overfeed
Markets that consolidated down to the 2 or 3 domestic champions had settled into a comfortable coexistence with their domestic "competition", and were coddled by the government will have a hard time remembering how to compete for real.
To use a food analogy; this approach may works very well for a boutique upscale food joint chasing Michelin stars. This doesn't scale to the size of, say, McDonald's. At that stage, variability and creativity are a negative, and guardrails are good. You want literate warm-bodies that can consistently follow well-established procedures step by step. When your engineering staff complement grows over 50,000 - it may get difficult to find (and retain) those who are "exceptional".
It's not just about scale either, if you're a content, sole-proprietor hole-in the wall operation, (expensive) exceptional people aren't going to move the needle,and may even be detrimental to your viability.
dude-with-hand-on-other-dudes-shoulder-meme.jpg
A lot of his voters saw the cruelty and crassness and liked it. Trump was their vessel for vengeance, as one disappointed Florida voter complained during Trump I "he isn't hurting who he is supposed to be hurting"
"It's shortwave radio"
There's a statement about reporters having to look through their own windows when reporting on someone who claims it is raining.
it takes a lot of of human hubris and incompetence to think you can herd AI agents to squirt out a better version of a multi-decade project in weeks, without a single human in the loop. I can't imagine the levels of hubris in prompting "Make a faster Postgres in rust. Make no mistakes" and immediately uploading the result for Jepsen testing without bothering to manually kick the tires a bit to see if your DB is fit for purpose.
I see a direct line between this attitude and the Proton CEO deriding "loser thinking" in a subordinate that dares suggest the AI output have some oversight before going to prod; there exist people who are so AI-pilled they get lost in the sauce.
Postgres was already mature by the early 2000s - it was started in 1986 by a veteran who had written Postgres' spiritual predecessor in the 70's (Ingress). So no, Postgres didn't have amateur-hour data loss bugs in the early 2000s. What Postgres was missing out of the box were some features, like replication.
It should go without saying that a white sub-human (according to the bigots of yore) is not a white person.
You are mistaken: the Irish may be white now, but they weren't then! Whiteness evolved over time, it used to be exclusive for WASPs, and excludes the Irish, Italians, Jews, slavs, and other groups widely accepted as white today. Race is a sociological construct.
Microcenter sold it for $5, limited to 1 per customer. Additional ones were $15 each.
The article mentions CXMT has committed to 4F^2 server RDIMMs by end of 2026, and a linked article mentions that CXMT has halved the fab build time from the industry-standard 2 years to just 1. How would Micron catch up with that?
The memory triopoly is in the midst of a slow motion wreck because pumping the brakes will reduce profits; the incumbents are going to have their butts handed to them by CXMT in 2-5 years, and they'll all act like they didn't see it coming as they are bailed-out by the US or SK government, for "National Security."
I suppose the bunkers are just go-bags for the ultra-rich bastards, except I don't delude myself into thinking I can hang onto my go bag against a handful of people who desperately want it, after civilization collapses. The other difference is I don't think my neighbors will go feral on me and fight me for my supplies, but the billionaire have no faith in humanity - I suppose they deduce that everyone will act as they would act.
> Isn’t the Waymo version used for some small portion of the dataset to cover edge cases that they have no real data for?
The usecases in the blog post[1] paint a much more expansive picture of how Wayne used Genie 3: including enabling transfer learning to convert single-camera dashcam footage into simulated full-take sensor data (multi-camera and Lidar), which is extremely valuable.
> My basis is that I’m a real human and these videos look far more like bad video games than the real life I’ve experienced
1:1 fidelity isn't required for a lot of training tasks; for instance, computer vision can do a lot with 32x32 resolution images, which is trash quality to humans. I think the potential customers for this technology will constrain the model to fit their use cases.
1. https://waymo.com/blog/2026/02/the-waymo-world-model-a-new-f...
If someone comments that "Qatar produces a high amount of greenhouse gases", a reply that says "Every country on Earth produces some greenhouse gases" is low quality; can we instead talk about CO2 production, compare trends, dig into why Qatar's per capita emissions are so damn high, and what can be done - if anything - to achieve the idealized version of the world with less climate change.
A proprietary version of this technology is already training Waymo's robot cars. Waymo blogged about their DeepMind-provided world model a year ago.
> It’s so obviously not representative of the real world.
That entirely depends on the use-case and level of granularity needed, doesn't it?
The post-apocalyptic stable-state doesn't require a billionaire at the top; without the threat of state violence, I posit that the most stable configuration is the people who control the guns calling the shots. Things will get feudal fast medieval-era food tasters included; explosive collars do nothing against the chef intentionally lacing food with botulism neurotoxins from canned food.
> [...] with security team to have enough incentives to not challenge leader and enjoy what they have.
This is what we have with the current system, underpinned by the "rule of law" - which has seen better days tbh. These billionaires have been cynically chipping away at its foundations without realizing if society falls, they fall with it. In the dystopian fantasy, the nerds in charge have to pacify all their employees all the time, an ambitious or dissatisfied lackey only needs to get lucky once.
I enjoyed those parts too, for multiple reasons. Typos often forced me to ask a colleague for a fresh pair of eyes (who'd almost always spot the problem immediately. Humbling, and great for team cohesion) manually running git bisect to isolate a problem helped me understand how different changesets interact, making me better at reviewing code. Debugging race conditions taught me resilience, and deepened my understanding of the code base, and the satisfaction from fixing a gnarly problem was fantastic. I also enjoyed gaining mastery and becoming the go-to person for a certain class of problem. These are all gone now, sacrificed on the altar of efficiency gains. Now it's "BTW, I noticed a subtle race condition while refactoring variable names, and fixed it. Review my 6,000 line diff now please."
The unexpected variety of activities gave texture to my working day, and was probably good for the brain. Does anyone actually achieve flow state while using agentic AI? I don't[1], and I hate it.
1. except occasionally during design, which isn't thr same. AI side quests inevitably overflow detailed mental models.
- signed, Management
Consequences from the aforementioned functioning society that would have collapsed in the billionaire's dystopian survival fantasy.
Also, soldiers running amok in rhe absence of a strong government happens in our world - ask any civilian who's been caught up in an insurgency, or look at places like the Democratic Republic of Congo where armed militias run fiefs, and you have to pay tribute to operate a cobalt or lithium mine "their" area.