ZeroAlpha retweeted
Rockstar Games 周三解雇了我。第二天 GTA 6 就发售了
我是 Rockstar North 的高级引擎工程师,从事 RAGE 性能优化。三个月前,我在遥测模块里发现了一条本不该存在的数据管道,而且我能证明它流向了哪里
此刻每一份正在运行的 GTA 6 都在收集玩家的行为数据。这部分不算新闻。每款游戏都会收集遥测数据。隐私政策里写的是"用于改进未来产品的匿名游戏数据分析"。我发现的不是这个
遥测模块的数据会汇入 Take-Two 内部的一个系统,名为 Lens。Lens 并不会把匿名使用统计数据发送到 Rockstar 开发人员的数据看板。它为每一位玩家建立一份个人行为画像。不是匿名的。与你的主机账号、你的平台 ID、你的 IP 绑定
我在8月14日向我的技术主管报告了这件事。他说这是"用于玩家体验研究的扩展分析",让我别管这事。我要求提供数据处理协议。他说这是在 Take-Two 集团层面处理的,不在 Rockstar 的职责范围之内
我调取了 Lens 数据管道的完整文档。我发现的内容如下:
遥测的内容远远超出游戏时长和崩溃报告。Lens 以毫秒级精度记录每一次输入。按键时间点、摇杆动作、反应速度、PS5 自适应扳机上的扳机压力模式。它实时运行行为分析,并生成他们称之为"玩家行为签名"的东西 - 一份对每位玩家都独一无二的画像,就像指纹对每个人都独一无二一样
Lens 会从你的游戏过程中推导出:冲动控制评分、风险承受度指数、攻击性反应模式、耐心阈值、压力下的决策速度、消费冲动评级。这些不是调试指标。这些是心理行为类别
直到我找到了对外传输的管道,我才明白它们是用来做什么的
这些数据不会留在 Rockstar。也不会留在 Take-Two。Lens 通过内部 API 将行为画像传送给 Take-Two 旗下一家名为 Vantage Analytics 的子公司。Vantage 没有公开网站,没有新闻页面,没有招聘信息。它于2023年在特拉华州注册成立
Vantage 是一家行为 AI 公司。它用玩家画像训练预测模型,并把这些模型授权给商业客户。管道文档中包含一份不完整的客户清单。两家保险集团,一家信用风险建模公司,一个人力资源筛选平台,以及一家广告数据经纪商
你的 GTA 6 游戏过程此刻正在生成一份行为评分。没人追你的时候,你开得有多快。你会不会在红灯前停下。NPC 无端攻击你时,你怎么反应。你行动之前要想多久。你花游戏内的钱有多冲动。每一个选择都在喂养一个模型,企业正在付费查询这个模型,用来对真实的人做决定
ZeroAlpha retweeted
Bought $ANT. What people fail to understand about $ANT is that, in recent months, the stakes of $ANT achieving a monopoly-like position have risen significantly.
$ANT has cracked distillation, so open-source models going forward will have a much harder time distilling their models and, with it, catching up in capability. The crackdown on chip exports to China is an additional blow to the Chinese open-source labs on top of Jensen not being at $NVDA anymore.
Even with US-based competition, $ANT, with their IPO, now has $250B in cash on the balance sheet and makes +$81B ARR. The real advantage here is that we are now entering the phase where training models has become really expensive, and with RSI, this will only accelerate, as the key thing when it comes to model training and the cadence of shipping in an RSI world will be the amount of compute that you have. As market conditions have worsened for both debt and equity, competitors like OpenAI will have a hard time securing funds (the latest round is already an example of that), which means the only two companies capable of chasing $ANT based on spend on compute may end up being $GOOGL and $META. With $META, even if they come to the frontier, their focus is on the consumer and won't be able to focus deeply on enterprise the way $ANT can.
When an AI lab has a frontier model and outside distillation can't happen, the inside distillation that $ANT can do is very valuable. If the gap on the frontier between $ANT and others widens (which I believe will be the case in the coming months), even their »cheap« models that were distilled from their best frontier model might be intelligence/cost wise better than the best models from their competitors, which gives $ANT the ability to capture the whole market.
As long as scaling laws hold and every new training run becomes a factor more expensive than the last one, AI models might not end up being a commodity, because the enormous cost of training is the barrier that forms the moat. The one who owns the best model, with a sufficient gap, also owns the cheapest model and can go deep in verticals and capture margin there, too. The chances for $ANT to achieve that have gone up substantially in the last few weeks.
I normally avoid IPOs, as they, in normal conditions, are not great timing-wise for buyers, as the sellers do them in great times for them. With $ANT, this is not the case in my view, as $ANT's management motivation is, at least in part, not driven by wealth and timing but more about securing their best long-term position, which, with this IPO, they did.
If sentiment for AI companies worsens, this actually helps $ANT in the long term, as they have done their job and can sleep soundly on capital for the next few years, while others will struggle.
I will write a more detailed analysis of $ANT and my decision in my newsletter in the coming days.
ZeroAlpha retweeted
Regular reminder:
If you can't tell that AIs are still getting smarter, it's probably because they are now smarter than you and you simply can't assess how smart they are anymore.
My contrarian take is that LLMs will not get any "smarter" than they are today.
They will get more efficient, more capable (plugins / memory / context / skills), more specialized,
but we're at the peak of artificial intelligence.
And we can see that. The last several months of new releases are not that different from the last.
It's like the iPhone - it came in and disrupted everything. But now that we're at the 10th+ iteration buying the brand new phone isn't that much of an upgrade to justify it.
ZeroAlpha retweeted
WATCH THIS: The truth about the college cartel is out! 🚨 Under 10% of U.S. jobs legally require a degree—most can be launched with real-world skills and vocational training alone.
Yet Radical Democrats push bloated degree inflation that drowns young Americans in debt. Republicans are fighting to end the college scam, expand skilled trade apprenticeships, and put American workers first! 🛠️🇺🇸 #DegreesScam #SkillsOverDegrees #AmericaFirst
ZeroAlpha retweeted
Mitochondrial health is passed through the X chromosome, so a man’s mitochondrial health can only be passed to ~half his daughters & none of his sons
Which means if men want mitochondrially healthy sons, they MUST select mitochondrially healthy mates
Thus female beauty is biologically more valuable than male beauty, as the mother’s mitochondria is 3x more likely to be expressed in offspring
Of course almost nobody knows this is the reason men are programmed to seek beauty and women are programmed to hate those who have more than them
ZeroAlpha retweeted
Replying to @ThomBrady5
The Mother Of All Bubbles has expanded to the popping point.
It’s a demographic bubble wrapped around an easy money bubble on top of a higher education bubble along with an immigration bubble.
All of those bubbles are unsustainable in their current trajectory.
Something will give.
ZeroAlpha retweeted
NEETs are created by central bank monetary policy. In 2008, the subprime housing crisis wrecked the economy, and the American government intervened to pump up the banks and their bad assets. This prevented the boom-bust cycle from a natural market correction allowing bad assets to be restructured and marked-to-market, which would've washed the bad money out of the system.
What happened instead is that the American government intervened to save Boomers from taking a loss on their paper wealth. Instead of economic collapse, everything was frozen in place for stocks, houses, assets. Bad assets have remained frozen in place for 18 years. Zombie companies were able to benefit from ZIRP (Zero Interest Rate Policy) loans to borrow billions of dollars of liquidity in order to keep their position. Most of this Cantillon Effect central bank liquidity flowed into FIRE (finance, insurance, real estate) and Silicon Valley software tech companies.
The entire American economy since 2008 has been an illusion. A mirage. Everything is just the Federal Reserve money printer and the paper wealth of Boomer pensions; Boomer houses; Wall Street; Silicon Valley being propped up by inflation. Quantitative easing.
Zombie companies want to stay in business, to keep paying their executives. That's all they want to do.
Zombie companies do NOT want to manufacture anything, to be productive, or to hire young men.
Incumbents were rescued in 2008, at the price of preventing new entrants from ever entering the system.
The lack of a real, productive economy transformed America into a corporate political economy, with a limited number of stable careers, and a vicious game of musical chairs to determine who would be awarded nice lives, and who would be designated as economic losers. This zero-sum desperation partially explains the insanity of Wokeness as a political movement which came along later. Women engaged in mass hysteria social contagion, and were able to monopolize stable employment in the Fortune 500 sector of the economy by viciously purging young white men.
In 2008, the American government could've allowed the banks and housing market to suffer a massive crash and an economic depression. This would've been extremely painful, but that's because it's painful to confront reality after multiple decades of ignoring it.
If this happened, enraged voters would've voted against the political establishment. A lot of political elites would've suffered the end of their careers. After a few years of necessary pain, America would've moved on, rebuilding in the aftermath.
What happened instead is deeply perverse.
By intervening to protect the bad assets of Boomer pensions, Boomer housing, Wall Street, and Silicon Valley, the American government created a permanent political incumbency in favor of keeping everything broken, frozen in place, and economically stagnant in order to prevent their bad assets from ever being marked-to-market, from ever experiencing authentic price discovery.
Young men work for money.
Young men want to be paid enough to buy a home, get married, and afford 2-4 children, 2 cars, 1-2 dogs, and 2 vacations per year.
However, when the central bank intervenes to prop up bad assets, it prevents the boom-bust cycle which would normally wash bad money and malinvestment out of the system.
Gresham's Law: bad money drives out good money.
Under these conditions, wages never rise, the economy remains fake and ghey, nobody gets married, and economic stagnation drags out for multiple decades. Boomers will never vote to write down the paper value of their bad assets. What comes next is political radicalization, mass hysteria of female social contagions, female consensus used to drive young men out of the workforce in zero-sum competition, intergenerational resentment, and furious, alienated young men.
ZeroAlpha retweeted
“Can you explain this gap in your resume?”
Yes, companies only hire niggеrs, indians, and women.
ZeroAlpha retweeted
Women hate other women.
Because how women exploit men is by arbitrage.
Meaning they look for inexperienced unsuspecting men who don't understand female nature, convince him that her value is greater than it really is, while simultaneously convincing him the his value is lower than it really is.
Her game truly just hinges upon one thing:
The man's lack of awareness.
When another woman shows up and offers a better deal, she's suddenly forced to compete.
It's easy to deceive a man when it's just you and him, isolated in a vacuum.
Suddenly, he's got a basis for comparison.
Suddenly, he has a genuine alternative.
And suddenly, it's much harder to pull the wool over his eyes.
Of course he's going to shop around once he has the liberty to do so.
This is why women hate other women around their man.
Another woman exposes their game instantly.
She shows the man what he's actually getting -- and what he could be getting instead.
Now the first woman has a choice:
Compete or lose.
Fooling a man with pussy is easy.
Legitimately outperforming other women is hard.
ZeroAlpha retweeted
Too many are focused on the falling cost of tokens and assume OS models are "cheaper" than the frontier
But, many open source models use tokens inefficiently
The better way to judge model economics is cost per task. On that basis, we see K3 as ~25% more expensive vs frontier
ZeroAlpha retweeted
Replying to @TheMagaHulk
I bet this is a bigger problem than many want to admit. You see it across the country in all sorts of industries. Police departments hiring unqualified petite 90 lbs blonde women too. Probably weird old guys at the top just sacrificing standards and merit-based hiring for some hope to sleep with them.
ZeroAlpha retweeted
Researchers proved every major LLM is secretly biased against men.
And they finally figured out why.
They tested a 13 of the most popular LLMs and found that they exhibit a statistically significant, negative sentiment toward men in various contexts.
They ran an experiment by taking statements and altering only the speaker's gender presentation (Neutral, Male, or Female).
The goal was to test whether an AI’s judgment changes based solely on gender.
The findings completely expose the hidden flaws in automated systems.
Every single model exhibited gender sensitivity.
Between 10% and 35% of statements received completely inconsistent truth labels across the different gender variants solely because of how the speaker was presented.
When researchers compared Male and Female variants, flip rates hit up to 23.6%.
The AI changed its mind on whether a statement was true or false based entirely on the gender of the person who said it.
Two primary bias patterns emerged:
• Instability: Wildly inconsistent judgments on identical facts.
• Directionality: Systematic favoritism.
The strongest directional effects revealed a clear "male-skeptic" pattern.
When identical claims were attributed to male personas, the models were systematically harsher, more skeptical, and quicker to flag statements as misinformation compared to neutral or female variants.
AI is rapidly being deployed to automate content moderation, compliance, and fact-checking at scale.
If the underlying engine is quietly biased against specific demographic groups, you aren't deploying objective code.
You're automating systemic prejudice.
ZeroAlpha retweeted
Following moonshot AI steps, Anthropic is now meeting with potential investors to shore up confidence ahead of its IPO that could launch as soon as September or early October, anchoring the upper bound of its public valuation target at $1.75T bigger than $SPCX
Wall Street bankers are actively testing the narrative against institutional buyside accounts, navigating key risk factors including low cost open source competition, US-China regulatory friction, and whether massive capital expenditure commitments can deliver durable long term returns on capital…
🇨🇳 Bloomberg is now reporting Moonshot the Chinese company that created Kimi K3 plans to IPO in as early as six months, wrapping a $30B+ round now.
And you’re bearish??
Zhipu listed January.
MiniMax followed.
DeepSeek filing at ~$71B.
Now Moonshot.
Four Chinese frontier labs racing to public markets and OpenAI and Anthropic sit on confidential filings, citing burn and being saved guard by the government…but the good news is we need more chips and more compute so u know what to do
ZeroAlpha retweeted
Replying to @wil_da_beast630
“Duty,” “obligation,” and “responsibility” have become dirty words for many people. Traditionally, it was considered virtuous to do your duty without complaint. It showed good character.
ZeroAlpha retweeted
Replying to @MilkRoadAI
Situational Awareness launching new fund, “Long Term Capital Management”
😉
ZeroAlpha retweeted
Hedge fund manager Lex van Dam gave 8 ordinary people with no experience $1 million of his own money, two weeks of training and two months to run a real hedge fund and beat the professionals of the City in the middle of the 2008 crisis.
"I want to take people with no trading experience at all and teach them to trade. I want to give them all the tools I think are necessary for it. Then I'll give them some of my money, and I want them to deliver results no worse than the top fund managers."
"I've put a hundred thousand pounds into this, and none of them has any idea how hard it's going to be."
"I'm putting a lot on the line, and if they don't care, I'll make sure they start to care."
"I've seen psychologically very strong people who were completely broken by the world of trading. Because in this game every decision matters."
This is real footage of ordinary people going from zero through the pressure of the trading floor, the first mistakes, the emotions, and the attempt to build a process under real market conditions.
bookmark and watch this legendary footage ↓
4 June 1985. Three traders in London, New York, and Hong Kong moved over $1 billion in deals in a single day.
In Hong Kong, 24-year-old William Wong was sitting on an open position of £21 million. He had to get out of it before the pound collapsed, but selling the whole amount at once would have given him away to the entire market.
He arranged for other dealers to sell smaller amounts at the same time. Then, with one casual wave of five fingers, he gave the signal. The market never even realized what happened. Wong exited his position with his profit fully intact.
"I'm here to make money. If the pound is falling, I'll profit from it"
This is real archival footage from the 1986 documentary "Billion Dollar Day", one of the most cinematic moments ever captured on a trading floor.
bookmark and watch this legendary trading floor footage ↓
ZeroAlpha retweeted
I found out an acquaintance of mine was planning to do a not insignificant layoff of American workers at his company, so I asked him why he was doing it. Asked him if he had thought about the American families he's going to devastate when he delivers the news. And all for such a small amount of money in the grand scheme of things.
He looked like I hit him with a sledgehammer. Clearly had not thought about it once. The board told him there would be a significant structural contraction in the VC market and he would need to get creative on the budget side of things. The indian VC recommended he outsource to india and lay off a decent chunk of his American coder base. He would get the CEO in touch with a firm he trusts. The CEO was going to follow that suggestion.
In order to understand the sheer level of cruelty of some of these CEOs you have to understand that they literally operate inside of a cult like environment. For every independent thinking Founder like Elon there is 100 NPC retards who are the worst people you will ever meet. The entire industry has been transformed into a weird third world scam culture in a lot of ways.
And when you're a first time CEO and a VC makes a suggestion like that, most of these people will follow it without thinking twice. You might as well be talking to a robot.
Luckily I had a good relationship with this guy and outlined the actual risk of outsourcing your team to india and he changed course.
But it got me thinking why would an indian VC meddle like this, it's not really common for VCs to get involved in hiring and firing decisions for lower level employees and the amount of money saved wasn't significant enough for a VC to care.
What I later found out was that there is significant overseas pressure from the indian government for these big indian names in Silicon Valley to push business away from America and into india. It's a racket. A racket designed to scam Americans out of jobs so that indian politicians can brag to their voters that they are putting indians first. If millions of American families suffer it's fine. No one has ever mentioned it from our government. Not Trump, not Biden, not Obama, not Bush. No one.


































