11 minute read

I turned 30 recently and have always enjoyed Sam Altman’s The days are long but the decades are short post. I decided to write something similar. A few notes and caveats before we get going:

  • These ideas are roughly in chronological order by when I learned them.
  • They’re primarily career oriented.
  • Almost all of the paragraphs should be prepended by “I really should write a full post about this.” :)

(1) Knowing something is really hard.

Many of the lessons taught in my Chemistry undergrad were about measuring uncertainty. Do you think you know how much water is in this tube? Have three people measure it and they’ll come up with three different answers. Want to know the shape of a new protein? Be ready to commit a couple years of full time work. Want to know the solution for Middle East conflict? Ask your Uncle John; he’s got the answer. The cavalier people demonstrate for claiming they “know” something is astounding. Unfortunately, the last few years have demonstrated how disastrous these types of assumptions can be.

(2) People matter most.

In highschool, I thought it was pretty cool to believe that people were boring, and as a consequence, that people didn’t particularly matter. Maybe it was a mechanism to boost my own self confidence. But the truth is, people matter. Their well-being matters. Recognizing and accepting this fact is life changing. What’s potentially surprising is, if you attempt to improve the lives of those around you, your actions tend to improve your own life as well. This seems counter to a lot of modern advice around self-care, doing whatever makes you happy, etc. I haven’t learned to let myself be a doormat, of course, but instead that deeply caring for people is worthwhile. Shoutout to my friend Drake for teaching me this.

(3) Sometimes, perception is more important than reality.

It wasn’t until grad school that I started to realize the implications of how difficult knowing the truth of things. A vast majority of the time, it’s literally impossible to know what’s true.

  1. Which tool is best for aligning NGS reads to the genome? You don’t have enough time to come to a conclusion.
  2. Will it save money to move your workload from on-prem to the cloud? Probably, but the way to know would be to do an experiment you don’t have the resources for.
  3. Which of the two employees explaining their conflict from last week is telling the truth? You weren’t in the room…

I think I’d like living in a world where more things are knowable. But, the best I can do is acknowledge that we don’t and figure out how to deal with that. In each of the three examples above, my perception of the situation is bound to influence my decision:

  1. This code seems much cleaner, maybe it’s the better approach.
  2. My HPC seems like it’s down a lot, AWS probably wouldn’t do that.
  3. Employee B seems more trustworthy, maybe their story is more accurate.

Learning to leverage this differential between people’s perception and reality responsibly and appropriately can be a valuable skill.

(4) Speaking is a learned skill.

Even through college, I took it for granted that I was a “bad public speaker”. Like speaking was some inherent trait encoded in my DNA or something. Given what percentage of our lives are spent talking to one another, it’s remarkable how far in life most of us go before getting any formal training on how to speak. Most people never receive it. Thankfully, the startup class I took during grad school involved facing real time feedback on presentation quality every other week. That was enough to motivate me to record myself speaking. Listening to my own voice was a torturous experience, but directly resulted in a significant improvement in my ability to affect change in the world.

(5) Stories are important.

People think in stories. Regardless of if they’re conveyed via speaking, writing, film, or dance, people imbue everything from facts to events to randomness with meaning. And they construct that meaning through stories. If you want to be able to communicate with an audience, the most surefire way of doing it is by telling a story.

(6) A little bit of statistics goes a long way.

People only think in three probabilities, 0%, 50%, and 100%.

I’m not very good at statistics. I’ve only every taken one class, in grad school, and it was quite difficult for me. I’d love to be able to think more intuitively about how different situations translate to different distributions. Bayesian hierarchical models sound cool. But, even the statistics I do know come into use on a weekly basis even though I’m not doing research anymore. Statistical thinking is immediately applicable to nearly every facet of life. Again, the last two years would have been so much easier if a majority of people in the US had made it through an AP Statistics level course. A highschool-level understanding is enough to shift the way one think about the world.

(7) I like building more than I like discovering.

This is strictly a personal preference, but an important one for the trajectory of my career. While discovering is fun, the excitement is always tinged by the doubt of uncertainty in my conclusions. When I’ve built something it either behaves as I envisioned, or it doesn’t. And building something from nothing—which then functions as I saw it in my head—is one of my favorite feelings in the world.

(8) Genuinely finding people interesting is a great way to experience the world.

This lesson could be interpreted as having a lot of overlap with lesson 2, but something that has surprised me over and over is how quickly people find themselves drawn to others who actively listen to them. The TV show Love Is Blind is a ridiculously over-the-top example of this, but makes the point all the same. Take a few dozen people (many of whom appear to have never had an intimate conversation in their lives), force them into a situation where they have a deep conversation, and they soon want to marry each other. More relevantly, simply asking deep questions and actively listening to the responses is a straightforward way of building connection with people.

(9) Passion can be more important than talent.

I consider myself extremely fortunate to work with some of the most intelligent, educated, and talented people in the world. It’s genuinely inspiring to learn from them and watch them work. And yet, it seems that the strongest correlation between ability to affect change and/or produce results is not with raw intelligence or talent of an individual, but instead with their passion and enthusiasm. The “Outlier” employees are those that care.

(10) Making friends outside of cohorts requires more effort.

This one is sort of a bummer. But much of our early life, we’re involuntarily placed into cohorts with other people our age who are in the same situation. What’s nice about this is that everyone is simultaneously looking for friends. When you move or change jobs as an adult, you join a group of people who mostly have their friends and routines established. Making new friends is still possible, it just takes more dedicated effort.

(11) Academia would be significantly more efficient if the skills needed to become a top-tier PI weren’t antithetical to the skills needed to be a top-tier PI.

I’m not going to attempt to claim solutions any solutions on this one. But the idea that you should select the most analytical people in the education system, allow them to focus in near total isolation on a single niche interest for nearly a decade, and then promote them into people management with little to no training is downright bizarre. Any improvements that could be made in training early career PIs or improvements in deciding who should become a PI in the first place would have large ripple effects in our society over the coming centuries.

(12) I learned to program. And how little programming you do as a SWE.

It’s pretty wild for me to think about the fact that I had no idea how to program 10 years ago. I’ve had so much fun learning about the technology that makes much of 21st century life possible. What’s even more wild is that after all the hours that it takes for a human to learn to program, the average software engineer spends very little of their corporate life writing code. While some minority stay individual contributors, it’s easy to end up spending your life in meetings.

(13) Cities have very distinct cultures.

You often here people talk about traveling to other countries in order to experience different cultures. But I rarely, if ever, here people talk about traveling to another city to experience a different culture. Sure, the difference will be smaller. But I’ve lived in 6 different places as an adult, and every single one of them had a noticeably different culture than any of the rest.

(14) Step 1 is, “draw a smaller circle on top of a bigger circle”. Step 2 is, “draw the rest of the fucking owl”.

It’s true what we suspected as teenagers: the adults rarely know what they’re doing. When it comes to building new technology, there is no playbook or recipe. There’s no teacher you can turn to for answers. This has been a difficult thing for me to accept over the past decade. I like knowing that the thing I’m doing is “correct”. I want to be able to flip to the back of the book and double check my answer. When I’m unsure, I want to be comforted by going to my boss and asking, “Am I doing this right?” and they instantly respond, “Definitely.” But that’s never what happens. Instead, I ask, “How am I supposed to do this?” and the response back is a more polite version of “Follow step 1, then step 2.

(15) Making promises about the way you’ll feel in the future is difficult.

There’s not much to say about this one. It’s hard. Tread with caution.

(16) Don’t listen to other people about side projects.

I’ve heard so many different opinions on side projects over the last decade. “You have to have a portfolio if you want a job,” or “Don’t do side projects, they contribute to a culture of worker abuse,” and everything in between. I’ve learned to tune it all out. I’ll do them if I want. If I don’t feel like having one sometimes, that’s fine too.

(17) Repeating yourself is good.

I don’t like repeating myself. Not in an arrogant “I shouldn’t have to do this” way. It feels like I’m implying that people didn’t hear or understand me the first time or I think they forgot. But it turns out that repeating yourself is another important part of communication. As one example, marketing research shows that people need to be exposed to a brand around a half dozen times before they remember it. As another, I’ve lost count of the number of times I’ve heard Invitae’s CEO walk everyone through the company’s cultural tenets. Now when I have an important thought at work, I don’t say it once and feel sad if nobody immediately jumps on board. Instead, I make a note to talk about it again next week.

(18) Modeling a wide variety of things as hierarchical trees seems very natural to me, but other people don’t seem to do it much.

Want to build a video game? That’s a root node. You’ll need to flesh out a concept, write the code, and generate the art. Those are the three children nodes. Everything you need to do falls into one of these three buckets. We could talk about the children nodes to any of those three and continue deeper and deeper until we have a full description of everything we need to do. Personally, that’s the most natural way of describing a project. But I’ve had many conversations with people who are fully aware of what a tree is, but still seem surprised that one might use it to describe a project. JIRA is another good example of a popular task/project management system that doesn’t encourage a tree structure. It’s weird to me, but there are probably interesting things I could learn by discovering how other people outline projects in their heads.

(19) Edges are more important than nodes in a knowledge graph. That implies that knowledge is partially quadratic. This is why experts seem so far ahead.

When you think about your personal/internal knowledge graph, it’s tempting to count the nodes as the “number of things you know”. But I’ve learned that that’s probably not the best semantics. Thinking about the connections you can make between distinct facts is likely a better illustration of “knowledge”. This is probably part of the reason the frequency illusion is real. When you’re told a arbitrary fact, it’s nearly impossible to retain and you quickly forget it. It isn’t until you can connect a fact to a set of other information that it becomes valuable.

My experience has been that this means the rate of learning is superlinear when you’re entering a new domain. On the other hand, people seem to extrapolate that their learning rate (when comparing themselves to others in the field) will be linear. Keeping this discrepency in mind while attempting new endeavors has bolstered my confidence. I’m not as far behind as I feel, because I’ll learn faster than I expect. It’s the edges, not the nodes.

(20) Having an expiration date on life is both important and good.

I’ll admit it: I’m a little forlorn about turning 30. But even though aging and dying seem scary, life without them would be terrible. I picked this one up from one of my favorite shows, Black Mirror. In San Junipero, the scene at The Quagmire provides an artful demonstration of why an expiration date is important for humans.

So, even though I’ll mess it up sometimes, while my clock is still ticking, I’m going to do everything I can to enjoy life and to make sure those around me and those that come after me enjoy it as well.

Cheers to another 10.