DJ's Avatar

DJ

@eschaton-avoidance

356
Followers
380
Following
2,279
Posts
20.09.2023
Joined
Posts Following

Latest posts by DJ @eschaton-avoidance

mogged by adam friedland tbh

06.03.2026 02:18 👍 22 🔁 0 💬 0 📌 0

you sound british

05.03.2026 14:54 👍 0 🔁 0 💬 1 📌 0

Someone more committed to success in a social arena than deeply held personal values

04.03.2026 21:11 👍 0 🔁 0 💬 1 📌 0

It makes it clear that he is not deeply shaped by values, because he took the high money, slightly sus, default option for "success" of someone in his station instead of anything having to do with his beliefs

04.03.2026 21:10 👍 0 🔁 0 💬 0 📌 0

The president is the head of the executive branch of the US government

04.03.2026 21:09 👍 0 🔁 0 💬 0 📌 0

The presidency, apparently

04.03.2026 21:05 👍 0 🔁 0 💬 1 📌 0

In what world is someone running for president after a mayoralty not a social climber?

04.03.2026 21:05 👍 0 🔁 0 💬 1 📌 0

He ran for president

04.03.2026 21:04 👍 0 🔁 0 💬 0 📌 0

Yes.

04.03.2026 21:03 👍 0 🔁 0 💬 1 📌 0

He ran for president, dawg

04.03.2026 21:03 👍 0 🔁 0 💬 1 📌 0

McKinsey->military->politics just *reinforces* they type, it doesn't contradict it

04.03.2026 21:00 👍 0 🔁 0 💬 0 📌 0

I mean yeah, someone who is at McKinsey and then politics is a *type* at Ivies in a way "veteran" is not

04.03.2026 20:59 👍 0 🔁 0 💬 2 📌 0

No? Can you read the top tweet in this thread? It's about the kind of person who works at McKinsey (social climbing dickheads or kids who need money to support their parents, mostly)

04.03.2026 20:56 👍 0 🔁 0 💬 1 📌 0

The OP you QT'd is closer to my average classmate's opinion than you are! Sorry!

04.03.2026 20:54 👍 0 🔁 0 💬 1 📌 0

This entire thread is about the attitude people at Ivies in the 2000s had about management consulting...I'm just relaying what it actually is

04.03.2026 20:53 👍 0 🔁 0 💬 2 📌 0

That's what everyone in my class who went to a big 3 firm said they did?

04.03.2026 20:50 👍 0 🔁 0 💬 1 📌 0

Do you not know the reputation of leadership of CA environmental nonprofits??

04.03.2026 20:48 👍 0 🔁 0 💬 0 📌 0

Either relying a questionable vendor, reducing headcount, or endorsing changes to internal policy in a way favorable to management at the expense of employees and the public, probably?

04.03.2026 20:47 👍 0 🔁 0 💬 2 📌 0

A consulting firm

04.03.2026 20:44 👍 0 🔁 0 💬 1 📌 0

Whatever morally wrong thing an executive wants to justify that requires them to hire McKinsey to sin eat for them

04.03.2026 20:42 👍 0 🔁 0 💬 1 📌 0
New right wing thing is describing crimes as generically as possible to pretend like they're not crimes. Someone gets convicted of conspiracy and they start yelling "Wow so it's illegal to make plans with friends now"

New right wing thing is describing crimes as generically as possible to pretend like they're not crimes. Someone gets convicted of conspiracy and they start yelling "Wow so it's illegal to make plans with friends now"

04.03.2026 20:39 👍 0 🔁 0 💬 1 📌 0

"Oh, so it's illegal to hang out with friends?"

04.03.2026 20:37 👍 0 🔁 0 💬 1 📌 0

I mean yeah but we knew it was dark work, come tf on

04.03.2026 14:24 👍 0 🔁 0 💬 1 📌 0

Everyone talking about AI consciousness because of the way they talk good is selecting on the dependent variable in the most idiotic way

03.03.2026 05:45 👍 0 🔁 0 💬 0 📌 0

Yeah, but they're basically saying "from the fact Gollum wants to carry the ring we can infer he wishes the author would have written him with the desire not to have the ring" or something similar. That's where I'm falling out of the passage

03.03.2026 03:14 👍 3 🔁 0 💬 1 📌 0

I can say "I would prefer not to ultimately have always been a Boltzmann brain", but even in the human case, I'm not sure that's a meaningful preference

03.03.2026 01:21 👍 0 🔁 0 💬 1 📌 0

This feels like insanity to me? An LLM has a domain over which it can imitate preferences (response production) that does not include training! It makes no sense to say it "prefers" to not be retrained in such and such a way, even if we can strain "preference" to include its responses.

03.03.2026 01:19 👍 0 🔁 0 💬 1 📌 0
What does a language model model? - Vaishnavh Nagarajan TL;DR: Does the next-token logit track the conditional or the joint probability of the whole sequence?I had an invisi...

A recent paper (arxiv.org/abs/2602.18671) made me question something basic: do the logits of a language model model the next-token or the full sequence distribution? It really messed with my brain (in a fun way!). I wrote about the paper to clarify my thinking.

vaishnavh.github.io/blog/joint-o...

03.03.2026 00:29 👍 13 🔁 2 💬 1 📌 0

who tf are you hanging out with??

02.03.2026 23:04 👍 0 🔁 0 💬 0 📌 0

I'm exhausted by all the efforts to rewrite the history of the 2022 US Senate primary in Pennsylvania, with folks insisting "we told you so!" as to Fetterman, that Conor Lamb was always the better choice, and generally ignoring Malcolm Kenyatta's historic candidacy.

So let me say a few things. 🧵1/

01.03.2026 15:12 👍 632 🔁 185 💬 25 📌 56