cohost of The Big Hollywood Table Read. i wanna live in a lighthouse.
Cambridge, MA
Joined July 2013
- Tweets13.1K
- Following634
- Followers1.5K
- Likes12.5K
toyotathombudsman retweeted
First base coach talking to a runner: "Ok so stay here until you can go to the next one. If you get to that one try to get to the third one. A colleague of mine is waiting there for you"
toyotathombudsman retweeted
The Beatles 'White Album' is ultimately about fatigue of every kind: physical, spiritual, political and cultural. A soundtrack to the tedious indulgences of the 60's yielding only false prophets and dark portents. You say you want a revolution. It's been a long, long, long, time.
toyotathombudsman retweeted
concept albums are so cool. The Beatles made one where the concept was “what if we were a band” and then David Bowie made one where the concept was “what if I was a guy”
toyotathombudsman retweeted
Been noticing more people using “prolific” to mean “good.” Something to keep an eye on as the decades pass
toyotathombudsman retweeted
Biggest issue: If you're being drafted, people are saying "we want you around for 4 years" & I can't imagine anybody wants Enes Kanter or Royce White around them for even 4 minutes.
We've seen this illustrated whenever Enes is near Turkey or Royce comes close to logic/reality.
toyotathombudsman retweeted
One of the most evil things imaginable to me
🦔AI companies are bulk-buying rare books, scanning them through high-speed machines that cut the spines off, and shredding the originals. A service called ISBNdb facilitates orders of up to a million books and keeps buyers anonymous. Pre-2022 books are premium because they're free of AI-generated text. A federal judge ruled the practice is fair use because eliminating the original means only one copy exists at a time. Anthropic hired the former head of Google Books partnerships to obtain "all the books in the world."
My Take
This got to me. A bookseller told 404 Media that rare books with almost no surviving copies are being fed into this pipeline. Books that survived wars, fires, and centuries of handling are being shredded so an AI can learn to write a better marketing email.
ISBNdb's website literally says "'AI company destroys two million books' is not a headline that generates sympathy," and they still built an entire business around making it happen quietly. They offer NDAs as a feature. They coach clients to call it "digital preservation."
I've covered AI companies scraping the internet, torrenting libraries, and stealing music. This is worse because it's irreversible. You can re-upload a website. You can reprint a bestseller. You can't replace the last three copies of an 18th-century botanical text once someone shreds them for training data. And the judge said it's legal. So it's going to accelerate.
"We shred rare books and offer NDAs so nobody finds out" is a legitimate business model in 2026. What a timeline.
Hedgie🤗
toyotathombudsman retweeted
Here Bond, a cyanide pill disguised as a mint. You can use it to kill yourself in a pinch. And this Bond, is your watch. It’s fitted with a detonator to kill yourself if necessary. And of course your gun, the Walther PPK. Only one bullet though so save it for if you’re captured.
toyotathombudsman retweeted
I would love to see what a post-literate Harvard is like. Maybe smart kids who can’t read will invent some kind of new shape
toyotathombudsman retweeted
You’ll be able to tell the difference between Mexicans in America and actual Mexicans over the coming days because the former will wang on about shit food/“colonizers” while the latter will be posting videos of British aid workers being beheaded by ISIS
toyotathombudsman retweeted
I don’t really know the difference between pro biotic and pre biotic . But this soda trusts that I do. And to be trusted by soda? That’s the warm home of family to a guy like me
toyotathombudsman retweeted
Gfs bday. Drunk off 7% ipas. Playing golf. Whooping her ass. She's sick. Losing voice. Sounds like fran drescher. Food is mid. I love it. I love it all. I'm at war with everything around me all the time. What other way is there to live brother. Comanche
toyotathombudsman retweeted
I just got it - it’s called Letterboxd because you “let her box”, like in Million Dollar Baby, a famous work of Film
toyotathombudsman retweeted
missed opportunity: when Kevin Hart sat down in his chair that said KH, Kamala should have come out from backstage & said “your sitting in my chair you short lil bitchass fucker”. (Standing ovation) Then Kevin pushes a button & she gets slimed like on da kids choice awards 🤣
toyotathombudsman retweeted
My boss chucking: I will definitely check out no country for old men. Now I want to talk about the reason I called you in. You've called out 16 times these last months due to "the entity".
Adopting a southern accent for the first time in my life: Thats correct ma'am
toyotathombudsman retweeted
The hustle, the bustle — and the bodegas that fuel it all. The late nights that turn into early mornings. The hook ups, break ups, make ups and shake-ups. The fact that it may not be easy, but at least you know you’re having the time of your life. And — yeah — the freakin’ pizza.
toyotathombudsman retweeted
This is just like being alive in the 1600s when they got good at making complicated clocks and deduced that every complicated thing in the universe probably functioned exactly like a clock
There's a quadrillion-dollar question at the heart of AI: Why are humans so much more sample efficient compared to LLM? There are three possible answers:
1. Architecture and hyperparameters (aka transformer vs whatever ‘algo’ cortical columns are implementing)
2. Learning rule (backprop vs whatever brain is doing)
3. Reward function
@AdamMarblestone believes the answer is the reward function.
ML likes to use pretty simple loss functions, like cross-entropy. These are easy to work with.
But they might be too simple for sample-efficient learning.
Adam thinks that, in humans, the large number of highly specialised cells in the ‘lizard brain’ might actually be encoding information for sophisticated loss functions, used for ‘training’ in the more sophisticated areas like the cortex and amygdala.
Like: the human genome is barely 3 gigabytes (compare that to the TBs of parameters that encode frontier LLM weights). So how can it include all the information necessary to build highly intelligent learners? Well, if the key to sample-efficient learning resides in the loss function, even very complicated loss functions can still be expressed in a couple hundred lines of Python code.
