Every argument clarity score on this site is built from rows on this page. Each
question and answer was assessed with names hidden, the host's own answers included, on
four things from 1 to 5:
directness (does it answer the question asked), coherence (do the ideas follow),
precision (concrete details and clear references), compression (says a lot per word). The weighted
mix (30/30/25/15) is the exchange score. A person's published score averages their exchange
scores on raw tape only, at least 8 of them, shrunk toward the cohort mean.
Full method →
Not addressed raw tape
D 1 · C 2 · P 3 · Cm 2 1.95
Q and win the AI researcher race, we should ban high frequency trading. It feels like we might not even need to have that conversation because Mark Zuckerberg is willing to pay as much as Jane Street now. Um, but what is your take on whether or not, uh, economic value or American values are created through the process of high frequency trading? Should we ban it? Is there an answer?
A So two awesome, like quick and funny stories. The first is Citadel published a paper on back in the V 100 days. There's V V 100, A 100, H 100, and then V 100. So in the V 100 days, Citadel found a way to do mat moles faster than NVIDIA did. And it was like the most incredible, you know, find ever. And it's like, you know, how is that possible? And the, and the paper's fascinating because the techniques they used were just remarkable. The second story is, um, so, you know, they're brilliant people, obviously at these firms. Uh, the second story is I haven't hard launched this yet, but you know, um, I'm having a baby with a woman. She's my new partner.
AI assessment note: “So two awesome, like quick and funny stories.”
Redirected raw tape
D 1 · C 2 · P 3 · Cm 2 1.95
Q did the hard work to get the scoop. They should get 90% of the value and 10% should go to the repurposed rage bait version as opposed to the economic model now, which is maybe you get 90% more clicks on the, on the viral, you know, repurposed content and the scooper actually gets 10%. Do you think that's possible or how do you think that investigative journalism will evolve?
A Yeah, this, I don't know if this is an answer to your question, but, but it, but it's, um, Yeah, I thought it was, I thought it was interesting. I, as I said, I met with Daniel Ak, um, up in, in Stockholm, and, um, and he told me, he said, you know, we were talking about this, and he said, you know, one of the things we do at Spotify is we actually surface all of the different queries that are sent to Spotify that we don't feel like we have a really good response for. Um, so if somebody searches for, you know, I want to, uh, uh, a song to a reggae kind of beat, With, and it's all about how much it sucks when your sister steals your car and your dog. Um, like.
AI assessment note: “I don't know if this is an answer to your question”
Not addressed raw tape
D 1 · C 2 · P 2 · Cm 2 1.70
Q What does it look like when somebody sells someone?
A Well, I mean, there's a technical reason. These are LMs are probabilistic. They're not precise. The, the value of LLM is when it's essentially in an ontology wrapper, because to, to, to actually create value, you have to be able to take the output, serialize it and deserialize it in the context of the business. So the logic actions and security of the business and its tribal knowledge and what it's trying to accomplish. LLMs are vertically crucial, but the, but, but the error bound is very, very, very narrow. And the way you actually do LLMs in the real world, not in theory, not as like, Is that you essentially put them in a concatenated chain where each single thing has to be done as a street unit, because otherwise the underlying math is 95 times a hundred separate change. It's like totally unreliable. And if you do it any other way, you're getting a steak dinner and that steak dinner is super tasty. It's not going to work. And even worse than the steak dinner, honestly, is that you're being taught how to do something incorrectly. It's like, it's like, okay, I'm going to learn how to learn From a wokester.
AI assessment note: “Well, I mean, there's a technical reason. These are LMs are probabilistic.”
Not addressed raw tape
D 1 · C 2 · P 2 · Cm 2 1.70
Q So break it down for me. I saw a hundred K fellowship. I also see a school. Uh, typically I have to pay for school. Are you paying me? How does this all work?
A Right. So this is essentially, there's several things that are happening at the same time. Sure. Like obviously traditional academia is basically over as we know it. It's about 10 different things hitting it at the same time. Um, and, uh, it's, you know, lost trust and credibility and so on. And, uh, you know, what we've got here is we're doing global meritocracy. Anybody from anywhere, obviously Americans are welcome, but anybody who's what I call internet first is welcome, right? So if you're, You know, pro-Bitcoin, if you're pro-cryptocurrency, smart contracts, and so on, you're pro-AI, you're pro-biotech, pro-nuclear, and whatnot. And those kinds of people exist all over the world. And, uh, and so we're bringing them here. And so global meritocracy, basically. And, uh, there's, there's Vitalik, there's Brian Johnson. Can you see the video?
AI assessment note: “obviously traditional academia is basically over as we know it.”
