reply guy philosophy / markets / AI
Joined May 2024
- Tweets35K
- Following1.7K
- Followers360
- Likes80K
excel_rator retweeted
Actually things aren’t fine, sorry. The singularity is currently happening against the backdrop of the disintegration of the western political and ideological order
There’s nothing deep or intellectual about presentism. It genuinely doesn’t matter if your life feels “normal”
excel_rator retweeted
MAGA data guru Data Republican says the early vote data is so depressing for the GOP, she can’t share it with her audience.
I've been asked about early voting analysis.
The dilemma I have is that my platform is big enough such that being too blunt can have an effect on morale. I'll push out new states and tools out to my website soon, but won't announce them here on 𝕏.
I will be giving @realJeremyCarl my predictions a week in advance so he can hold me accountable after Election Day.
You MUST turn out and bring their friends, and brace yourself on election night. Even TX is going to be a fight.
excel_rator retweeted
so many gems and great tips in this talk from @yminsky about how to improve your codebase for agents
• agents thrive on feedback, and types are one of the best kinds
• make illegal states unrepresentable with types, and it enforces global rules across your codebase
• property based tests, fuzzing, and even full-system deterministic simulation testing
• expect tests: just print stuff out and capture it in the test. makes behavior legible. it's very similar to how compilers have fixture tests, it reduces the review surface area
• formal methods used to be very expensive: seL4 is about 8,700 lines of C and took around 25 person-years, with 23 lines of proof for every line of C code, but agents change the roi
youtube.com/watch?v=rUYP4C29…
excel_rator retweeted
Dario was right when he said in March 2025 that most of the code will be written by AI.
We are in October 2026 and 100% of the code is AI generated. Models are so good at coding it’s a liability to code manually.
excel_rator retweeted
This is happening across the board based on my conversations with junior analysts and associates
However, will add a couple of caveats
1. Modeling was never the bottleneck in these jobs. It was already a commoditized skill, with the real value being able to draw conclusions based on the analysis in your model
2. The tool will only ever be as good as its user. Ask someone who knows nothing about finance to build a financial model with Claude, and they will still produce something subpar compared to someone with years of experience in the industry
excel_rator retweeted
You can flip two bits in an LLM’s head and it starts fantasizing about killing you.
excel_rator retweeted
This is the best advice I have as well. You need to be trying stuff that is much more difficult than you expect and AI to be able to help you with. It is very easy to underestimate what the models can do now. Even as someone fully down the rabbit hole for a long time now, I struggle with this continuously. Basically whatever you are asking it to do, try the 10x more ambitious version of that. Maybe even 100x. This isn’t hyperbole. Like try the most ridiculous thing you can and see what happens
excel_rator retweeted
it's really unpleasant to see so many people making fun of her when this is a totally valid crash out. she's likely seeing less work, and on top of that, she's fighting for "the cause" perpetuated by bigshot twitter artists who can afford all the virtue signaling they usually engage in.
she repeats their talking points--"We know GenAI is theft," "We know GenAI is bad for the planet"--and can't bring herself to embrace new tools because doing so would make her feel like a traitor. there are comments under her original post asking her to "keep fighting GenAI," as if they genuinely believe all of this will come to pass automatically. so she's really caught between a rock and a hard place.
i genuinely feel sorry for her
excel_rator retweeted
willpower is like yelling at your child to overrule a conflict. it can be useful, but there is a long-term cost. ideally you'd already be on the same page
excel_rator retweeted
The right simply doesn't recognize any election it doesn't win as legitimate. In their soggy, conspiracy addled brains, the right can't be beaten, only cheated here and everywhere. Unfortunately this is just a feature of modern politics now, and there's a whole garbage ecosystem to support it that includes media organizations and nonprofits. It's simply a fact of life now that we have to deal with this every single time an election doesn't go their way.
excel_rator retweeted
At 37 Patrick Collison listed the reason not to start a company in every decade since the 80s.
Japan will win everything, Microsoft will dominate, the dotcom crash, Google, then Facebook, now the AI labs:
Every single time it was empirically a great time to start a company.
He arrived in Silicon Valley in 2010 and people were already saying the funding was drying up.
For more than a generation, betting on entrepreneurship has been a durable bet.
Save this.
excel_rator retweeted
i love that all these companies doing this predatory corporate surveillance are getting dickmogged from orbit right now
While I reverse engineering my Wi-Fi router, Claude found f*cking scary privacy problem...
Internet provider sees every site I open and can change anything in my router without asking me
here's what else it found and 5 steps to fix yours:
what Claude found:
• my router keeps a list of 29 devices that ever touched my Wi-Fi and see when new connects
• my provider can reach into my router any time, Claude found channel hiding in my connection name, and I have no way to turn it off
• every site I opened was looked up though my provider in plain text
• my provider's app on my TV box kept sending data while TV was off
• and much more...
and it's not just me:
- about 1 billion devices have this remote channel (University of Innsbruck)
- researhers pulled 2 billion+ router locations from Apple's Wi-Fi database, accurate to meters (University of Maryland)
how to fix yours:
1. install repo below (It also helps improve Wi-Fi speed) and send Claude Code a photo of your router
2. change admin password from the one on the sticker
3. turn on encrypted DNS, your provider stops seeing your lookups
4. turn on Private Wi-Fi Address, your phone stops showing its real ID to every router
5. add _normal to your Wi-Fi name, Apple and Google stop adding it to their location maps
Do it as soon as possible. Then read article below on how to find these problems in other tech:
excel_rator retweeted
In my book Kakistocracy, I note that Brazil locking up Bolsonaro after he attempted a coup put its democracy on a healthier course. Compare to America, where the prosecution took too long. Brazil dodged a bullet tonight.
You can be right-wing! Stop trying to be authoritarians.
excel_rator retweeted
Having Tourettes while sitting between your mistress and wife for 4 hour livestreams daily is like some high level CIA mind breaking torture shit
excel_rator retweeted
A CNN architecture from 2015 beat GPT-6 Astra and Claude Fable 5.1 on skin lesion classification.
Dermatologist Magnus Lynch fine-tuned Inception v3 on 68,011 images in about 20 minutes using one H100.
On 200 biopsy-confirmed lesions:
• Inception v3: 62.5% accuracy
• GPT-6 Astra: 42%
• Claude Fable 5.1: 42.5%
But there’s a catch.
At its preset threshold, the CNN would have discharged 18 of 30 melanoma cases, versus 3 for GPT and 4 for Claude. Adjusting its threshold changed that tradeoff substantially.
This was a small test comparing a fine-tuned specialist with general-purpose LLMs at default settings.
The specialist won on diagnosis accuracy. Whether a lesion could safely be dismissed was a separate question.
Choose your benchmark carefully before declaring a winner.
