The BLS has an interesting piece on household computer ownership in the 1990s broken down by education level and race. The overall rate went from 15% in 1990 to 35% in 1997, but households headed by a college graduate went from 23% to 56%. It didn't take long for it to become a majority though - that was around the year 2000.
What's the Indian education system like in terms of academic integrity? I was a TA in college and Indian international students were disproportionately getting in trouble for plagiarism, handing in identical assignments, or trying to cheat on exams. A common excuse was "this is a cultural difference, this was allowed (explicitly or tacitly) in India". With all the weight for admissions being placed on one entry exam, is there genuinely just a more cavalier attitude about cheating on other work, since its perceived as busywork that doesn't matter?
Actually digging into this a bit further, there have apparently been major NEET cheating scandals in 2024 and this year. I suppose with so much of a student'a academic (and therefore economic) future riding on one test, there is a lot of incentive for it. American admissions criteria might be less legible, but they are at least less vulnerable to being gamed by cheating on a single test.
On one occasion, he said, he was held down, stripped naked, and as he was blindfolded and handcuffed, a dog was summoned. With encouragement from a handler in Hebrew, he said, the dog mounted him.
Another possibility consistent with the victim's statement that doesn't require "trained rape dogs" - they easy could have had the dog jump on him and then penetrated him with some other object. The guy tied up, blindfolded, and panicking is going to have no ability to tell the difference. Letting him believe he was raped by a dog would still be incredibly degrading and traumatizing.
The Israeli military uses dogs pretty extensively, this is well documented by non-controversial sources. In the context of a large detention center run by the Israeli military, I'd imagine they are particularly useful to search for contraband like drugs, or to incapacitate combative detainees in a non-lethal manner.
This seems like it could easily be a case of "assholes taking the opportunity to do sadistic shit" especially because they know the Muslim perception of dogs, and therefore it would be even more humiliating and degrading than other forms of sexual assault.
Okay, I'll be more precise - Israel is not a recognized nuclear weapon state ratifier or acceder under the NPT, but still possesses nuclear weapons. That puts them in the same category as North Korea, which followed the agreed-upon withdrawal procedure and is also not bound by the NPT.
Like a lot of infrastructure, it's dual use. Iran is a sovereign nation and can spend money on whatever it believes is in its strategic interest, as long as it abides by its international commitments.
That begs the question though, if Iran has been trying to develop nuclear weapons for decades, why don't they have them already?
From Iran's perspective, they are next door to a hostile nuclear power with illegal nuclear weapons. Iran knows that they won't get the same special dispensation from the international community, so they have to try to ride the line of maintaining some form of deterrent without ending up as an international pariah like North Korea. Historically that has meant having a robust civilian nuclear program, and then using their degree of further enrichment as a bargaining chip. That was the whole point of JCPOA - Iran's breakout time would be regulated through limits on enriched uranium stockpiles and centrifuges in exchange for diplomatic normalization and sanctions relief. These sorts of negotiations have been going on for decades with no prospect of Iran actually producing nukes, despite Israel's constant claims to the contrary. That's my understanding of Joe Kent's statement about IC reporting - he was saying that there was no indication that Iran was planning to break this holding pattern and try to actually produce a nuclear weapon.
From a game theory perspective a way to think about this I suppose is that they are deliberately giving away the information that they are incapable of performing a first strike, but maintaining the potential capability for a delayed second strike. That adds significantly more risk to an Israeli first strike without incurring the full diplomatic consequences of having nuclear weapons. The issue is that the potential for nukes in weeks/months is not the same as having nukes ready to launch - I highly doubt they would have risked a decapitation strike on Iran's top leadership if there was the prospect of immediate nuclear retaliation.
From America's perspective the old status quo was fine - or arguably even beneficial because it discouraged Israel from doing anything too disruptive. The American interest here is essentially just "don't fuck with the oil supply" and by that metric, this conflict is a complete own-goal.
Yeah, there's no way that strike on Iran's top leadership would have been approved if they had nuclear weapons. Blowing up the top layers of the chain of command is too risky when you don't know who's going to end up with the launch codes, and what they are going to do with them. Just the existence of nuclear weapons alone is a powerful deterrent to regime change attempts.
There seems to be no evidence that human to human infection is something the virus is really capable of, and even if this is a strain that mutated to do that, it also doesn't spread through the air.
Scientists are pretty sure the Andes strain is capable of human to human transmission. It's not particularly transmissible though since it requires either transfer of bodily fluids or exposure to respiratory droplets. A cruise ship with lots of people in close, prolonged contact is probably the worst case for transmission which would explain the mini cluster of cases.
Last I had heard, they thought the patient zero was a Dutch birdwatcher who visited a rodent-infested landfill with his wife.
This probably won't blow up into a global pandemic, but it's novel, scary (50% fatality rate), and like you said, people still have COVID trauma. Maybe the coverage is a bit overblown but I don't think it has been too egregious. Eg. Some of the cruise ship passengers exposed hopped on airplanes immediately after disembarking, which is another great enclosed space for human to human transmission... It's possible we might see cases pop up elsewhere if any of those passengers were infectious at the time.
It seems like everyone (except the victim) basically got what they wanted here. The defendant got a very lenient deal, the prosecutors are making a show of protesting but still get to count this as a successful conviction for their stats, and the state of California can use a limited prison spot on a criminal who poses an actual danger to the public. It's not like this guy is going to make a habit of murder by megaphone. It's hard to muster up much outrage when this was a one in a million unlucky outcome from a typical scuffle at a protest.
Canada did that after a high-profile sexual assault case imploded during trial thanks to the defense catching the accusers lying and colluding behind the scenes to coordinate their testimony. The solution, to prevent such an embarrassment going forward, was to hold mandatory pre-trial hearings where the defense is forced to lay all their cards on the table ahead of time if they want records admitted as evidence, making it even easier for accusers to tailor their testimony.
It's more of an acknowledgement of the reality in the US. You can't remove guns from the dialogue tree of those most likely to be a problem - criminals who ignore gun laws. Making it legally harder to carry only impacts the law-abiding segment of the population, who very rarely instigate incidents.
I don't think the falsity is hard to prove here. Opening a bank account for an entity that does not exist must involve making a false statement somewhere, especially given the giant stack of forms banks make you sign to do anything.
I largely agree with your argument regarding the wire fraud charges, it seems like quite an uphill battle.
After doing a bit of googling I found some model jury instructions for the bank fraud charges though, and there does not appear to be any requirement for it to be a loan. The criteria are simply:
First, the defendant [made a false statement or report] [willfully overvalued any land, property or security] to a federally insured [specify institution]; Second, the defendant made the false statement or report to the [specify institution] knowing it was false; and Third, the defendant did so for the purpose of influencing in any way the action of the [specify institution]. It is not necessary, however, to prove that the [specify institution] involved was, in fact, influenced or misled, or that [specify institution] was exposed to a risk of loss. What must be proved is that the defendant intended to influence the [specify institution] by the false statement.
1 and 2 are met, and 3 is met because opening a bank account is an action, and the false statement was intended to mislead the bank into opening an account it otherwise would not have opened.
Maybe this is interpreted differently in Alabama, but those Ninth Circuit jury instructions make it seem like a slam dunk.
That was kind of doomed from the start with a DC jury. In this case they were able to file in Alabama, which should have a more level playing field.
The SPLC is interesting because they don't have a real incentive to shut down the organizations they target. Police eventually want a big press conference with drugs or guns on the table and a bunch of arrests, that's how they demonstrate their efficacy. The SPLC has no authority to do any law enforcement, the best they can do is say "we helped" when law enforcement eventually does act. But they have a direct incentive to have more hate organizations to add to their hate map, more scary news headlines to use to raise money from their donor base, etc.
So if they are able to prove the SPLC funded some of these hate groups for decades and made no serious effort to shut them down, just farming them for content, that does start to look like a deceptive use of donor money. It requires more of a stretch compared to the other counts, but if they do get convicted, their reputation is ruined.
The SPLC has been federally indicted on six counts of wire fraud, four counts of false statements to a federally insured bank, and one count of conspiracy to commit concealment money laundering. And the charges were filed in the Middle District of Alabama. 14-page indictment PDF here.
In brief, the indictment alleges that the SPLC raised money under false pretenses by claiming to fight right-wing extremism, instead funding extremist informants with roughly $3 million dollars of donor money. The informants included members of the KKK and an organizer of the infamous Charlottesville unite the right fiasco. They allegedly did this using illegal means, creating fictitious cutouts and lying to banks to open phony bank accounts to obscure the flow of funds from the SPLC to their informants.
I can't help but feel some schadenfreude here - "no one is above the law" also applies to left-wing NGOs who think they can larp as spies. They even named one of their cutouts Center Investigative Agency... It seems like they flew very close to the sun thinking that their brand and political affiliation would shield them from scrutiny. Project Veritas got a lot more heat for doing a lot less.
From a layman's perspective the indictment seems pretty compelling but I'd be curious to hear what the legal commentators here think. Of course this is only one side of the argument, but those statements to the bank in particular seem quite incriminating. Also, what exactly would be the consequences for the SPLC if the DOJ succeeds on some or all counts?
Anthropic's move here (combined with them handicapping Opus 4.6 a few weeks ago) seems to clearly be an attempt to achieve profitability. The free/subsidized rate train for end users has pulled into the station, and now you have to pay more for the same (or worse) capabilities you were enjoying before.
I guess it depends whether you think this is a forced move due to running out of money or if they have run their internal numbers and think people are willing to pay the increased prices. VC money is a runway, it's not intended to be a permanent subsidy. If they reduce the amount of money they are burning on subsidized inference, that's money they can put into R&D, more GPUs, etc.
It's hard to speculate without knowing more about their internal metrics, but based on the complaints I have heard about Claude being slow, laggy, etc, it sounds like they are quite oversubscribed. If the demand exceeds the supply, increasing prices is the logical move.
There are some Chinese contenders within striking distance as well. GLM-5.1 is open weight and seems to perform somewhere between Opus 4.5 and 4.6. It's pretty incredible that there is open weight competition that's less than six months behind frontier state of the art.
It's a big reason to maintain a two income household as well, even if childcare ends up costing most of the second paycheck. Marital assets are generally going to be 50/50, but it's a lot harder for a woman to argue for alimony if she has been working full-time for most of the marriage.
There are really two different types of paywalls. The lazy version just hides the content using JavaScript but the site still serves it, so it can be displayed with a browser extension. Or the site gives you "n free articles" so an extension can just reset the counter. This has become less common over time.
Services like Substack use a more advanced version, that only serve the paywalled content to authenticated users. To get around that, you generally need:
-
Someone with a paid account willing to scrape the content (or allow their creds to be used to scrape it).
-
A site willing to host the content that won't fold after the first DMCA request.
I'm not aware of a Substack service for this, but archive.today does it for major news publications that are paywalled, and there's a fairly well known one for Patreon called kemono dot cr where people can upload paywalled content, or provide their Patreon auth token to enable it to be scraped automatically.
These tools don't generate 1-shot perfection - you need to create a feedback loop that will iterate until it reaches the goal. That can be either test coverage or using tool calling to hit a live service with a test API key or something. Even just prompting it to use a linter or a compiler to catch syntax errors makes a huge difference. Claude would fix most of the issues you flagged in a few loops of trying to test the library, failing and getting an error message, adding the error to its context, editing the code, and repeating. Then at the end once you have something that works, instruct it to write some regression tests, clean up the code, and make sure everything still works as intended.
You're doing the equivalent of handing an intern a sheet of paper, telling them to write down their program based on a vague problem description, and then calling them an idiot when it doesn't work on the first try.
Today, if you are in a situation where you need to be respectful, you are in a very bad place. You are either in court, in a ghetto, or in prison. In court you must pay deference to the judge (who, being very high class indeed, is often very disrespectful- “I just can’t with that history.” I wanted to jump her too.) In the ghetto you must pay deference to the top dogs, the alphas, the crime mob bosses, whoever is in charge. In prison you must be respectful of everyone else or you’re going to suffer the consequences.
...
When I’ve traveled in the Middle East and Greece, men are extremely respectful of each other. They never talk poorly of their local community and those around them.
Do you not see the connection there? Yes, places where perceived disrespect results in violence tend to have different norms regarding politeness. That isn't to say that disrespect goes unpunished in the US - it can easily lose you social capital. It's just less likely to get you punched in the face.
293 - the pyramids question, makeup, and poetry were my main misses. Also I didn't realize it was supposed to be exactly five until most of the way through...
Wan 2.2 had the lead for a while, but LTX 2.3 came out recently and might have changed that. They are pretty impressive considering the VRAM limitations they have to work within, but definitely a ways off proprietary SOTA.
- Prev
- Next

https://archive.ph/dr885
Not surprised they avoided the entire topic of race-based jury nullification, which seems to have become more frequent post-BLM. Out of curiosity I tried digging up some stats on the prevalence of the phenomenon (eg. conviction rate of black defendants if they have a majority/plurality black jury) but everything seems to reference a single study from Florida, which of course claimed that white jurors are the only biased ones.
More options
Context Copy link