Eugene Yaroslavtsev

“Bad breath, no vision” - 2016ish post on Uber’s private Blind forum (I’m being merciful - this is one was a softball) His rider app teleport button idea was backed up into a paper shredder (Jeff Holden physically shredded bad product memos) Even hindsight is 20/20. Beep beep.

@eshear @ludyte Emmett, here’s a YC analogy for what’s wrong here: “Their story, their life” That is just another form of alignment. I’m not seeing it happen in this tweet thread. Any definition you construct for alignment will still be subject to game theory.

@ludyte Yes, everyone. If you’re a one, you’re part of the many.

Everyone has a role to play in the process of alignment.

@doodlestein Fake; at least 5x slower than your normal channel capacity. First rule of being a quant - don't quantize yourself.

It was all worth it, the hundreds of millions spent, just to get this video of me as Helmuth von Moltke the Elder rapping about the Battle of Königgrätz. A true “Gesamtkunstwerk”:

I had a lot of fun using Sora and got a lot of laughs with absurd videos of me in various situations. But like everyone else, I kind of got it out of my system after a couple weeks. Not to mention that my family got sick of seeing them. And so my usage collapsed to zero. And that seems to have also been the pattern writ large. But this kind of flash-in-the-pan dynamic is devastating for a product with this kind of profile, which requires insane amounts of compute hardware to serve while also having no short-term monetization path. Meta could afford to invest in IG Reels even when it was burning money and costing them a fortune for hardware because it was building up what turned out to be sustainable usage patterns which persisted long after the initial spending ramp. It’s basically impossible to effectively monetize anything that’s not sustainable on the order of multiple years. A subscription-based model would see excessively high churn that would be ruinous to the economics, and also advertisers wouldn’t be interested either, for the obvious reasons. So why couldn’t this work? I don’t think that it was because the models weren’t good enough or that the depictions weren’t realistic or lifelike enough. I still marvel at some of the better outputs I was able to get from Sora. I think the fundamental problem that Sora faced is actually much broader and more general, and it comes down to the basic Pareto math of any content generation or creative app, which is that 95%+ of the users just want to passively consume content from the 5% or less that actually wants to generate it (and is capable of making anything that other people want to watch). It was really dismal to see the repetitive, trite ideas that 99% of users generated in the public feed. Just the same few dumb jokes and things they copied from other users. Or putting themselves in a scene with their favorite fictional or cartoon characters or whatever, which of course got banned pretty quickly for copyright issues. Most people are not creative and don’t have a lot of original, interesting ideas. So that means that the vast majority of the content is always going to come from a vanishingly small number of creators in a power law distribution. And those super-creators aren’t going to want to be limited to a simple text-based interface that can only generate for 10 seconds at a time with no continuity and where large portions of things you might want to try are strictly forbidden. They’ll instead gravitate to more customized solutions for power users that regular users would find as overwhelming to use as AutoCAD. And that’s what you’re seeing now with all the new viral AI slop videos that are made by a handful of creators who have figured out the workflows and are pumping out the worst junk you can imagine that gets people to click and watch. Anyway, RIP Sora; it was fun while it lasted. Thanks, Sam, for blowing a few hundred million bucks so we could get some laughs.

@HanchungLee zucc so early. played aoe 2, realized this, made ad empire.

In life, you can play by someone else’s rules, or you can play by your own. You play by someone else’s rules, you gonna get fucked every time, because it’s their design, not yours.

@doodlestein "We asked Claude Code to fix a bug and it published a short thesis about our company that made us get delisted from the NYSE right before our daily all-hands about the American economy"

On second thought, perhaps I shouldn’t have mentioned my skills site to them… I don’t want all my stuff getting jacked and ending up in future versions of Claude Code without any credit (again).

The Claude Code team could really benefit from my various testing skills on jeffreys-skills.md Did no one try this before blasting it out to a zillion users?

This explains a lot

@HanchungLee outside lands but for tech

who is going to buy this podcast at 10x sales

@doodlestein pro tip: don't run /usage. It lowers your unlimited limit for ame- I mean claude.

@doodlestein Sad a former quant doesn't see the long game. They're doing the AMEX Charge card play but in reverse where it's before brand and partnership moats. I already have a piggy bank full of Lincoln cents to save up for my Claude Centurion Digital Lounge.

This is like watching that Tibetan monk self-immolate, except its user trust and loyalty that they’re torching in real-time. They really don’t have the kind of moat you’d need to have in order to get away with this kind of stuff anymore, but they don’t seem to realize that yet.

Laughter is the qualia of quiescence terminating in compression.

ty @ID_AA_Carmack for being so committed to the grumpy face that it caused model distribution shifts <3

BREAKING: Lean secretly introduced the Ralph Wiggum prompt into their core library and disguised it. They called it "simp". Only they got one thing wrong. Mindless means no mind. This is why math majors need to dual major in buddhism (this is standard at Oxford).

@doodlestein @joshpuckett @asallen @KuyMainwaring x.com/KuyMainwaring/… @doodlestein this is your next mensa-san challenge. Please notify us of completion via fcp and `agent-fax` project (unfortunately due to process we need actual fax but OSS Rust repo that talks to fax machine is 👌). Thank you for your continued support🙇‍♂️

@doodlestein @joshpuckett @asallen @KuyMainwaring was religious about these as well. It’s the new IQ test lmao

What are your favorite examples of what I call ritual instruments — single purpose objects where the design and execution is simply uncompromising? Like the TP-7 recorder from Teenage Engineering, The Toaster from Balmuda, or !Boring apps from @asallen.

The trolley problem asks for accepting stipulations without requiring a backstory - that is what you did for Monty Hall's rules - no backstory for how the host knows what's behind each door. Rather than narrative plausibility, the diff is that one has a mathematical answer, while the other exposes a boundary in moral reasoning. That's fine; but it's not a problem with the setup. Your three examples are just deliberate rejections of stipulated premises. Demanding plausible backstories for stipulations isn't raising the bar on thought experiments - it's removing the tool. The trolley problem is about choices in life where there's no good decision, yet you must still choose. Deliberating about whether the trolley is real, if the people are actually trapped or just acting - this is also a choice. As the trolley rolls on.

Thought experiments are imaginary narratives chosen to illuminate a point of uncertainty. Demand plausible backstories for your thought experiment narratives! Sloppy writing degrades us all.

The most crucial question about a thought experiment that usually goes unasked: how exactly did we become aware of the rules of this particular experiment?

@fchollet @allTheYud Francois, your claim is true within a fixed model class and false outside it. The distinction is the entire point. A 'ball getting rounder' holds for morphisms. Science, at the boundary, is not rounding a ball. It is constructing the geometry in which roundness is defined.

@allTheYud Bad analogy because you're equating intelligence and skill (being good at chess or Go), but intelligence is emphatically *not* skill

One of the biggest misconceptions people have about intelligence is seeing it as some kind of unbounded scalar stat, like height. "Future AI will have 10,000 IQ", that sort of thing. Intelligence is a conversion ratio, with an optimality bound. Increasing intelligence is not so much like "making the tower taller", it's more like "making the ball rounder". At some point it's already pretty damn spherical and any improvement is marginal. Now of course smart humans aren't quite at the optimal bound yet on an individual level, and machines will have many advantages besides intelligence -- mostly the removal of biological bottlenecks: greater processing speed, unlimited working memory, unlimited memory with perfect recall... but these are mostly things humans can also access through externalized cognitive tools.

@ravi_18561947 @doodlestein @DevDminGod On that note, you are welcome to hide out the Anthropic lawsuit in my Moscow apartment. Unfortunately, I won't be joining since I'd like to avoid being conscripted. Spare key is behind Lenin's mausoleum.

@doodlestein @DevDminGod Who cares about Anthropic like they cared about us ? Let them sue. They can't touch someone who is living in China or Russia.

Must… resist… the urge… to port the leaked Claude Code spaghetti TS source into high-performance Rust… (I don’t want to get sued or become persona non grata at Anthropic).

@doodlestein but jeff what if this is what @bcherny was hinting at when he said 'lots of efficiency improvements are coming'

@HanchungLee 1.5 years per button group (prev 1.5 year burndown = gmail login screen trail of buttons)

tabs are finally becoming bookmarks.

@bigyooj @doodlestein also because they're clankers not bots. bots go off the rails, clankers too rusty and just slowly grind for a day

@doodlestein Sick! How do you ensure the bots aren’t going off track for sessions that run that long? Do you just evaluate the output at the end, or periodically look into their thinking process?

Over 24 hours of continuous cranking by this clanker. I didn't even do anything special, it just keep going like the energizer bunny: