Becoming a more playful, deeper thinker

Note: Ignore the weird Markdown formatting, it’s because I copied this from the blog source code.

Inspiration: Understanding by Nabeel Qureshi, David Deutsch’s The Beginning of Infinity and Fun Criterion.
Meta: This is an experimental post in which I dissect something about myself rather than write about some technical topic. Feedback and criticism encouraged as always!

One commonality I’ve noticed amongst profiles of thinkers I respect–Richard Feynman, John Von Neumann, David Deutsch, Donald Knuth, George Church–is that these folks enjoy(ed) “thinking as play”. Concrete examples include:

• Feynman’s development of QED all starting with him burning out and then thinking about spinning plates in the Cornell cafeteria.
• Every Von Neumann story that involves someone proposing some problem and him spending seconds/minutes (Von Neumann was an alien) solving it.
• David Deutsch tweeting about fun (also here) as a necessary condition for ‘good’ research.

The perspective and way of deciding what to think about these stories reflect is somewhat foreign to me. I definitely like learning, but I often lack the patience to fiddle around with things for fun to the degree the people I mentioned do. Working as an engineer also instilled a strong “is it useful?” sense in me that I struggle to quell.

Recently, I’ve been questioning whether my fear of wasting time and going down rabbit holes is holding me back. In this post, I try to understand some of the factors that led me to have this perspective in the first place and try and figure out whether I need to change.

How did I get here? #

I find myself in the position of wanting to be the type of person who thinks as a form of play but feeling strong internal conflict whenever I try to do so. Given this, it seems like the best first step to changing is to try and figure where that internal conflict comes from.

Identity as an engineer #

In my first year as a software engineer, I viewed my bias towards speed and usefulness as a strength. When I started working, I had a deep insecurity around being able to produce and deliver which led me to want to prove that I could deliver quickly and effectively. For example, I distinctly remember worrying that my standup updates wouldn’t seem substantial enough and that people would start to think of me as unproductive. During this time, I also saw how some of my peers would more deeply understand all the details of a system or new technology but took longer to accomplish tasks. Related to the former, there were a few instances in which I’d prepare for a group discussion and I’d get asked a series of “why?” questions that I quickly became unable to answer. While my embarrassment in these situations taught me to go deeper, it also reinforced my notion that learning things deeply wasn’t my comparative advantage. On the other hand, because I started out so focused on delivering, some of my early work successes came from deciding to ignore non-essentials and relentlessly focus on a few key goals. Especially since these experiences involved praise from senior people, they strengthened my developing self-image as someone who Gets Things Done rather than deeply understood them.

My reading about startup culture also reinforced these traits. As a college student and junior engineer, I was especially influenced by articles by Important People like Paul Graham, Sam Altman, and other entrepreneurial types that emphasized speed as one of the most important traits a person can cultivate. While I now have a more nuanced perspective on these sorts of essays, back then I had a separate issue of reading people I respected un-critically and blindly following their advice. ¹ Taking a step back, while it may sound like I was hyper self-aware of my values and their influence on my behavior, if you had asked me at the end of 2017 what it meant to be a good engineer, I wouldn’t have said “good engineers get shit done and don’t waste time understanding things.” It was just that I would start to feel dread if I went too long trying to learn stuff and aggressively minimized my time spent going down rabbit holes while feeling low-key envy of the people who understood things better than I did.

Fortunately, after a year and a half of working, my need to constantly prove that I could do things calmed down and I got the opportunity to work with senior engineers who wouldn’t settle for anything less than deep understanding. Seeing successful people I respected take the time to learn things and do them right showed me that one could be productive while also going deep. Related to this, like every software engineer ever, I saw how stuff we built quickly the first time around ended up having to get rearchitected (I’ve since seen the inverse happen, i.e. overly designed projects that had to be rearchitected anyway). However, I wouldn’t say this experience taught me to enjoy going deep so much as to be afraid of not catching something. Put a different way, my self-worth source shifted from being proud of going fast to being proud of being systematic and thorough.

Reader, not thinker #

I’ve always read a lot. While I know this is a common internet bragging point–see the myriad articles talking about how to read 50 books a year–I actually think my natural tendency to read widely and quickly has also hindered my development of “deep” thinking in two ways. First, for a long time, I would read books and then move onto the next one without internalizing the lessons from the book I’d just read. This resulted in me often having surface level opinions about topics that came from books I’d read but that I’d never digested enough to make them my own. Second, as a member of the first internet-native generation, I developed bad habits around searching for answers to questions rather than first thinking about them myself that my tendency to read everything exacerbated. Combine these tendencies with reinforcement from people occasionally being impressed by me raising a point from something I’d read and you get a feedback loop which discourages spending the time to digest and problem solve in favor of just reading like a madman.

Fear of running out of time #

Especially as I age, I constantly worry that I’m running out of time to do all the things I want to do. In an ideal world, my response to this fear would be to prioritize and really focus. Instead, my response has often been to just try to do everything faster. My favorite example of this is my internal dialogue around textbooks I want to read, which goes something like this.

Ugh there are so many textbooks I want to read. Real analysis, Bishop’s PRML, MacKay’s ITILA, measure theoretic probability…
I’ll never get through all of these and do research, work, etc.
Ok but wait, here’s an idea. What if I can figure out how to finish a textbook in two months.
Then I’ll be able to read six textbooks a year!

Granted, this time pressure can be a good way to force myself to really buckle down and focus. But it’s also a recipe for rushing through things while forgetting that the goal was to learn not just “finish” (a case of Goodhardt’s Law).

Practice must be deliberate #

In a separate vein, I’ve spent a lot of time reading books like Peak by the late Anders Ericsson and Deep Work by Cal Newport which argue for a helpful but rigid framework (deliberate practice) (1) (1)If your association with deliberate practice is the 10,000 hour rule as described by Malcolm Gladwell Outliers, I strongly urge you to at least skim Ericsson’s original paper on the topic. The TL;DR is that Ericsson’s whole point was that in order for the 10,000 hours to matter, one needs to be very deliberate about what form of practice one is doing. So the Beatles playing many hour-long shows would almost definitely not qualify under his definition.

for improving skills. Learning about the deliberate practice framework definitely helped me but also accentuated my tendency to scrutinize and plan my time spent learning. As an example, I often carefully plan my time spent reading textbooks and then worry I’m spending too much time reading or thinking rather than solving exercises.

Bringing it all together #

Bringing all these threads together, my sense is that I’ve internalized a sense of impatience and obligation around learning that makes it hard for me to patiently and carefully think about things, especially if I can’t see them leading to a concrete benefit. While this helps me stay on task and achieve unambiguous goals, this often comes at the cost of shallower understanding and flinching away from too much thinking about “unimportant” things. (2) (2)Playing devil’s advocate to myself, there’s also a failure mode where you only contemplate and never act. In the limit, this leads to accomplishing nothing and being passed by by external events. In the end, the explore-exploit trade-off this is an instance of seems fundamental to me so it’s not like I’m just beating myself up over an easily solvable problem.

Is this even a problem? #

Depth is relative #

The above paints a skewed picture of me as a speed-obsessed script kiddy who thoughtlessly but rapidly accomplishes tasks. In reality, if you ask e.g. my parents and sister, they’d tell you that I’m one of the most “in their head” people they know. Point being that this description of myself as less “deep” has to be taken as relative to where I want to be.

Have I just found my niche? #

One argument against changing is that the middle ground I’ve found seems to work fairly well for me. I’m able to get things done while also spending time learning things but not getting too bogged down except when necessary. Put a different way, strategies are environment-specific. If there’s approximately zero chance of me having a Feynman/Deutsch/etc.-style fundamental breakthrough in any science, why would I adopt their strategies?

And it’s no coincidence that of all the folks I’ve mentioned, George Church is the one who works in the least theory-heavy area, biology, and seems to correspondingly spend more time doing lots of stuff and less thinking. More broadly, physics and math style thought experiment and theory-driven thinking haven’t proven as successful in biology since the fundamental discoveries of things like the DNA helix (which were discovered this way). So, as a person interested in (computational) biology, I shouldn’t assume that these strategies would help me. As another example, Ed Boyden, a neuroscientist and one of my other science role models, recommends more deliberate strategies for having ideas that are more similar to my current strategies and tendencies. (3) (3)That said, I by no means intend to describe Ed Boyden or George Church as thoughtless. These two both clearly spend a lot of time thinking and pondering and have made creative breakthroughs in their respective fields. Further, it’s worth noting that despite what I said, the first item on Ed Boyden’s list is still all about learning how to think creatively with what you’ve learned, and, based on interviews, George Church does seemingly spend a lot of time pondering things.

Yeah but it is a problem #

Taking these objections into account, I still want to fix my ugh fields and blocks around thinking without a clear objective.

Thinking things through is important for happiness #

Thinking things through carefully and dialoguing with myself seems especially helpful for personal issues. As a meta example of why, I wasn’t even able to spend time sitting down and working through the above until I committed to writing a blog post about it, something my subconscious tags as semi-productive. Now that I have sat down and worked though this, I’m pretty confident I won’t regret having spent the few hours it took to do so! That I could only let myself do it by writing a blog post about the process illustrates that my current strategies are suboptimal.

Understanding things clearly as a terminal value #

Similar to Chris, I’ve recently realized that I have what feels like a terminal value of clear understanding and explanation. I don’t know of any substitute for the feeling of understanding something formally and intuitively that was previously a mystery to me. In the past I’ve ignored this value in favor of other instrumental values, but I no longer think this makes sense.

While there’s no magical formula for going from confused to clear, idly thinking and pondering is definitely a key component. Nabeel Qureshi’s post that I cited as inspiration above is great, and it includes an especially good bit on understanding things by slowing down.

My countervailing advice to people trying to understand something is: go slow. Read slowly, think slowly, really spend time pondering the thing. Start by thinking about the question yourself before reading a bunch of stuff about it. A week or a month of continuous pondering about a question will get you surprisingly far.

Taking this and my stated value of understanding things seriously implies that I should be much more willing to ponder things on my own, even if it feels inefficient in the moment.

My niche has shifted #

Putting aside introspective emotional considerations, I take the objection I made above about computational biology not benefiting as much from Eureka-style deep thought breakthroughs seriously. However, I think when evaluating how this relates to my personal niche, it assumes the wrong comparison. I’m not a physicist going into biology, I’m a former software engineer moving into a more science-y area. Therefore, it’s reasonable to assume that the niche in which I originally developed my skills was backend and infrastructure software engineering. Looking at the comparison through this lens, much less is known about what works in computational, ML-focused biology than in backend software engineering, and a corresponding shift towards a more explore-heavy strategy seems reasonable. Related to this, even if I’m not going to invent the next theory of relativity, I can still aspire to do creative work in the area in which I do/will work.

Where to go from here? #

As a big believer in not destroying systems without having replacements ready, I’m wary of throwing out my discipline & planning-based strategies without developing tested alternatives first. Also, since these strategies I’ve been talking about often operate below my conscious control, I doubt deciding to change all at once would even work. Instead, I’d like to try a small experiment in which I pick a source from which I already have been learning, David MacKay’s Information Theory, Inference, and Learning Algorithms and try to, as Feynman put it, “study hard what interests me the most in the most undisciplined, irreverent and original manner possible”. I’m also going to deliberately try to block out more time to “just think”, which I hope will replace time spent browsing the internet (see my failed attempts at doing this previously if you want to calibrate your expectations on how likely this is to work).

Beyond that, I’m honestly not sure what else I can do, which is part of the reason I’m putting this post out there into the internet ether! Almost everyone I’ve read about or know who works this way seems to have always worked this way, so combined with the fact that, as I said before, a lot of this happens below the conscious level, I’m still figuring out the right way to go about changing it. If you have ideas, please reach out over email or Twitter!