Molochinating One Year On

Some follow-ups on a year's worth of posts

Molochinations started a year ago on July 2, 2025 with the incendiary claim that you should always insta-sell all RSUs the moment you vest them. I’d like to share some reflections after publishing a full year of weekly articles and also update you on my thinking on several posts.

Fireside True Story™ Time: My mom passed away from advanced liver cancer exactly two weeks after my first post. I had quit my job at OpenAI six months earlier and moved my parents into our basement to take care of my mom. It may seem surprising I started posting during those final weeks of her life, where the suffering went up greatly and the number of medical tasks similarly increased, but I found it a reprieve, as I always have, to spend time writing about unrelated topics.

Writing has always been a source of joy and flow. It’s an act of thinking. For this reason, I have never used AI to generate any English prose — not in Molochinations, not in emails. I even gave up dictating Molochinations after trying it for a few paragraphs because the slow act of typing enables a form of parallel contemplation I can get no other way. It’s like the way many insights come to us when washing the dishes or showering, instead of when laying still or focusing exclusively on a thought. The idle physical engagement allows the mind to roam semi-free as the body keeps it on task through rhythmic, near-autonomic action.

So here we are a year later.

If you look at AI-generated images in my posts over the past year, you’ll see they’ve become better on average by leaps and bounds. This exception proves the rule.

Reader Favorites

From the past year, here are the most viewed articles:

• Insta-Sell All RSUs, Always (24k). Most people have correlated risk bound with their employer by holding RSUs (not to mention owning a house in their employer’s city). You shouldn’t hold what you wouldn’t buy. I continue to abide by this, but made an exception last year when selling OpenAI shares because I was persuaded by Steve Na’s rebuttal to keep a small portion of the stock as a way to support my former coworkers.

• How I Got a Job at OpenAI (22k). I’m surprised how little people prepare for interviews when they dictate so much of our future lives. The majority of getting a job has already happened by the time you interview, based on a series of past experiences which should hopefully build on each other. But you can significantly improve your odds of getting an offer by intensely preparing for interviews.

• No More Code Reviews: Lights-Out Codebases Ahead (20k). It’s self-evident that humans can’t possibly keep up with code reviewing every line of AI-generated code. We need to build systems for this new reality. And there will definitely come a day — I’d argue soon — where it’ll be a liability to your company’s reputation to let human beings touch code. My convictions have only gotten stronger since authoring this piece given interim developments in AI.

• AI Killed the Individual Contributor (10k). Developer jobs are becoming far more like manager jobs and PM jobs. The age of a lone developer thinking hard on implementation details is over. This may not be good news for many people as AI erodes craft/flow by requiring we split our attention amongst many parallel agents for the sake of maximum efficiency.

• You Should Change Companies (9k). Companies, like countries, can give us a mistaken sense we know how best to organize and behave in groups. Only by switching companies can you truly experience different cultures and draw insights otherwise not possible about yourself and about how to build successfully.

• Waxing Asymptotic in Career Velocity (9k). Everybody has a terminal level, a maximum they’ll reach in their careers. Acknowledging this may help reduce the urgency to speed-run the entire experience. It may also ameliorate feelings of incompetence or languishing if we accept our limits. That said, most of us probably underestimate the levels we’re truly capable of.

• Sometimes a Thing Is Just Expensive (8k). You often get better quality by paying more, but many items are Veblen goods, where demand increases as price increases (i.e. where the point is paying more). I’m not sure why this article resonated so well with many people. Maybe we like hearing that expensive things aren’t always what they’re made out to be. Maybe it feels like pragmatic advice.

Note that power laws seem to apply: AI Killed the Individual Contributor had half the views of the three articles above it.

The things that are popular tend either to be controversial (sell everything! no code reviews!) or self-improvement-y (i.e. of potential career benefit to the reader). Though this is not surprising, it’s clear why these dynamics would cause more people to post extreme claims in order to get more viewership.

If your favorite article wasn’t listed above, I’d love to hear about it in the comments.

The Duds

You might be wondering which articles did the most poorly. Here they are:

• The Design of Everyday Things. I just got back from a fancy conference center where I watched many people struggle with push doors which had pull handles. The constant violation of simple design principles seems to be a pastime of sorts, so the fact that this article was the least read of the 53 that have been published bodes poorly.

• The Morass of Idea Blindness. To paraphrase William Gibson, “Great ideas are already here — they’re just not evenly distributed.” We seem willfully blind to applying well-understood, time-tested ideas to different geographies or domains. Digital technologies will hopefully hasten the spread of good ideas as it becomes easier than ever to learn what others are inventing and doing.

• Whither Titlelage? Whether to have publicly visible titles in a software organization is often debated in small companies, whereas the same question rarely comes up in other fields. It’s probably because software organizations often have meritocratic ideals which bristle at the notion of title-based authority. I’ve seen hidden titles work well in smaller organizations but have never seen them scale.

• Form Line! Form Line! Serpentine lines (i.e. lines which bifurcate into multiple lanes at the last possible moment) have been shown to not only be more fair but also to reduce average wait times. Yet we see them in places like Europe and the Far East but rarely in the US. The time for serpentine lines has come.

• Technological Progress Leaves the Dregs of Employment. Is there a chance that technical advancement brings with it the degradation of job satisfaction? We saw this during the Industrial Revolution as well as the introduction of Ford’s assembly line, but might it be a general principle that new technologies sometimes hollow out the most enjoyable parts of a job, leaving behind only the worst parts in the name of efficiency?

Each had under 2k views. Three of the articles are thoughts around design, so perhaps lack of interest reflects a readership more interested in technology and career topics.

But I’m fascinated by the impact design has on our everyday lives, so brace yourself for more unpopular posts about this essential but neglected topic. I’m willing to lose readers over this. 😆 (I’m told by youngsters that the use of laughing emojis alone reveals my age; it’s apparently so passé.)

The Year Ahead

I’m partly unsure whether to continue Molochinations. As you probably know, this publication is named after the concepts in Meditations on Moloch, reflecting my belief that systems determine outcomes. My hope is that some of these articles help us consider the broader systems around us, and to focus more on changing broken systems as opposed to merely addressing the bad outcomes they predestine.

But the other part of me wants to simply enjoy the act of writing and not be captured by my audience to produce only what’s popular. It’s a selfish act to foist writings about random unpopular things on people… but at the same time, I’m trying not to get lost in chasing the big clickbaity topics du jour.

Please comment below if you have thoughts on how where we could take this publication. I’m hoping for a year ahead where I can continue enjoying the act of writing while at the same time producing things which are worth reading, at least to some niche audience.

Thanks for your support of Molochinations over the past year!