expect to spend in a field) to create mental ‘toys’ that you can fluidly manipulate, will lead to net faster understanding and storage, and potentially more and better novel ideas. This would mean at least 20-30 hours for a field you expect to spend 100 hours interacting with.
- Letting yourself do a lot of weird things that don’t look like learning, if they ‘feel’ right in the first phase, is important.
- I dance and move to create spaces in which my brain ‘sees’ the answer to things. I’m sure everyone has a similarly unique way in which they assimilate new information. Our default way of intaking and processing new information (sitting at a desk craned over a computer) seems really, really terrible at enabling and discovering these modes of thought.
Specifically, to create a mental model, you might try the following:
- Generate a list of numbers, as long as possible, relevant to the object under study.
- Attempt to ‘play’ with each number at least once. Generate a problem in which you manipulate that number, and solve it mentally. Then, attempt to visualize an object that would represent the problem, and ‘see’ the solution.
- Try to ‘see’ the thing you want to understand. Generate as many visualizations as possible, of objects in the cell you are already comfortable with. Watch each of them to see what happens - is the phenomenon you want to understand clear from the visual? If not, can you see where the photons or electrons came from in the machine that generated the result? What did they interact with, in the biology? Can you see any places where this feels uncomfortable to you, or as though it would miss something?
- If you feel really uncomfortable with the areas of the cell involved, checking whether there are basic equations or simulations that might give you a clearer intuition for the objects involved. Would better mental models for relevant equations regarding the length, timescale, information content, force, or energy involved help you generate more visualizations?
'Visualizing' could mean anything from a several seconds-long attempt to tag a word with a picture, to a mental action requiring 30+ minutes of continuous concentration, analogous to what Tesla describes here. I mean the latter, not the former.
There are two ways to solve a problem. Get a bunch of other people to work on it, or do it yourself. I do the former, professionally. I don’t expect a deeper understanding of biology to be the critical thing limiting returns in my venture fund - that would be stupid, because it’s only a peripheral aid in talent identification, and entrepreneurs don’t (typically) care if you’re intellectually quick.
But the longevity problem is hard, because it’s currently (still) decoupled from what companies are doing, ultimately. We don’t yet have the concepts to solve it completely, we just know enough to do useful things that will be very profitable. And conceptual progress is hard, particularly in companies motivated to get a product to market quickly. So, outside of the financial constraints of my day job, I think it’s worth really worrying a lot about personal conceptual understanding of the field, to make sure what I’m doing makes sense toward the ultimate goal I care about. And even if an academic has figured out an idea, it can take a surprisingly long time to be able to fluidly understand and use something that’s fast to parse. Thus, the above.
 Perhaps these sound similar to Polya’s heuristics - I think it would be wonderful if we could just directly use those in biology, but generating the right question seems to be as big of a deal as solving it, so there’s some extra creative work required.
 Great books to seed your mind here include “Cell Bio by the Numbers”, William Bialek’s “Searching for Principles’, “Principles of Physical Biology”, and a thorough review of David Goodsell illustrations.
Thanks to Sebastien Zany for suggesting many of the personal experiments that lead to the ideas above.