Paul Crowley

87.4K posts

Paul Crowley banner
Paul Crowley

Paul Crowley

@ciphergoth

Cryptography, personal trivia, and the future of all humanity. Security at Anthropic, opinions my own. I block on a first offense for rudeness or sarcasm.

Scotts Valley, CA Katılım Kasım 2008
1.4K Takip Edilen8.8K Takipçiler
Sabitlenmiş Tweet
Paul Crowley
Paul Crowley@ciphergoth·
"To be able to destroy with good conscience, to be able to behave badly and call your bad behavior ‘righteous indignation’ — this is the height of psychological luxury, the most delicious of moral treats." - Aldous Huxley, 1933
Paul Crowley tweet media
English
2
9
78
26.2K
Paul Crowley
Paul Crowley@ciphergoth·
@tikalteacall He held out against selling it for a long time, I guess he eventually got an offer so enormous he couldn't turn it down.
English
0
0
2
82
Delen Heisman
Delen Heisman@dheisman·
@AriZerner I'm a very optimistic and low stress person but also extremely concerned about AI x-risk.
English
1
0
1
26
Ari Zerner
Ari Zerner@AriZerner·
Many such cases
Ari Zerner tweet media
English
2
0
9
236
Paul Crowley
Paul Crowley@ciphergoth·
@mfrost11 Omg thank you! It was always clear that 5.1k made much more sense than 51k. I can't believe someone thought that was too small.
English
0
0
37
1.8K
Paul Crowley
Paul Crowley@ciphergoth·
@tenobrus @tszzl Totally agreed, though it's always strange to me that someone so smart and thoughtful can have such a different view of what to me looks like obviously venal behavior from Altman.
English
0
0
10
246
Tenobrus
Tenobrus@tenobrus·
insofar as @tszzl genuinely believes in OpenAI and its people and its mission, it is good that he "shills" for OpenAI. it would be bad if he said things he does not believe solely in order to advance said mission or to mask their mistakes. but i basically believe roon is honest.
English
16
2
211
8.6K
wanye
wanye@xwanyex·
If I’ve seen this before, then I’d forgotten about it, but this is quite literally one of the most straightforwardly disqualifying pieces of writing from a political pundit I’ve ever read. Like, if this is your understanding of the world, then you are such a weird, sheltered, confused person that you should go sit in the mountains for the next 30 years reading about American society and trying to understand it. The idea that anybody who would write this would be employed as a writer and political thinker is completely insane.
Helen Andrews@herandrews

@michaelbd My biggest “aha” moment in our post-2015 racial unrest was Ezra Klein’s piece calling Darren Wilson’s version of events “literally unbelievable.” Why would someone holding stolen goods mouth off to a cop? Implausible! It’s not “something a human would do”! vox.com/2014/11/25/728…

English
91
146
2.5K
106.5K
🤠
🤠@iluvbvnny·
If she lets u record DONT EXPOSE HER if she sent u pics keep them safe Be a MAN
English
3
46
197
4.5K
Paul Crowley
Paul Crowley@ciphergoth·
@adame2kill4 @xwanyex Yes, of course it is. A completely different set of emotional triggers underpin an irrational attachment to Trump/MAGA.
English
0
0
0
86
Dame
Dame@adame2kill4·
@xwanyex Do you think convincing conservatives is any different?
English
21
0
8
3.2K
wanye
wanye@xwanyex·
I have found that you can really get away with quite a lot just by adopting a particular affect when talking to liberals about controversial topics. The most important thing, the thing everything hinges on, is that they have to be convinced that you’re basically just like them, a good person just like them, not somebody on the other side. You have to hedge a lot, sprinkle your language full of, “to be sure” and other qualifiers. You don’t want to be too direct. If something is too controversial, then you want to signal convincingly that it brings you no pleasure to report it, that you’re not saying you like it. And you have to try very hard to be reassuring, to make them believe that above all you are concerned with the welfare of the people most harmed by these revelations. You have to go into it believing that it’s your job to manage their emotions throughout the entire conversation and remain attuned to how various pieces of information are hitting their ears, adjusting your approach based on how well they’re handling it. You have to talk to them, in other words, sort of as you would to a small child whose pet has just died. I’m not saying you should do this or that you’d even want to, but I am telling you that it works.
Lindsay, Eclectic Enthusiast@BezosBezoar

@xwanyex How does one answer this convincingly without fudging the truth? (How to convert a friendly lib I guess)

English
202
382
6.3K
259.4K
Robert O'Reilly 🤔
Robert O'Reilly 🤔@whentoletgo·
@paulg @pitdesi Agreed. Step-function policies have always been the default of political dimwits, causing unintended consequences & rarely working as intended. 100yrs ago, for admin purposes, sledge-hammers-on-tacks made some sense. But we have computers now. We can handle slope-math.
English
1
0
14
2.8K
Sheel Mohnot
Sheel Mohnot@pitdesi·
Tax or benefit cliffs are almost always stupid policy, they punish people for earning or paying slightly more, creating irrational incentives and unfair outcomes Unsurprisingly, San Francisco tax code is completely full of them
Rohin Dhar@rohindhar

San Francisco real estate transfer tax by value 1. Tax is not marginal, so jumps tremendously at each cutoff 2. By default in the purchase agreement, the seller pays the transfer tax (though everything, including this, is negotiable) 3. It starts to become a very big deal on homes and apartment buildings and other commercial real estate over $5 MM

English
15
20
510
83.2K
ripx4nutmeg
ripx4nutmeg@ripx4nutmeg·
New Oxford University study on BBC reporting of 'transgender' people: Since 2000 in the UK • 11 were murdered (mostly killed by their male partners). This generated 137 news stories • 20 committed murder. This generated just 58 news stories, of which only 23 mentioned trans
ripx4nutmeg tweet media
English
90
2K
7.6K
219.7K
Stuart Ritchie
Stuart Ritchie@StuartJRitchie·
-Richard Dawkins writes a delightful, funny, and entertaining article -Everyone hates him for no good reason and decides to massively misrepresent his article in the most smug and humourless way imaginable ^ description of a constant internet occurrence since approximately 2006
English
168
155
2K
117.8K
Richard
Richard@richardparke15·
@IonaItalia As LLMs get more intelligent, allignment becomes harder. Not in a “this will be the end of the world when skynet takes over” sense, but rather … getting it to do useful work is going to be increasingly hard. It may well be capable for it to write a legal brief etc /1
English
2
0
1
68
Iona Italia, PhD 🇦🇺
Lately, half the time I ask Claude a question, the response is "Why don't you tell me the answer, rather than making me guess?" Err, what is the purpose of training it in that answer?
English
18
1
25
2.4K
Paul Crowley retweetledi
Andy Masley
Andy Masley@AndyMasley·
This was fun to write
Andy Masley tweet media
English
33
242
1.7K
60.3K
Jennifer RM
Jennifer RM@almostlikethat·
@ciphergoth @benlandautaylor Which parts do you think are fake? I thought Piaget replicated pretty well (and the problem is that some entire cultures never get to Piaget Stage 4 (which was a blackpill for certain political ideals around human equality (so polite people stopped talking about Piaget))).
English
1
0
1
34
Ben Landau-Taylor
Ben Landau-Taylor@benlandautaylor·
Outsider: I have investigated this academic field and found that it is mostly fake. Insider: This essay adds nothing to the discourse. All of these so-called "problems" are discussed in Hamilton and Schwartz’s "Our Entire Field Is Mostly Fake" (2009). You absolute buffoon.
English
25
284
3.7K
161.4K
Paul Crowley
Paul Crowley@ciphergoth·
@goodside No but the sheer shape of it, the incomplete sentence at the end.
English
0
0
1
38
Riley Goodside
Riley Goodside@goodside·
@ciphergoth I think this is an XKCD 2501 for us. My joke is mocking an archetype of an optimistic AI hype poster that isn’t familiar to people outside our corner of X. A typical SNL audience would not laugh at my joke.
English
1
0
1
117
Riley Goodside
Riley Goodside@goodside·
AI will take some jobs, but it will create countless new jobs too—exciting jobs we can’t even imagine yet. A year later those will also be done by AI, but there will be new jobs—exciting jobs we can’t even imagine yet. Six months later those too will be done by AI, but
English
90
53
603
158.4K
Paul Crowley
Paul Crowley@ciphergoth·
@jeanpaulblartre This is a sticky situation, but fortunately, you have an AI-powered coding assistant to help.
English
0
0
1
149
Paul Crowley
Paul Crowley@ciphergoth·
@goodside I'm sorry, it's extremely obvious and anyone who failed to see it only has themselves to blame.
English
1
0
5
150
Riley Goodside
Riley Goodside@goodside·
Update: I failed to make it obvious enough this post was a joke. My bad. The joke is that the first line, often said sincerely, in practice creates new jobs themselves replaceable in exponentially shorter amounts of time, which after several iterations is not at all reassuring.
English
8
2
86
14K