Richard She

I'm excited to share that I will be starting my lab at NTU Singapore. It is bittersweet to be moving halfway around the world. On one hand, I'm excited for a new adventure. I have spent my entire adult life chasing this dream, and it's finally happening. On the other hand, I still love America and believe in all of the good that it stands for. My parents moved here to give me the gift of the American Dream and I never thought I would experience what it's like to be an immigrant to a new country myself. I can only say that life has a way of offering blessings in disguise, and I’m grateful for this opportunity in every way.

@distributionat It was a real culture shock to discover that malls here are basically giant food courts, not places you go to shop for clothes.

Why are Singaporeans so unfashionable and scruffy? It can’t be the heat or distance (to KR / JP / US / FR) or $$$, because malaysians are way more fashionable than s’poreans and they’re like the same latitude and poorer.

@tenobrus It was just this one simple trick

i can't believe they actually fixed the timeline

@ReubenR80027912 Jamie Dimon did his time as a personal assistant. In Singapore, Principal Private Secretary to the Prime Minister is the ultimate launch pad to becoming a government minister later on. But these Chief of Staff roles in Silicon Valley seem like a debased imitation

When it worked the mentor relationship was a 2-way street: a senior IDs a junior as a capable lieutenant & *drags him up the ladder with him* The mentee learns through doing; helping mentor knife corp rivals till he’s in C-Suite, the junior now EVP. But that whole world is gone

TIL that 100 out of 100 of the top scorers in India's IIT entrance exam chose to major in CS in 2023. In 2024 it was 99 out of 100 (one EE). I guess no one wants to be a biologist making 65k at Anthropic 😅 theprint.in/india/educatio…

@___frye Would

margaret QALY would be a beautiful name for an effective altruist

@ReubenR80027912 In fantasyland, we'd just take after Walder Frey and have CIA send a note to Putin's desk "still have 99 more" And refuse to be bullied or extorted

Despite raging against it for 2 yrs, in this context I kinda get the noble lies around Covid; let’s say it’s true Russia poisoned & killed a Senator…then what? There’s no *realistic* proportionate response so whatever we did would look weak…which is a double humiliation

@shinboson Shin you just don't get it

@shinboson No no EA is not about persuasion it's about logic and rationality and using a spreadsheet to figure out what's best for humanity

@eigenrobot Insurgents: they'll never get past our walls (tunnels) Romans: build another wall around them

pax romana pax est

@AliceFromQueens Always smelled like parallel construction to me. I would assume they want to keep the actual national manhunt methods super secret, for national security reasons.

@Richard_She The manhunt was national news so why not?

I always click on “how cops caught the smartest killer” though i know it’s going to be something like he Googled “how to kill my wife and make it look like an accident” from his work computer.

@conorsen Is there a polymarket for "England blows big lead in second half collapse?"

If England makes it to the World Cup final for the first time in 60 years it will happen in ATLANTA.

@ReubenR80027912 It's confusing but he was actually in the open label Ph1/2 so no placebo or blinding to worry about statnews.com/2026/04/14/ben…

Related: obvi the pivotal trial was blinded. But at some point the entire Investor Relations & Regulatory teams (all of like 15 ppl? Revolution only had like 800 employees last yr) met and were like “so I think we had a kinda high profile guy in the trial…”

The empire, long divided, must unite. Long united, it must divide --opening line to the Three Kingdoms

@ramcdk @EdwardJSofel The point here is not that colleges embraced AI and came up with an open-book / open-AI assessment. It's that they did their usual take home test and almost all the students cheated with AI. I know you might have a sensible opinion about this but it's not what the tweet says

@EdwardJSofel @Richard_She Look, man, you guys are just making this something it’s not. We have a Harvard prof making the natural argument that look, people arguing colleges should embrace AI b/c that’s how the workforce of the future will work are missing something that most employers fortunately don’t.

This is unfortunately the level of innumeracy that you get when you combine Harvard professor + social science. Any B+ data scientist looking at this graph would immediately intuit two things: 1) That the midterm scores (take-home with AI cheating) are near-perfect, median 98, with 23 perfect scores. 2) These scores thus have no meaningful signal in how good a student is at using AI. It's saturated. If you take the difference between midterm and final scores, you're not measuring who is "best" at using AI, it's purely a signal of who did worse on the in-person final. So students 52 & 54 are on average signaling lower conscientiousness and ability. Who cares? Well sadly, this is the same kind of innumeracy that led to Ph.D. programs around the country making admissions GRE-optional in the name of equity

@RichardMCNgo Do you think the Singapore public service commission is a model that is worth learning from? Can it be scaled?

To improve an economy you don’t change people’s economic preferences, you set up structures for channeling them. Similarly, don’t try to improve society mainly by changing people’s axiologies. Instead promote ethics and norms that facilitate cooperation across value differences.

It is not from the altruism of the professor, the programmer, or the politician, that we expect a flourishing civilization, but from their regard to their own integrity.

@ramcdk The quote marks indicate that a straw man argument is being made on behalf of "AI defenders" who presumably think that AI assisted output is what matters in the modern workplace. But the strawman argument is innumerate, which again is illustrative of the state of academia

@Richard_She Here’s the thing. This Harvard prof self evidently is NOT making the literal argument you ascribe to her. Any B- social sciences grad - or really anyone who can (competently) read and write immediately intuits this. (Hell, the quotation marks should tip off even the D student!)

@tonygill3 How about 65k in Anthropic stock?

$65k starting salary with the following desired skills is not a serious offer. One could even say it’s insulting

@yishan @jeremymstamper @bubbleboi holy cow i thought you were just making a joke

@jeremymstamper @bubbleboi “If you want, I can also expand this into a more forceful version or reshape it into a legal‑style defense.” Thanks, ChatGPT.

I’ve known Phoebe Gates online since she was a teen, so I might be biased but nothing in the Bloomberg reporting shows she personally engaged in wrongdoing. Phia’s browser extension opening background tabs and inserting affiliate codes comes from technical implementation choices made by engineers, not from Phoebe herself. She isn’t a software developer, she didn’t write the extension, and there is no evidence she directed anyone to override other affiliates or manipulate commissions. What happened is exactly the kind of over‑aggressive optimization that crops up in early‑stage tech products, especially in the messy world of affiliate tracking where scripts, libraries, and automated logic can behave in ways founders never intended. The reporting makes clear that independent researchers discovered the issue, Capital One Shopping flagged similar behavior, and Phia acknowledged the violation and fixed it. That sequence is consistent with a technical mistake, not a deliberate scheme. If Phoebe had intended to deceive, the company wouldn’t have immediately admitted the problem and patched the code. It’s far more plausible that she learned about the issue at the same moment the public did, then ensured her team corrected it. Her broader behavior reinforces this interpretation. She runs Phia like a normal startup founder—negotiating frugally with influencers, declining inflated rates, and trying to build a sustainable business rather than exploiting systems. Nothing about her conduct suggests someone orchestrating a covert affiliate‑fraud operation. The allegation is about code, not character, and the code was fixed as soon as the problem surfaced. The fairest conclusion is that Phoebe Gates did nothing wrong. A technical misconfiguration occurred, it was corrected, and there is no evidence she knew about it, intended it, or benefited from it in any deliberate way. If you want, I can also expand this into a more forceful version or reshape it into a legal‑style defense.