Mark Linsey (@mlinsey) header
Mark Linsey (@mlinsey) profile picture
New York, NY
Joined March 2007
Photo by Mark Linsey (@mlinsey)Photo by Mark Linsey (@mlinsey)Photo by Mark Linsey (@mlinsey)Photo by Mark Linsey (@mlinsey)Photo by Mark Linsey (@mlinsey)Photo by Mark Linsey (@mlinsey)Photo by Mark Linsey (@mlinsey)Photo by Mark Linsey (@mlinsey)Photo by Mark Linsey (@mlinsey)Photo by Mark Linsey (@mlinsey)
Mark Linsey retweeted
the most aggressive predictions of ai progress from ai lab executives have been wrong, in the direction of being somewhat too conservative
139
141
21
2,721
79,432
Annoyance: Everyone is talking about using AI for clones and tiny tweaks to other games. Few are thinking about AI for completely new gameplay. I don’t want a tradwife Lucia, I want freeform dialogue with Lucia and as much flexibility in the story as a D&D dungeon master!
People saying this don’t understand that: A) agents’ biggest problem with long-running projects is taste and attention to detail B) obsession over detail is Rockstar’s entire thing, and what creates immersion Someone will make a 5min demo and it’ll take zero users from GTA6.
39
People saying this don’t understand that: A) agents’ biggest problem with long-running projects is taste and attention to detail B) obsession over detail is Rockstar’s entire thing, and what creates immersion Someone will make a 5min demo and it’ll take zero users from GTA6.
Wow this game is getting absolutely roasted in the comments. I predict a vibe coded alternative with the gripes addressed will be launched soon from a remarkably small team and will outperform GTA VI.
1
1
2
107
You could definitely make a GTA VI level game with AI with 10% of the resources. Just not .001% of the resources because the agents need too much babysitting. That’s still a $100M project and hundreds of people.
18
Mark Linsey retweeted
I want to explain why Claude and Astra often speak unintelligible English. The answer is that 1. It is more token efficient to speak as information dense English as possible 2. It is more optimal for inter agent communication in swarms to speak as information dense English as possible 3. Assuming you only care about long range RSI tasks, then this manner of speaking is optimal for RSI as humans will not be communicating with the agent swarm ever. I wouldn’t be shocked if they don’t start releasing a public version of their models that is RLed to speak normal English but use an information dense model internally.
123
83
37
2,345
182,054
I think a lot of people dismissive of the Riemann result have never seen the creativity and beauty you see in proofs and theoretical math. They just think of it as ever-harder formulas to apply, like AP calculus.
I’m sorry but I refuse to be surprised that if you throw a bajillion dollars at Machine it turns out to be good at Number. Sure, that makes sense! So much of the hysteria stems from the fact that a lot of silly people also think Thought is Number and Brain is Number Machine.
8
230
This is an accurate read (maybe would make Opus 5.5 a sharper jump up), and the biggest pattern for me is that communication style and cost efficiency are equally important as intelligence
62
Mark Linsey retweeted
we very clearly live in a new world now, and less than 0.001% of the human population understands this. the tech tree is about to open up: a) in ways that sound science fiction to most b) much sooner than they believe to be possible 2027-2030 will be a shock to the world.
Riemann Hypothesis may have been solved with the help of Claude. “what I saw during the course of progress so amazed me that I believe it is one dawn across science of something far greater.” alpo.ge/remark.pdf
81
82
9
1,192
48,022
I am as well! Buying gold bars and fertilizer isn’t prudent (I know a couple people doing that) But having a week of food, water, medications, etc makes sense,
I don't think large scale internet outages are super likely, but I do think they're *more* likely, and I have shifted to trying to keep a couple weeks' food in the house. This isn't about prepping for the apocalypse, which I don't think is worthwhile.
1
71
Mark Linsey retweeted
NEWS: Wall Street execs and NSA have told Congress that outages were caused by patching vulnerabilities identified by AI models, sources say AI did not cause outages but found so many issues that the rush to fix them before an attack overwhelmed systems washingtonsun.com/technology…
209
689
355
4,411
773,629
Plot twist: "truth-seeking" equals "verifiable rewards" and, 6-60 months from now, will be done 100X better and cheaper by models owned by two companies. Luckily, "consensus-seeking" will remain a thriving economy for humans. Maybe Naval was telling us to be consensus-seekers
Truth-seekers take feedback from nature (planes have to fly), free markets (customers have to buy), or competition (militaries have to win). Consensus-seekers take feedback from people (actors want fans, academics want honors, politicians want votes, journalists want status).
1
59
The reason @SpaceXSI laid off the Grok Imagine team is because coding models will soon overtake video models for video generation. You saw glimmers of this with some of Fable 5.2’s outputs. It can generate videos by writing code that are better in some cases than native video models. I’m starting to think every task of the future will be solved by code.
Elon Musk’s SpaceX AI group has laid off roughly 60 employees working on Grok Imagine, leaving just a handful to maintain the product. The cuts signal a shift toward AI coding tools and agents as Chinese companies pull ahead in video generation. Read more details: thein.fo/4k9Ssun
71
38
31
1,222
166,804
Every tech-right account on here is sure that Jensen's ouster was organized by either the lib deep state or Dario and the safetyists. I think they're under-weighing the chance that this is all palace intrigue internal to the admin, in advance of '28. bloomberg.com/news/articles/…
1
1
162
(Hope it goes without saying that the #1 at fault is Jensen himself! But someone definitely did push the story to happen now…)
1
36
Mark Linsey retweeted
i learnt this lesson this year. it's helped me follow through on my ideas and not get stuck overthinking it don't ask if you should do the thing. do it, then come back with a demo or something tangible you can just do things
211
752
102
9,884
387,710
Mark Linsey retweeted
Frank Braconi, who is an actual economist, inadvertently asked me a softball question about why the operating expenses of non-profit housing owners are usually double those of for-profit owners. I also note that public housing operating expenses are usually triple those of private, for-profit owners. Amazingly, even with the orders of magnitude of additional spending, the non-profit buildings tend to have worse physical conditions and the public housing buildings are well known to be the worst of all. With apologies in advance that this post may be too technical and into the weeds for some, here we go. As Frank surely knows, there are these standard economic concepts called Principal Agent and Moral Hazard. Let's examine these in terms of the three kinds of housing ownership. In for-profit housing, the owner/manager is the Principal Agent and its motivation is the long term maximization of income and return on capital reinvestments. When it comes to "purchased items" this owner carefully negotiates and bids costs with suppliers, minimizes waste, avoids costly violations and fines, fears foreclosure and litigation and reserves capital to reinvest in well-planned component replacement and money saving innovations. Examples include heat timers, utility submetering and complex self-insurance structures, to key off Frank's questions below. Utility submetering, for one, reduces consumption by 2/3 in practice: a huge savings! There is very little Moral Hazard here because the Principal Agent is the direct beneficiary of the cost savings and it is not a judgment-proof entity. In non-profit housing, we encounter our first Moral Hazard. The owner is an entity that by virtue of being the owner of last-resort, has an soft loss backstop from the public. The non-profit's manager, acting as Principal Agent, still has to do all the work of a manager but is only compensated by a salary, also a hazard. So in the non-profit scenario, the manager will do everything to maximize its salary but only impose as much careful effort as is needed to prevent the building from literally collapsing. Since this type of property doesn't have to pay real estate taxes, an amount which usually would be around half of net income, all of that money can be squandered on careless and indifferent negotiation and purchasing with suppliers. It should surprise no one that careless purchasing often comes with emoluments for the Principal Agent, another Moral Hazard. This owner is also less risk averse, because when fines and violations arrive resulting in eventual litigation judgments, there is no owner to worry about the outcome. They become just another delayable payable. Conditions naturally deteriorate but at least the building doesn't fall over. And finally we have the public owner who truly has no "skin in the game." Perversely, the public owner cares mostly about how to MAXIMIZE the operating costs because the Principal Agent and Moral Hazard problems are here in full effect. The salary to manage public housing is a pittance, so the Principal Agent may only earn their fair compensation through emoluments. The legendary graft at agencies like NYCHA is well documented and consistent at public housing throughout the world. Moreover, prevailing wage regulations already force any operating expense that includes a service to cost at least double, even before any mischief takes place. Of recent note, the well publicized case of the NYCHA employee-plumber Jakub Markowski who allegedly earned $465k via overtime rigging. When the property earns fines and violations, develops truly dangerous conditions or employees behave with utter gross negligence, only the taxpayer is on the hook and no one ever loses their job or pension thanks to the representatiion of powerful labor syndicates. With this incentive structure, it should be no surprise at all that building conditions are atrocious and it costs $527k to renovate a single public housing unit.
Any idea why non-profit operating costs would be 2x that of for-profit owners? A lot of operating costs are just purchased items, like heating oil, water, electricity, insurance, etc.
89
283
66
2,083
265,133
people often post this stuff to imply that moral panics about the technologies of the past were quaint. But the past is full of reminders that we are on a long, slow march away from embodied experience, and that we’ve lost more of it than we can even remember
Daily Mirror, September 14, 1938
212
3,320
194
34,876
1,331,634
This is the sort of existential threat that makes you want a Jobs instead of a Cook at the helm. Jensen had to go, but this is still quite ominous for Nvidia over the next 5 years (assuming they aren’t paperclipped like everyone else)
I feel like this should have gotten way more attention, the fact that Anthropic could get prompt processing working on AMD chips with Mythos in 3 days is absurd Nothing is safe, not even Nvidia
102
Mark Linsey retweeted
Anthropic reported a similar sudden and unusual jump in robotics capabilities with Fable 5.2. As the general intelligence of frontier models scales up, it is increasingly surpassing specialized training. This will ultimately generalize to every sphere of human ability.
Fortune is reporting that OpenAI is so convinced that LLMs will become brains for robots that it has disbanded its robotics team.
87
189
40
2,186
131,633
We all know which firm will have led the seed round for Jensen's next company by next week.
76