martythemaniak

There's been a massive change to public markets in the last decade and the retail path to making money seems to have closed. I made a some money on IPOs using a laughably simple heuristic:

"Is the company market cap low? Do they have a decent product? Is it plausible they'll 10x? Yes -> Buy some amount I can afford loosing"

For example, Tesla IPO'd at $5B cap, it was perfectly plausible to believe they'd be worth $50B some day. Shopify IPO'd at $1.3B, Square at $3B, 10x was perfectly believable. Uber IPO'd at $75B, I did not believe they'd be worth $750B any time soon, or ever. Do I believe SpaceX will be worth 20T in like 10 years? Lol. Fmao even.

Today's IPOs at $1T+ means that private money figured this out and cut the retail public out, IPOs seems to be a really terrible deal these days.

Well, the 90s were a time of liberalization/deregulation so definitely a moving target, but the monopolies/layer of mediation were held back by variety cultural, regulatory and legislative checks. Glass-Steagall was repealed in '99, fairness doctorine was repealed in '87, but the mindset and expectations remained in place, the idea of Citizens United would have been considered absurd, agencies had stronger regulations, etc etc. Today you have a handful of people who control these companies (non voting shares) paying money directly to the president in return for preferential access. Laws and the public interest are not a part of any consideration.

Here's a practical example: your gmail account today is probably more important to you than your phone number was in the 90s and you can run afoul of some random ML subsystem and lose access. There is no recourse or accountability. Randomly loosing your phone number and not having recourse was not a thing back then.

But yes, I was on the internet in the 90s, the little playground we fled to has grown into a terrible panopticon run by unaccountable people for their own personal interest. NIMBY DC opposition is a terrible proxy to tackle this, but it may be the only tool available.

Here's a view that I've not seen AI/DC proponents engage in (for example, Carmack's recent pro DC post)

AI is an exciting and promising new tech, much like the web/internet in 90s and smartphones in late 2000s. Back in those times, the tech industry was far, far smaller, tiny in the 90s and maybe like 1/20th of the current size in the late 2000s. Tech companies were not a big part of people's every day lives, so these technologies could be seen as something exciting happening off to the side that you didn't need to engage it if you didn't want to.

Today, Big Tech is absolutely ginormous and huge parts of people's lives are mediated by one of a half dozen companies that together form an interlocking set of barely accountable duopolies. It is this overbearing unescapable structure that is causing the backlash, because many people understand intuitively that this exciting new tech will be leveraged against them in every way possible by this structure. We cannot treat AI as neat new thing to play with, experiment with, find novel uses for, we have to put our guard up and defend against Big Tech and DC opposition is a very easy and straightforward way. DC opposition is also highly compatible with existing NIMBY networks and mindsets, which are bipartisan and widespread. Thus

All that is to say is that it's not the technology, it's that bad people are in power and are weilding it to make your life worse in myriad ways - layoffs, increased electricity rates, slop, etc.

Restricting access to guns is actually a pretty mainstream thing in most of the world.

I wonder if he'll ever realize that his silly definition of "freedom" is precisely what enabled the techno-oligarchy he rails against.

I ran across this two-part interview with Yann Le Cunn and I thought it gave a pretty good broad overview/background into JEPA: https://www.youtube.com/watch?v=kYkIdXwW2AE

Lol, this is so Q2. Today you need the LARP Orchestror to spin up a dozen SubLARPs, then hook them up to an external ledger verifier so you can run them in a Enron loop.

I clicked expecting Bane Proctoring, where Bane monitors your exam and if he catches you cheating...

Here's an answer for you: we need a new temperance movement.

Sounds pretty ridiculous at first, but that's because people know little of the temperance and progressive movements of the late 19th and early 20th century besides "prohibition failed". There was a whole lot more to it than mere anti-alcohol push, it included labour rights and reform (Upton Sinclair's The Jungle, etc), Anti-trust and anti-monopoly laws, anti-political corruption, consumer protection (food regulation), women's suffrage, etc. The 20th century progressives essentially saw a huge concentration of money and power turning society into a cruel, exploitative machine that ground people up and spat them out. Rampant alcoholism was only one part of it.

A modern version would be different in details but similar overall. Alcohol and cigarettes are not major problems, but corruption, monopoly, and engineered addiction are very real and very problematic. A new temperance movement would be about tackling rampant gabling and addiction (everything from gas station drugs to social media algorithms), political corruption, tech monopolies, misinformation etc.

Big problems need big solutions, but I suspect things will have to get worse before people realize the scale of the problem they're facing.

It's dreadfuly simple - Bluesky users can choose what content they want to see - including selecting feeds, algorithms, moderation services etc. This situation is intolerable to people who believe that their right-wing views must be forcefully shoved down users' feeds, because not liking and not wanting to see their shitty content is considered an attack on freedom of speech.

This is a good observation, because this tactic is a hallmark of Putin and authoritarianism in general. What he does just lie about something where he knows it's a lie and the audience knows it's a lie, and he knows that the audience knows that he's lying, but the audience is powerless to correct him, so it is his way of demonstrating his power over the audience. He is saying to them that he is so powerful and dominating that he is in charge of their reality.

Masha Gessen has written a fair bit about this.

Honestly, are people not getting what's going on? The US is turning into a personalist regime, there is no "government" per-se, there's a dude. There are no 'rules' there's only the dude's opinion and you'll do whatever he feels like today.

The way you know this is true is to imagine The Others in power. Sacks used to scream about government interference, but now that he's running (this part of) the government, obviously things are different.

The only constant is that David Sacks (& co) always believed he should have all the power.

The problem is not the technology but the size and dominance of the tech sector. Current AI is a promising new technology that can do a lot great stuff, just like the internet and smartphones. But in the 90s the tech sector was tiny, and even in the late 2000s the current behemoths were like 20x smaller than they are today.

Today the tech companies dominate people's lives and attention, everything people do goes through a few companies in one way or another. The google of 2007 was a plucky underdog who promised to Do No Evil. Today they are so huge, the size itself is scary. Instead of being happy you got 2GB free email, you're slightly anxious about accidentally getting flagged by some automated recourse-less system that'll lock you out of your data and disrupt your life significantly. AI startups don't offer a respite, they small but even more scammy and slight-of-hand-y. AI industry marketing is best summed up as "You're a worthless piece of shit NPC and we'll take your job, your personal data and your attention. Buy our shit"

That is to say, people probably understand (on vibe level mostly) that this great new tech will be leveraged to the max by bad rich people to make their lives worse. Nothing innate about the tech dictates this, but it is the way things are turning out. I expect more backlash in the future.

Disclaimer: I'm a robotics noob, but I've been working on robotics for a few months now.

I'd say virtually all robots you've seen in the real world today rely on classical approaches - you build a rudimentary map, then use classical algorithms to find paths/do area coverage. The robots do no reason or understand what they're looking for, they're more like in-game units. At most there's some bounded, lightweight image classification going on.

LLMs can understand and reason about the world natively. nVidia has a Tesla FSD/Waymo competitor which simply their 7B reasoning LLM but instead of outputting tokens directly, its outputs are fed to a 2B diffusion model that outputs 1.6 second long trajectory for the car, and this is enough for an L2 system. But to make this work, they need the model to run at 10Hz, so they use super high-end hardware to do it (Jetson Thor) and the car is still "blind" for 100ms at a time (they have a parallel classical safety system).

With on-chip LLMs you could run this loop at like 100Hz on a chip that costs a few hundred bucks, rather than 10Hz on a board that costs several thousand.

In a chatbot, 17k tok/s is a neat but nearly useless showcase. In a coding agent it is a meaningful improvement. In robotics, it could be an absolute revolution.

8B models aren't useful in general, but for specific use cases they can provide an enourmous amount of intelligence - nVidia's Tesla/Waymo competitor is a 7B LLM with a 2B diffusion model, and running that at those speeds could be an order of magnitude cheaper than existing solutions.

No, that's a Mercedez EQE. MB is the launch partner for nVidia's alpamayo-based autonomous efforts.

People in the immediate future will be far less happy simply because there are enormous resources being spent to make them so.

The world's first trillionaire spent the day he became the world's first trillionaire celebrating by trying to incite a racial pogrom in a distant country.

I think you are misreading this situation. The Musk-Thiel cluster, of which David Sacks is a major member, is one of the main groups trying to destroy the rule of law and turn the US into a personalist regime. They are doing this because they think they'll be major beneficiaries of the new regime. As they see it, the long term success of the US, ie the rule of law, is against their personal interests.

Yeah, afaik the approach people take today is always some form of bi or tri level hierarchical control, with a slow LLM doing planning and sub task management and diffusion or VLA doing the motor control at higher frequencies. Major differences seem like where and how you draw the boundaries. For my project I'm personally trying to use ROS2 as a low level tool call (instead of diffusion), with an agent /LLM doing the main decisions.

Having said that, this scheme seems like it might just be a reaction to current hardware limitations. When I saw Talaas demonstrate a 8B model running on a custom chip at 17k Tok/sec, first thing I thought was "wow, you can just run an LLM in a control loop"

Yes, obviously. People really need to update their priors on how the US operates now. Laws are out, loyalty is in. There is an extraordinarily powerful unitary executive in whom the will of the people resides and it is the job of the government and society to work towards the will of this powerful executive. There are no checks or balances or alternative centres of power (civil, political, clerical, etc) allowed.

This particular executive loves money, praise, and submission. When Anthropic submits (eg agreeing to whatever the DoD demands), issues public praise and makes appropriate donations (eg ballroom, memecoin, etc), then they can do as they please.

Students of history will find this new MO very familiar and very depressing.

Yeah, there's commercially available snow plow robots, you can buy a Yarbo for your house today. As far as I can tell, they all operate on a classical robotics stack - for the Yarbo you install an RTK antenna to give the robot cm-level precision, define a map and a routine, then the Yarbo can execute that routine by itself.

But can it deal with arbitrary lots without extensive premapping, manage piles, handle obstacles intelligently, correct itself (ie spot needs a second clearing ), tackle windrows, etc? It can't, and my hunch is that LLMs are the first tech we have that can plausibly handle all the various cases that a proper robot would need to handle.

I come from a regular swe background, but I've spent the last few months getting into robotics and trying to build a snow-clearing robot, so here's my noob notes:

First, very much expected. Both Google and Qwen have been building explicit spatial reasoning and spatial output capabilities in their models since last fall, gemini 3 was released with support for outputting trajectories for example. I only took a look at Robonav (more relevant for my needs) and its architecture and capabilities are inline with other similar models (eg nVidia's alpamayo).

Second, the overall architecture they describe mirrors what I've been working on: You have general purpose LLM that takes a look at the works and the task in front of it and reasons to break it down into subtasks and tool calls, and you can think of RoboNav and RoboManip as tool calls here. The harness keeps a memory and manages the context of the LLM and tools and keep looping until the objective is complete.

Consider the task of clearing snow off a driveway using this suite: An LLM (Qwen 3.7 plus) takes look at the driveway and decides which areas to clear. The harness then tells robotnav to go to an certain location, then robotnav takes over an runs in a loop until the robot is that that location. Then the harness tells robotmanip to use the plow to clear strip of snow. The harness will then call the planner LLM to plan an execute the next clearing and repeats until the driveway is clear.

So what' the issues? Well, they didn't release the weights, nor the training scripts so you can't actually use it. But also, it's all very research-y still, the models are "small" but still huge/expensive for current edge hardware. You'd still need lots of data collection, HITL, and fine-tuning and evals to make it work for your task. You'd also need a secondary safety system to make sure the models don't wreck something. But overall, I do expect robots to use an agent/model combo like this in prod in a few years.

Yep, people are expanding way too much mental energy on basic bribery. Anthropic will agree to work with the DoD, WH insiders will get some lucrative pre-IPO allocation and Fable will be magically "fixed" and available again.

Folks, it's just a simple, straightforward shakedown. Anthropic will agree to work with the Department of "War" and give some lucrative pre-IPO allocation to WH insiders and Fable will be back.

I think so too. What strikes me most about "pro fertility" people, particularly the SV type, is that they don't seem to particularly like kids, as in spending time with them, playing with them, deriving joy from them. They seem very enamoured with the kids-as-cogs-in-the-system, with "legacy" with all sorts of other silliness except for the kids as kids.

There's no population or fertility crisis. The world's population isn't going to fall, it'll level off around 2100, nobody old enough to read this will be around, our kids and grandkids can decide for themselves.

Local populations will see very different trajectories, yes. Africa will see population growth and many other places will see steep decline. Societies can choose to keep their current system and take in immigrants, or choose to keep their "national character" (or whatever) and rejig their societies so the remaining productive parts pay for increasing numbers of old people. Grifters (Brexiters, MAGA, Le Pen, etc) will attempt to sidestep such obvious tradeoffs, but they will fail, hastening the decline of these societies. So the only crisis we have is people refusing to deal with reality's tradeoffs.

There are no long term strategies, neither the "establishment" nor the public (ie you) seem to grasp that simple fact.

Americans simply do not wish to engage with reality any more, and that's why they're perfectly happy to elect a TV personality for a president who filled his cabinet with TV personalities who can do nothing more than LARP as statesmen.

I think your comment is another instance of this LARP, where the public can't tell the difference between reality and TV.

There's a simple answer actually, FIFA decided to cash out and their objective is now to get the maximum amount of money out of attendees, even becoming their own scalper.

This video is a good overview: https://youtu.be/ocxngraLbV0?si=XzeJf3Chfsbr_BeD

Protip: you don't need buttermilk, you need milk and acid. 1 cup milk and 1tbsp vinegar do the same job and are already in your kitchen.

Source: Frank Proto's pancakes

https://youtu.be/vkcHmpKxFwg?si=a9GeGHKp0WzTmqPr

Am I the only one that doesn't with about this "crisis". The world's population isn't going to fall, it'll level off around 2100, nobody is enough to read this will be around, our kids and grandkids can decide for themselves.

Local populations will see very different trajectories, yes. Africa will see population growth and many other places will see steep decline. Societies can choose to keep their current system and take in immigrants, or choose to keep their "national character" (or whatever) and rejig their societies so the remaining productive parts pay for increasing numbers of old people. Grifters (Brexiters, MAGA, Le Pen, etc) will attempt to sidestep such obvious tradeoffs, but they will fail, hastening the decline of these societies.

The most talent dense place I've worked had a dead simple process - two one hour chats, one with your potential manager/team and one with the CTO or CEO. If things didn't work out, well you got sent away, probably happened twice per month. There was a particular meeting the CTO/CEO used and if you saw someone meeting with them there on a Friday afternoon, you would not see that person on Monday.

The place was not big - never got much beyond 100 engineers, but produced dozens of founders, VPs, GMs etc at well known companies, as well as engineers with very notable OS projects and lots of high-placed engineering in FAANG companies.