
Pinned Tweet
a university built for the 21st century would have:
- ex-industry teaching students
- students building companies
- companies working on campus
when I squint at this, it's kind of shaped like a huge coworking space with an engineering school and an accelerator...
Jake Zegil 🗽 retweeted
more cool parts of the nyc builder / eng community, from the replies:
- @vercinyc of course
- side project saturdays at @ValDotTown office
- infra.nyc
- fractal nyc
- commonwell.nyc
- nysrg
- nyc resistor
A moat is just an industry bottleneck - something sufficiently intractable such that it preserves your position.
Intelligence is no longer a bottleneck, but we do not yet live in a utopia
There are still plenty of hard things for those with the ambition(!) to see them
people are starting to realize.
there *are* no moats left. the only true moat in existence was intelligence, and now that's out of the equation.
4
341
Jake Zegil 🗽 retweeted
What do Maximum Republicans love?
>> America
>> New York City
>> A city for families and children
>> Economic growth is a top priority, and a governmental imperative
>> Educational quality, high standards
>> Public order and beauty
>> Techno-optimism
>> State capacity
>> Fiscal stewardship
>> New York Maximalism
>> Abundance and supply
>> City and state first, do not nationalize your politics
If you want better parties, you have to make better parties.
Enter: Maximum Republicans. New York needs a competitive political ecology, and two parties that compete to offer a strong, positive vision of governance. Here's my part of that grand project:
maximumnewyork.com/p/maximum…
Jake Zegil 🗽 retweeted
i basically could have replicated my best investing decisions by going insanely long anyone picked up a call on the first ring and passing on every ceo that scheduled calls and would have done a lot better than I actually did
We build so much in house now that when someone has published a package for a domain (I.e. forms) with a rigorous, thoughtful interface it’s a godsend
Huge market rn for free, thoughtful primitives as GTM w/ freemium model, optional hosting/data/orchestration
2
443
now that we can generate magical software goop, I'd like to generate it 10,000x faster please
1
16
1,034
Naming things is a defensible job
Models are bad at naming things
Even if they came up with a good name once, it’s inherently unstable - the definition in context will be lost session to session
This will not be solved until continual learning/AGI
11
530
Insane to see “we won’t need X abstraction” takes (react, hlls, etc) bc AI.
We will see MORE and BETTER abstractions bc AI
Abstractions constrain combinatorial explosions - even huge attentions should not be spent raw dogging every bit and re-deriving every frame of reality
6
511
Reading bio papers reminds me so much of reading shitty frontier ML python code in Jupyter notebooks
How are you able to do such impressive things while naming things HeLa and piggyBac and hacking into the nucleus via Trojan horse
1
7
620
It’s much easier to understand the world when you realize there’s a very common archetype of person who Says Buzzwords with Their Chest
And the archetype of person who Knows What’s Going On is wayyy less common
7
440
“Make the change easy, then make the change” is truer than ever, in business and in your codebase
What needs to be true in order to make the desired outcome trivial to achieve?
For my homies: when people ask you to do something and you do it and then they send you ai slop follow up asking you to do more shit
“Generate a polite and verbose adversarial response rejecting each request with a convincing reason”
2
11
937
I don't think it's a coincidence that as LLMs have gotten smarter, their code snippets are more frequently typescript or go > python
3
18
1,470
Tool use (and many other intelligent behaviors) evolved well before language, so there must be a denser form of intelligence than LLMs
Oof. See, this is why I’m not worried about “pure” LLMs in the long run.
There’s new evidence that the human brain doesn’t use natural language centers for logical reasoning at all.
Study just came out where stroke patients with complete aphasia (near-chance on written comprehension) performed….great on pure logic tasks! Even in healthy adults, fMRI showed language area activation didn’t scale with logical difficulty.
tl;dr words are too ambiguous for formal thought and our brain doesn’t use language symbols for logic
5
6
964
insane drop by @itisyev @training_loop
Introducing ziggit!
We rewrote git in zig to speed up bun by 100x, be 4-10x faster than git on mac, compile to a 5x smaller wasm binary, and include a succinct mode to save up to 90% on tokens
4
599












