Gazette

Vol. 72 Sunday, 7 June 2026
Vol. 72 7 Jun 2026

The author shares their journey from being an electrical engineer to becoming a programmer, highlighting that programming initially felt unnatural to them. They emphasize that many successful programmers have non-traditional paths and that skills can be developed over time, regardless of early talent. Additionally, they stress the importance of learning how to solve complex problems and the benefits of collaboration in programming.

7 highlights

writing a tic-tac-toe game and AI)

First, no one's going to stop you from spending time reading at work or spending time learning (on most teams).

Second, I learned that there are a lot of engineering problems we can solve without anyone knowing why.

I learned which has been even more useful is the fact that writing a test-case generator and a checker is often much more productive than the manual test-case writing that passes for automated testing in most places.

the meta-skill of not being afraid of low-level hardware comes in handy pretty often, especially when I run into compiler or chip bugs.

Bug tracking is another meta-skill that everyone thinks they have, but when when I look at most projects I find that they literally don't know what bugs they have and they lose bugs all the time due to a failure to triage bugs effectively.

If I hadn't interviewed regularly (about once a year, even though I was happy with my job), I probably would've wondered if I was stunting my career by doing so many different things, because the big chip companies produce specialists pretty much exclusively. But in interviews I found that my experience was valued because it was something they couldn't get in-house. The irony is that every single role I was offered would have turned me into a specialist.

  • 2025 letter #
    , 4 highlights

    Moravec is better known for observing that what’s hard for robots is easy for humans, and vice versa, but in a note titled “Bombast,” he marveled:

    The enormous shortage of ability to compute is distorting our work, creating problems where there are none, making others impossibly difficult, and generally causing effort to be misdirected. Shouldn’t this view be more widespread, if it is as obvious as I claim?

    However hard I try, I don’t think my descriptions will even move you much. You need to pick your own test. It should be a problem you know well, one you’ve worked on for years, one you’re supposed to be an expert in. You need to predict the result, year after year. Then, you need to watch the AI confound those predictions anyway.

    Humans are supposed to be clever. This impudent “large” model, this jumped-up matrix multiplication, it can’t possibly understand all the dearly-won nuances at the state-of-the-art. There are confounders that are supposed to stop it, like data and algorithms and other bottlenecks. It shouldn’t be this easy. It would make more sense if one of those reasons for why scaling is supposed to break had kicked in by now.

    The pandemic’s arc is also instructive. That it would be a big deal was public information long before it was physically obvious. In late February, a week before the market began to crash, and weeks before its lowest point, Centers for Disease Control director Robert Redfield admitted that coronavirus was “probably with us beyond this season, beyond this year.” I remember leaving college for ostensibly a long spring break. Friends embraced with “see you in a month!” Only a few understood that we weren’t coming back. You could conceptually know what was happening in Wuhan, in Northern Italy, but it was hard to face. In a few weeks everything changed, until the whole world was in the same story.

    At the end of this Matryoshka doll of scaling curves is progress itself. Progress is a smooth trend that obscures jagged details. The historical record attests to thousands of years of stasis, before hundreds of years of growth. This smooth trend belies countless details of contingent and precarious human ingenuity, and countless confounders like the data and the algorithms that make it exist. Progress is certainly empirical. There’s no reason it had to be this way, but now we can hint at one—progress improves on itself. We have so many good reasons to hesitate extrapolating progress, why every time we’ve made the jump to a new regime of progress in the past, it was industry and accident and not providence.

  • Abstraction, intuition, and the “monad tutorial fallacy” #
    , 3 highlights

    when presented with a mathematical definition for the first time, most people (me included) don’t “get it” immediately: it is only after examining some specific instances of the definition, and working through the implications of the definition in detail, that one begins to appreciate the definition and gain an understanding of what it “really says.”

    The problem, of course, is that if Joe HAD thought of this before, it wouldn’t have helped: the week of struggling through details was a necessary and integral part of forming Joe’s Burrito intuition, not a sad consequence of his failure to hit upon the idea sooner.

    What I term the “monad tutorial fallacy,” then, consists in failing to recognize the critical role that struggling through fundamental details plays in the building of intuition.