Chad Solanyk

Earth observation and beach volleyball

[Chad Solanyk @chadbryant274]

What's the "wait calculation" for hardware companies building
Invest now or wait till it's cheap then build

[ᴅᴀɴɪᴇʟ ᴍɪᴇssʟᴇʀ 🛡️ @DanielMiessler]

Hey @mattpocockuk , love your stuff.

I feel like you’re for sure in the top 5 of all hardcore developers who went on AI. And one of the best at explaining your approach as well.

But I now wonder / worry that you’re absolutely maxing AIs capabilities in the OLD way we build software, but possibly hurting the overall ability to build good software using AI.

Your set of skills on Guthub is completely insane. Best of the best. But they seem very much oriented around separating dev into the traditional pieces we’ve always used. Specs. Plans. PRDs. Etc.

And also keeping the engineer in the loop to a significant degree.

So my question to you is whether you’ve started thinking of a Bitter Pill version of what you do.

In other words something way lighter that gets more out of the way of the model, and that focuses way more on capturing the WHAT of what you’re building vs. the HOW.

I’ve personally used dozens of different skills and systems for doing AI dev since 2023, and I don’t even have a development skill anymore.

I’ve got a single system for articulating and hill-climbing towards my desired output, and a bunch of super light skills that get my desired tools and configs, and I let the model handle everything else.

Are you messing with any lighter approaches like this?

Of anyone I can think of who I’d love to see a bitter lesson approach from, you’re at the top of the list.

🫶

[Tim Sweeney @TimSweeneyEpic]

One thing that's abundantly clear when doing general vibe math in Programming Language Theory is that one navigates related fields and topics at a vastly higher rate than when doing old school research, maybe 100x or 1000x, because it's so much easier to explore related topics across fields at varying levels of detail. In the old days, each sidestep would require buying books, finding papers, and spending days or weeks getting up to speed. Now just minutes.

This has the effect of exposing the missing superstructure connecting related fields. A few of the random things I've found:

- Programming language theory denotational semantics meets universe polymorphism via Reynolds parametricity to keep sets compressible.

- Parametricity meets clone theory to explain uniformities in both types and values across languages.

- Set theory with elementary embeddings meets nonwellfounded set theory to explain universe polymorphism in multiple ways.

- Positive set theories like GPK+/infinity meet parametricity to explain uniformity topologically.

- The surprisingly-missing denotational semantics of mathematical notation, and in particular how set theory and functional logic programming are the same thing separated by missing denotational semantics mechanisms.

This is all just in the context of formalizing the Verse programming language. I bet anyone working in similar fields is finding similar results.

I imagine AI model makers could do a lot of good by accumulating an open body of work describing the literature of each of the 10,000's of fields of academic study, populate it with papers and relatedness details over time, and attempt to have AI map out and accumulate the missing connections among them. There are probably a million breakthroughs that could be made but haven't yet due to old school interdisciplinary friction and limited resources.

There is definitely an opportunity to advance frontiers (albeit speculatively, until the work is validated or proven) far faster than the historical rate of academic publishing.

[Daniel Jeffries @Dan_Jeffries1]

The answers to this are simple:

1) Either you believe in freedom, and many intelligence agents (the human kind) making their own decisions about how to use power tools or you think that a closed, tightly guarded, authoritarian path is better.

We believe that most people can be trusted with kitchen knives to cut vegetables and you punish the few people who stab someone, like everything else.

AI is not magic that somehow magically transscends the entire history of civilization and technological development and economics.

2) I don't believe that their are "special people" with special knowledge who can guide and steer intelligence or the economy more widely than the distributed intelligence of mankind.

This didn't work in the Soviet era and it won't work now.

3) Better defense against risk is when everyone has equal access.

When Huggingface was under attack they had to turn to open models to defend themsleves because the closed model and their ham fisted safeguards infantalize people and block legitimate use.

We cannot have the bad guys with better tools while we have crippled, gated tools.

4) We believe in the Proactionary Principle, not the Precautionary Principle.

Precautionary means we have to prove a negative that something may or may not happen in the future. This is nonsense.

We legislate real things when they happen in the real world. Harms must be actually demonstrable not in people's heads and imaginary and most of the harms are exactly that, imaginary, kind of mass hallucination.

5) There are many smart folks on the labs but smart is meaningless.

You can be a genius in one area of your life and have no actual ability to effectively predict the future or understand how a technology plays out. Raw intelligence does not equal wisdom.

Many of the foolish predictions have continued to play our exactly the opposite of how "smart" folks predicted, like jobs. But we are just supposed to take the rest of their fantasies seriously?

6) In fact many of the folks at the labs are actively and deliberately and transparently lobbying for regulatory capture.

This is totally and completely unacceptable.

Win on merit. Win on competition.

You do not get to win with protectionism and locking out competition and delivering an inferior, gated, crippled product for "our own good" and then legislating that is how it has to be.

[Stephen Pimentel @StephenPiment]

@emollick Nope. They believe that open weights models are the best way to guard against those risk.

If you believe that regulation and restriction are the best way to do so, that’s a belief about something other than AI.

[Chad Solanyk @chadbryant274]

Hardware companies have 5ish? years to take advantage of using AI to leverage their hardware moat. After robotics comes like AI has to software, the moat will move to voblen goods
Hackers can get 80-90% of the way to pure software now. In 5 years, 80-90% of hardware.Act now!

[Vivek Ramaswamy @VivekGRamaswamy]

There’s never been a starker contrast in the choice for Governor. Winning vs. whining. Creating wealth vs. destroying wealth. Lower taxes vs. higher taxes. More energy vs. less energy. It’s pretty simple actually.

[Elon Musk @elonmusk]

Real photo of Starship taken by Starlink satellites

[Chad Solanyk @chadbryant274]

Small teams with agency > "agile" teams
Agile is dead

[NASA Administrator Jared Isaacman @NASAAdmin]

Pretty clear from this shot that @SpaceX knows where we are going! Congrats on getting Flight 13 underway. Excited for what will be learned from this mission. When Starship comes online, its capabilities will be game-changing, not least of which will be ensuring we never give up the Moon again!

[Elon Musk @elonmusk]

This has my full support.

Jensen is right.

[Chad Solanyk @chadbryant274]

@petewilz Freedom is open models + compute 🇺🇸

[Jensen Huang @JensenHuang]

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.

AI will transform every industry, power every company, and be built by every country.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.

The world needs both frontier closed models and frontier open models.

https://t.co/AUKzoQ5Ikb

[Christian Keil @pronounced_kyle]

I just got pushup-mogged by my 7-month-old daughter

[Matthew Berman @MatthewBerman]

GPT-5.6 Sol Medium is my most used model.

[Chad Solanyk @chadbryant274]

@mikepat711 It's not clear to me his claim that China stopped giving out AV licences was due to young people uprising. Seems more to do with the fact that their AV all of a sudden stopped at the same time in the streets

[@levelsio @levelsio]

Incredible interaction here between @_mattwelter and @pedrotech_ showing exactly what's going on

No judgement on both btw, just interesting to observe:

Matt complains how he improved his product 10x but revenue is down by 1/3rd. Shouldn't it go up if you make the product better? The thing is every competitor also made their product 10x better using AI. So AI raised the baseline quality of everyone's apps (easier to make great apps) and increased competition (easier to enter the market)

But then one of his customers @pedrotech_ replies how he loved the app so much he decided to vibecode his own version for himself "so he didn't have to pay"

This exactly shows:
- everyone thinks they're improving their app but everyone is improving the same amount with AI (but not many realizing), it's like a race but everyone gets nitro now without realizing everyone else has nitro too
- barrier to entry is gone, anyone can just clone an app and enter as a competitor
- personal software is here, as he just cloned an existing app to not have to pay for it anymore
- he then only pays for API tokens for LLM or image models to the model hosters and model owners

So you went from a business chain of
- customer > app > ai model provider

To
- customer > ai model provider

The app layer is possibly getting removed!

[@levelsio @levelsio]

Also some thing I forgot to add

There's so many things now where I don't even use the app anymore but just generate my own app to do things

Like WHOOP, for years I wanted to know the correlation between stuff I do and my workouts and sleep, like going to sauna, or what I eat. WHOOP has a journal but it's practically unusable, I logged sauna for ahwile, then I realized you have to ALSO log the days you DO NOT go to sauna, by hand, every day! Useless

So in those cases I just ask Claude Code to build a little web app for me that pulls all my data in from my WHOOP, and what I eat, my @wip logs when I go to sauna, or go tan, or go gym etc. It just pulls all the data in and gives me the answers to questions I have

Half the time I don't even use the web app it generates, I just ask it directly stuff

I have to find a hotel now for some trips we go, I didn't even go to my own site Hotelist, I just asked Claude Code on my Hotelist VPS server to find me a hotel, etc.

So many times where it just saves more time to ask Claude Code directly stuff than try to do it with some startup/company's app

That's what I mean, I'm obviously more tech than regular people, but we usually slightly ahead in how we use tools, and if I stop using websites and apps, and just use Claude Code for everything in my life, it's not without logic to think that something like that will be the future for most people

And of course it's already happening with Claude Cowork and Codex app which regular non-tech people use

My point is that the entire app and website layer is just getting decimated

And me being a person who lived kinda in that layer (well more websites ofc to be honest) means I have no clue where it's going to except this, or what a person should do, for the first time, I have no clue

The only thing I know is that the superintelligence clanker can't reach into IRL and hardware yet for the next few years at least, and there's lots of stuff there to do that's differentiating, because you need a space, tools, machines, make things, and no you can't generate that in 3 minutes with an AI

Blood, sweat and tears are not automated yet by AI, yet!!!!

[Chad Solanyk @chadbryant274]

@gdb @OpenAI Computer use does my time sheets for me 👍😅

[the tiny corp @__tinygrad__]

The time is long past to drop all export controls on NVIDIA. Some Americans are foolish enough to think they can win an AI race against China. That's unlikely. If the world switches to Chinese chips, the US will be dependent too. Let's make a global market for AI for everyone.