Paul Crowley
@ciphergoth.org
I don't really come here any more sorry!
Cipherpunk is more of a technical/political movement than a literary genre I think!
October 14, 2025 at 2:57 AM
Cipherpunk is more of a technical/political movement than a literary genre I think!
What was this meant to be a link to, do you remember?
September 8, 2025 at 2:15 PM
What was this meant to be a link to, do you remember?
What caused you to believe that Eliezer Yudkowsky is wealthy? His salary is public info in MIRI disclosures, and it's vastly below what Silicon Valley people at his level make.
July 24, 2025 at 6:23 PM
What caused you to believe that Eliezer Yudkowsky is wealthy? His salary is public info in MIRI disclosures, and it's vastly below what Silicon Valley people at his level make.
Oh how frustrating :( Is there a way to preview the Editors Copy or is the only way to build it myself?
May 6, 2025 at 8:35 PM
Oh how frustrating :( Is there a way to preview the Editors Copy or is the only way to build it myself?
Ah yes, it has been renewed - thanks! However on github.com/dconnolly/dr... "Editor's Copy" is still a 404 :(
GitHub - dconnolly/draft-connolly-cfrg-xwing-kem: I-D for a general purpose KEM (key encapsulation mechanism) that includes a hash at the end
I-D for a general purpose KEM (key encapsulation mechanism) that includes a hash at the end - dconnolly/draft-connolly-cfrg-xwing-kem
github.com
May 6, 2025 at 7:00 PM
Ah yes, it has been renewed - thanks! However on github.com/dconnolly/dr... "Editor's Copy" is still a 404 :(
@katyha.bsky.social I'm no fan of Eigen, but he's not being serious. The lack of punctuation before "chat is this real" is a tell. The reply is continuing and intensifying the joke.
February 5, 2025 at 3:27 PM
@katyha.bsky.social I'm no fan of Eigen, but he's not being serious. The lack of punctuation before "chat is this real" is a tell. The reply is continuing and intensifying the joke.
You can't, and it's bizarre that so many seem to believe you can. The nonlinear components are central to how the AI works.
January 24, 2025 at 5:03 AM
You can't, and it's bizarre that so many seem to believe you can. The nonlinear components are central to how the AI works.
Wow only 707, I would never have guessed it was so few!
January 17, 2025 at 1:50 PM
Wow only 707, I would never have guessed it was so few!
How are you seeing this number?
January 17, 2025 at 4:57 AM
How are you seeing this number?
Link to the story here? If I search for "asmongold dms" what I find says Elon "leaked" DMs between Asmongold and Elon.
January 17, 2025 at 4:52 AM
Link to the story here? If I search for "asmongold dms" what I find says Elon "leaked" DMs between Asmongold and Elon.
Astonishingly, the unsubscribe link in such emails practically always works!
January 17, 2025 at 4:39 AM
Astonishingly, the unsubscribe link in such emails practically always works!
Where does this number come from? Isn't "several trillion" closer to the correct answer, even if we only count bees in beehives?
January 17, 2025 at 4:31 AM
Where does this number come from? Isn't "several trillion" closer to the correct answer, even if we only count bees in beehives?
This is a good negative example - it shows that you can meet this spec without really making progress. You need a stronger form of uniformity, like random self-reduction.
January 11, 2025 at 3:24 AM
This is a good negative example - it shows that you can meet this spec without really making progress. You need a stronger form of uniformity, like random self-reduction.
How do you relate the hardness of the instances you generate to the hardness of the hardest instances?
January 11, 2025 at 2:45 AM
How do you relate the hardness of the instances you generate to the hardness of the hardest instances?
I think Paul's observation cuts against that, that the richness of structure required to be NP-complete is the opposite of the uniformity required for average-case hardness.
January 11, 2025 at 2:27 AM
I think Paul's observation cuts against that, that the richness of structure required to be NP-complete is the opposite of the uniformity required for average-case hardness.
I think discrete logarithm is a better example than factoring for this!
January 11, 2025 at 2:01 AM
I think discrete logarithm is a better example than factoring for this!
Many have tried to build cryptosystems out of NP-complete problems, but it repeatedly turns out that the actual instances of the problem generated by the cryptosystem are not hard in the average case.
January 11, 2025 at 1:51 AM
Many have tried to build cryptosystems out of NP-complete problems, but it repeatedly turns out that the actual instances of the problem generated by the cryptosystem are not hard in the average case.
But this reflects that it just doesn't have much internal structure. An instance of an NP complete problem has so much internal structure you can use it to encode any other problem. That strongly cuts against any hope of proving anything useful about average-case hardness!
January 11, 2025 at 1:51 AM
But this reflects that it just doesn't have much internal structure. An instance of an NP complete problem has so much internal structure you can use it to encode any other problem. That strongly cuts against any hope of proving anything useful about average-case hardness!
Yes! I asked Paul Christiano this question and his answer has stayed with me since. The discrete logarithm problem (eg) can't have "easy" and "hard" instances, because you can do "random self-reduction", so you can use an algorithm that breaks 1% of instances to break any instance in 100x the time.
January 11, 2025 at 1:51 AM
Yes! I asked Paul Christiano this question and his answer has stayed with me since. The discrete logarithm problem (eg) can't have "easy" and "hard" instances, because you can do "random self-reduction", so you can use an algorithm that breaks 1% of instances to break any instance in 100x the time.
I gave him kisses and told him they were from the Internet
January 10, 2025 at 10:16 PM
I gave him kisses and told him they were from the Internet
we have a large, a medium, and a small! Thor is the large!
January 9, 2025 at 10:49 PM
we have a large, a medium, and a small! Thor is the large!
I want them to support the Democrats but push back on their excesses, towards being a more electable party.
January 9, 2025 at 9:02 PM
I want them to support the Democrats but push back on their excesses, towards being a more electable party.
he is in fact a lovely sweet baby and allows us to huff him right in his middle
January 9, 2025 at 7:13 PM
he is in fact a lovely sweet baby and allows us to huff him right in his middle