John Loeber 🎢

[John Loeber 🎢 @johnloeber]

https://t.co/cjgTayf00k

[John Loeber 🎢 @johnloeber]

@andrewziperski Acting in financial markets has been unfairly maligned for a long time. It turns out that providing liquidity, managing risk, and enabling for exchange at arbitrarily granular speeds and quantities is a tremendous public service, and one of the great pillars of our civilization

[John Loeber 🎢 @johnloeber]

@jeremiahrogers the question is mostly a joke, the substance of the tweet is in the contrarian-to-consensus time window that you really have to anticipate correctly

[John Loeber 🎢 @johnloeber]

Timing -- predicting the future -- is the hardest thing about being an entrepreneur

Specifically, you want to bet on an idea that is contrarian when you set out, but becomes consensus once you are deep enough into your journey

Many companies die because their contrarian-consensus window is too long. Kite, the AI-first IDE, shut down the week before ChatGPT was released. Tons of early-2020s "AI" service businesses failed only for them to start cleaning up in 2025.

Many companies also die because their contrarian-consensus window is too short. They're the first mover, but that doesn't translate into a durable advantage -- one year and a few million in ARR in, they have twenty competitors and the incumbents are catching up.

The most successful businesses of our day all got that contrarian-consensus window exactly right. And you might ask: is that luck? or skill? The eternal question.

[John Loeber 🎢 @johnloeber]

@varunram killer 📈

[John Loeber 🎢 @johnloeber]

When I was a child and lived far from the mountains, my parents would take me skiing every winter, because it was special and rare

When we lived near the mountains, we never went -- because we could always go tomorrow

And similarly I keep thinking "wouldn't it be cool to make an AI generated video game, something exactly how I'd like it" but this doesn't move me to action at all, because it's so trivially there, the supply is readily available, and that changes the nature of desire

[cold 🥑 @coldhealing]

The technology for ai video games has gotten really good really fast but as far as I can tell no one is making games people actually want to play. Just like "woah look what ai can make now"

[John Loeber 🎢 @johnloeber]

it is exciting to be “cooking”. it is great praise to have “cooked”. but you never want to BE “cooked”. take note

[John Loeber 🎢 @johnloeber]

I really like it whenever Past Me has done something good in the service of Future Me: when there's a problem to solve, but it turns out I already did it long ago, precisely to prepare for this moment.

I feel grateful in those moments! People say "take care of yourself" in ways that are often either empty or immediately near-term oriented, but it is actually possible to feel cared for, by yourself, when the temporal distance is large enough that the person of today is not quite the person of the past. Just as with another person, the Past Me didn't have to spend time anticipating and solving some problem for Future Me, Past Me could've hedonistically prioritized a nearer-term concern, but I did care about Future Me, and now that's nice.

I had to catch a flight on Friday and pack for it last-minute. Most of packing is easy: I can pull together the right clothes in thirty seconds. The time-consuming part is all the finicky stuff: toiletries, eye mask, hat, chargers, IDs, and so on. What a relief it was to find everything already in my backpack. And then I remembered that two months ago, I had specifically made the effort to pre-load my travel bags such that I could pack and go instantly. (I made my flight with three minutes to spare before boarding closed.)

[John Loeber 🎢 @johnloeber]

Maybe the real purpose of an apology letter was to let the recipient know that *someone* has actually verbalized and internalized the error that was committed, and that this will in some way percolate into better decisionmaking in the future.

Interestingly, AI-generated apology letters totally obviate this. I think there are many instances of this, where once we are able to AI-generate the artifact, it turns out that the desired outcome of the process was in fact not the artifact.

[John Loeber 🎢 @johnloeber]

@archiemckenzie_ @martin_casado @a16z @generaltxn Congratulations! 🥳

[Archie McKenzie @archiemckenzie_]

General Translation has raised $27m led by @martin_casado at @a16z to localize the world’s best companies.

Today, @generaltxn serves Ramp, Cursor, and Sierra by translating apps, sites, docs, slide decks, and even product videos into dozens of languages.

a16z invested alongside @upfrontvc, Peter Thiel, Fei-Fei Li, and more. 🧵

[John Loeber 🎢 @johnloeber]

@broyojo42 @leanprover this is so cool 🫨

[John Loeber 🎢 @johnloeber]

@banisgh 🥰

[John Loeber 🎢 @johnloeber]

Agreed, but for what it's worth, I meant something different with my observation:

It is weird that people talk about how much AI is accelerating their engineering process, but then we're just not seeing that in large open-source projects.

And surely you wouldn't see it in all projects because they're run differently, but you'd see it in some. And yet I basically see it in none. Linux desktop environments are exactly as good as they were one year ago. LibreOffice is just as clunky as it has always been.

What I'm really getting at is the question of why the AI software development speedup seems to be so unevenly distributed -- why it's missing in an arena where you'd expect it to be winning easily -- and what those gates to velocity are.

[John Loeber 🎢 @johnloeber]

@wolfie_ please add villagers walking around and doing stuff. i'm imagining Age of Empires style

[John Loeber 🎢 @johnloeber]

@brandonwilson while i always thought that i had preferences exacting enough such that i would enjoy designing/building my own house, this process has taught me that i in fact do not have sufficiently particular preferences to warrant this act

[John Loeber 🎢 @johnloeber]

Never Renovate

Some people go crazy renovating their homes. After having done this for a few months, I understand why:

1. It trains you to be hyperaware of errors. Your contractors are inevitably going to try to cut corners, and you want to both get your fair money's worth and have your vision executed, so you have to really pay attention, to things that normally you never would: was that scratch there before? Is this cabinet level?

Of course, once you're trained in this way, your hypervigilance extends to everything, even far outside your contractors' work: there's a half-inch gap between the doorframe and the floor? Sacrilege! (Doesn't matter if it's been there for 40 years.) A home is a complex thing with lots of detail, so inevitably, no matter where you look, the flaws will be endless.

2. But the other thing you learn is that whatever problems you see, you can fix them! Don't like it? Tear it out and do it again. If renting, you treat things as fixed -- but a house turns out to be shockingly malleable. Literally everything can be changed. You can move lights, pipes, doors, tear out walls, there are no limits.

These are the two core learnings of doing renovations, and combined, they are very dangerous: for people who are not used to saying "good enough, whatever", this becomes an endless perfectionist loop. They'll have you changing stuff forever until you either run out of money or sanity.

Worse, you'll find that changing one thing often implies changing another thing. I have a painting that looks like it's not level even though it is. Why? It's not aligned with the doorframe. Why? The doorframe itself isn't level. Why? The floor itself isn't level because it's 100 years old. But you can't live there while you're redoing the floors, of course. This is a scope-creep, yak-shaving classic: all you wanted was to hang your painting straight and now you live in a storage unit. Beware.

[John Loeber 🎢 @johnloeber]

@AdamMayer After going through this process, 100%. Buying one lamp on your own is fun, but buying ten is a chore. I really understand why people hire architects/designers now -- delegating away a few hours of work a week for a 6-month project is actually very, very valuable.

[John Loeber 🎢 @johnloeber]

@ryantcwynar i did of course run cat6 + conduit into every single room