I’m here at the US Capitol this morning to observe what, if anything, may happen here today. Word is this will be a quiet day. I will report anything newsworthy or notable on this thread.
Anemic showing so far. More cops and media than demonstrators. Also a lot of dump trucks and snowplows.
LARP Moses is here. So far that’s about the most noteworthy development. He was talking about drinking beer.
The vibe here is “we love the Capitol police” and “anyone who committed violence should be held accountable.” The big gripe is perceived lack of due process. Remains unclear how much of that complaint is based in reality; much hyperbole in use.
They have a string of “Tu Quoque” logical fallacies and false equivalencies, suggesting that the violence at and inside the Capitol is the same as various other non-Capitol violence and peaceful protests. A festival of whataboutism. Also several claims of warrantless searches.
As this winds down, Capitol Police are staging a shield blockade behind the barrier fence. Energy here is low and I sense no appetite for violence or conflict.
And that’s a wrap. The final call to action, “Go home now please.” Folks are generally dispersing. Thankfully this can be characterized as a non-event with the worst offense being the dissemination of some very exaggerated and distorted claims.
Perhaps the most disturbing aspect of this event was its apparent civility and reasonableness, with leaps of logic just below the surface. Whoever collected the names/data for this event has a pool of people to feed into future radicalization pipelines. The funnel must be fed.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
1/Increasingly bothered by the argument that the path to “intelligence” requires the theft of all known intellectual property. That’s not anything but theft, and if you were actually building “intelligence” it theoretically requires near zero training data.
2/If, as proponents argue, we are simulating human intelligence “in silico” then we should be able to build what amounts to a 12 year old’s mind. A 12 year old can be extremely “intelligent” (genius level in fact) but has necessarily trained on very little data.
3/Few are talking about solving that problem, because we really haven’t figured out the neuro-symbolic architecture to do it. Instead, we are vacuuming up copyrighted works (stored intellectual work of others) and billing that as “PhD level intelligence.” What utter bullshit.
The funniest part about Zuck throwing billions into AI is his naming of the effort as “superintelligence,” as if that alone will manifest the gods. The second funniest part is he desperately tried to manifest the metaverse just 3 years ago — and failed at that, too.
What this tells us is that the core FB properties are aging cash cows and Zuck is desperate to invest billions in anything that will ensure his personal relevance as a tech titan 10-20 years from now. Good luck, but throwing money at things is no guarantee of success.
As @GaryMarcus and I have been saying the last few years, AI is up against some very challenging limits, and despite the massive investments in the space, progress has plateaued. Plenty to be done with what’s been built, but unclear the investments will be recouped anytime soon.
1/Yesterday, OpenAI demonstrated that even after spending untold billions on scaling AI models, LLM’s alone aren’t the ticket to achieving superintelligence. This is linear growth, not an exponential leap, and that has several implications…
2/First, it vindicates people like @GaryMarcus who have correctly argued that language-only architectures lack capacity for reason and actual world modeling, and that neuro-symbolic architectures are needed. I agree, and while we can say “I told you so,” that’s not important…
3/relative to other implications. Much bigger is the coming reckoning in the marketplace. Models are commodities. Application developers have opportunities to make money, but there is likely to be a big contraction in the AI sector generally. That’s fine and a natural…
As the Epstein blackmail/espionage story continues to fall apart, bookmark this for future reference. At some point people will figure this out. gameb.wiki/index.php?titl…
This aligns with the data-driven analysis we performed. What Acosta may have been indicating is that Epstein was involved as a confidential informant on a case; but there is little else to indicate ongoing large-scale involvement or a sprawling blackmail ring.
That doesn’t mean there isn’t stuff we don’t know and that some of it is material. But it may mean that the government doesn’t have details on it, and that it’s incumbent on journalists to find out more. Our analysis here. america2.news/we-mapped-jeff…
That said, journalists have doggedly picked over reams of data and leads on Epstein for decades. The idea we can find more when incentives were already so high seems improbable. But let the govt show what they have and let journalists do their work; we’ll see where that leads.
1/🚨 BREAKING: A State Department unit closed effective July 11 is now amplifying corporate messaging on behalf of X. The Bureau of Democracy, Human Rights and Labor posted this message yesterday. But the agency's website no longer exists.
2/The New York Times reported on July 11 that the bureau was closed. So why is it tweeting X corporate propaganda out to its 59.3K followers, which was also amplified by Elon Musk personally? nytimes.com/2025/07/11/us/…
3/The (former) website for the Bureau of Democracy, Human Rights and Labor no longer exists. Do defunct State Department accounts simply become Musk promo accounts by default now?