I was writing even as you were tagging :) bsky.app/profile/tedu...
Everyone’s chiming in with good ideas.
My additional 2¢ is that if you have a free text “notes” field handling the mess with comments like “name unreliable; also goes by Childers or Chidders” … that may actually be just fine these days. 10 years ago those would’ve been harder to use at scale.
My additional 2¢ is that if you have a free text “notes” field handling the mess with comments like “name unreliable; also goes by Childers or Chidders” … that may actually be just fine these days. 10 years ago those would’ve been harder to use at scale.
October 27, 2026 at 11:16 PM
I was writing even as you were tagging :) bsky.app/profile/tedu...
these are people of the land—the common clay of the new west
Main takeaway from the NYT Iowa focus group is that none of these voters really know what’s going on and they barely know anything about the candidates running in either statewide election lmao
October 23, 2026 at 6:03 PM
(apart from all that: the famous heterophenomenological jungian synchronicity, more interesting/revealing in fact than all that cross-reading and cross-referencing all those papers & and all those comments on all those conscious ghosts and bots)
bsky.app/profile/tedu...
bsky.app/profile/tedu...
The word “proprioception” is a cognitohazard. Because after you read it you wonder “how” you know the relative position of your body parts in space. And if you question that too closely while sitting, you lose faith that you do know, and start to feel like a gyroscope tumbling through space
October 21, 2026 at 12:44 PM
(apart from all that: the famous heterophenomenological jungian synchronicity, more interesting/revealing in fact than all that cross-reading and cross-referencing all those papers & and all those comments on all those conscious ghosts and bots)
bsky.app/profile/tedu...
bsky.app/profile/tedu...
so far the sole voice of dissent bsky.app/profile/tedu...
“You Can See Everything” is so deeply not my genre that these positive reviews are like reading “try their pineapple-anchovy pizza; honest to God, best pizza you ever had.” I can imagine a creature who would enjoy that, but cannot imagine being the creature.
October 18, 2026 at 7:17 AM
so far the sole voice of dissent bsky.app/profile/tedu...
But I majored in philosophy with a heavy continental focus when post-structuralism was at peak popularity so am maybe just hopelessly Foucault-pilled
bsky.app/profile/tedu...
bsky.app/profile/tedu...
Yeah, it’s by no means unique to me. I was trained in disciplines that have argued for more than a hundred years that human thought consists quite largely in predicting the next word. I spin it here as a structuralist insight, but … critinq.wordpress.com/2023/06/29/t...
The Empirical Triumph of Theory
Ted Underwood 29 June 2023 A graduate student who fell asleep in 1982 and woke up in 2022 might see large language models as a triumph for cultural theory. It is hard to imagine a clearer vindicati…
critinq.wordpress.com
October 13, 2026 at 6:52 PM
But I majored in philosophy with a heavy continental focus when post-structuralism was at peak popularity so am maybe just hopelessly Foucault-pilled
bsky.app/profile/tedu...
bsky.app/profile/tedu...
Something not sufficiently registered in the convo around AI/IP: the value that was gained from pretraining (on our collective legacy) is now common property.
People do pay more for frontier models than open ones — but the differences they're paying for, at this point, come from post-training.
People do pay more for frontier models than open ones — but the differences they're paying for, at this point, come from post-training.
Fortunately, that bridge itself is easy as heck to copy and make free.
At this point pretty much everything that was gained from the initial pre-training is in fact, universally shared. Differences between models that people pay for are now all post training.
At this point pretty much everything that was gained from the initial pre-training is in fact, universally shared. Differences between models that people pay for are now all post training.
October 11, 2026 at 10:51 PM
my favorite post on such subjects bsky.app/profile/tedu...
You've entered definition-land, @hikikomorphism.bsky.social. Nothing new is ever learned in definition-land, and nothing unexpected can happen. It's impossible — by definition. Anything that happened would just be an event, and we've seen events before.
October 11, 2026 at 1:46 AM
my favorite post on such subjects bsky.app/profile/tedu...
Anyone know why all the sites run by Ted Unangst, the developer of Honk and azorius, seem to have gone down at the end of last year? Is Ted OK?
https://web.archive.org/web/20251207001422/https://honk.tedunangst.com/u/tedu
#honk #azorius #tedunangst
https://web.archive.org/web/20251207001422/https://honk.tedunangst.com/u/tedu
#honk #azorius #tedunangst
tedu
Purveyor of modest software.
web.archive.org
October 10, 2026 at 1:13 AM
Anyone know why all the sites run by Ted Unangst, the developer of Honk and azorius, seem to have gone down at the end of last year? Is Ted OK?
https://web.archive.org/web/20251207001422/https://honk.tedunangst.com/u/tedu
#honk #azorius #tedunangst
https://web.archive.org/web/20251207001422/https://honk.tedunangst.com/u/tedu
#honk #azorius #tedunangst
anyone know what happened to @tedu of honk fame? Is he alright or why did he disappear from the internet suddenly?
https://web.archive.org/web/20250408034245/https://honk.tedunangst.com/about
https://web.archive.org/web/20250408034245/https://honk.tedunangst.com/about
honk.tedunangst.com
web.archive.org
October 9, 2026 at 11:54 PM
anyone know what happened to @tedu of honk fame? Is he alright or why did he disappear from the internet suddenly?
https://web.archive.org/web/20250408034245/https://honk.tedunangst.com/about
https://web.archive.org/web/20250408034245/https://honk.tedunangst.com/about
related point on 'social value of AI training on the common crawl' bsky.app/profile/tedu...
Something not sufficiently registered in the convo around AI/IP: the value that was gained from pretraining (on our collective legacy) is now common property.
People do pay more for frontier models than open ones — but the differences they're paying for, at this point, come from post-training.
People do pay more for frontier models than open ones — but the differences they're paying for, at this point, come from post-training.
Fortunately, that bridge itself is easy as heck to copy and make free.
At this point pretty much everything that was gained from the initial pre-training is in fact, universally shared. Differences between models that people pay for are now all post training.
At this point pretty much everything that was gained from the initial pre-training is in fact, universally shared. Differences between models that people pay for are now all post training.
October 7, 2026 at 4:05 PM
related point on 'social value of AI training on the common crawl' bsky.app/profile/tedu...
Oh sure. It's my pinned post, but I'll connect it here too! bsky.app/profile/tedu...
Paper: arxiv.org/abs/2609.23178
Code: github.com/Historical-A...
Team: Ted Underwood, Ziliang Qiu, Laura K. Nelson @lauraknelson.bsky.social, Sarah Griebel @sgriebel.bsky.social, Edwin Roland @teddyroland.bsky.social, Wenyi Shang @wenyishang.bsky.social , Matthew Wilkens @mattwilkens.bsky.social.
Code: github.com/Historical-A...
Team: Ted Underwood, Ziliang Qiu, Laura K. Nelson @lauraknelson.bsky.social, Sarah Griebel @sgriebel.bsky.social, Edwin Roland @teddyroland.bsky.social, Wenyi Shang @wenyishang.bsky.social , Matthew Wilkens @mattwilkens.bsky.social.
Chronologic: Measuring Language Models' Ability to Represent the Past
Language models are appealing tools for research on the past. But to trust the evidence a model provides, researchers need to know whether its responses fit the period represented. Validation is chall...
arxiv.org
October 7, 2026 at 6:27 PM
Oh sure. It's my pinned post, but I'll connect it here too! bsky.app/profile/tedu...
This is a fuller explanation of the point: bsky.app/profile/tedu...
Something not sufficiently registered in the convo around AI/IP: the value that was gained from pretraining (on our collective legacy) is now common property.
People do pay more for frontier models than open ones — but the differences they're paying for, at this point, come from post-training.
People do pay more for frontier models than open ones — but the differences they're paying for, at this point, come from post-training.
Fortunately, that bridge itself is easy as heck to copy and make free.
At this point pretty much everything that was gained from the initial pre-training is in fact, universally shared. Differences between models that people pay for are now all post training.
At this point pretty much everything that was gained from the initial pre-training is in fact, universally shared. Differences between models that people pay for are now all post training.
October 5, 2026 at 11:54 PM
This is a fuller explanation of the point: bsky.app/profile/tedu...
The faster we can get to this (or some other form of legal certainty around usage of copyright work for training) - the more opportunities we will have towards more democratized AI, where people can share data and findings freely and improve on each other's work
bsky.app/profile/tedu...
bsky.app/profile/tedu...
It's okay to learn from everything, and train on everything. That's everyone's birthright, and we shouldn't apologize for it.
Prohibitions on copying were a late invention, and a narrowly tailored one. They do not express a general ethical law that all information ought to be priced and paid for.
Prohibitions on copying were a late invention, and a narrowly tailored one. They do not express a general ethical law that all information ought to be priced and paid for.
October 5, 2026 at 8:09 PM
The faster we can get to this (or some other form of legal certainty around usage of copyright work for training) - the more opportunities we will have towards more democratized AI, where people can share data and findings freely and improve on each other's work
bsky.app/profile/tedu...
bsky.app/profile/tedu...
You've entered definition-land, @hikikomorphism.bsky.social. Nothing new is ever learned in definition-land, and nothing unexpected can happen. It's impossible — by definition. Anything that happened would just be an event, and we've seen events before.
September 16, 2026 at 5:07 PM
well known problem; you just have to keep enlarging the video until you see James Clerk Maxwell messing w/ us like a sophon
September 15, 2026 at 4:57 PM
Or in better form: bsky.app/profile/tedu...
But this comes back around to why I do support the statement Tao et al made. I like it more than the Leiden Declaration, because it doesn’t pretend it can regulate or limit change. It explains why we value knowledge-creation in the first place, and says “whatever comes next must do this too.”
This is not just about math, of course. It’s about the relation between SV and academia.
I think we’re right to say “good knowledge production is reflective about what has been discovered, how, and how that relates to previous inquiry.” But that will not stop our fields from being transformed.
I think we’re right to say “good knowledge production is reflective about what has been discovered, how, and how that relates to previous inquiry.” But that will not stop our fields from being transformed.
September 13, 2026 at 12:58 PM
Or in better form: bsky.app/profile/tedu...
bsky.app/profile/tedu...
I assume that OpenAI and Anthropic are constantly spying on everything the other does so that they can do it first or simultaneously, so it did not even occur to them that it was wrong
I assume that OpenAI and Anthropic are constantly spying on everything the other does so that they can do it first or simultaneously, so it did not even occur to them that it was wrong
Yeah. I'm sure it doesn't strike them as weird because they're in a space where "we threw lots of compute at this in order to scoop our competitors, who we heard were doing the same thing" is totally normal competitive behavior.
However, I don't think that's the social norm in math or science.
However, I don't think that's the social norm in math or science.
September 11, 2026 at 1:21 AM
bsky.app/profile/tedu...
I assume that OpenAI and Anthropic are constantly spying on everything the other does so that they can do it first or simultaneously, so it did not even occur to them that it was wrong
I assume that OpenAI and Anthropic are constantly spying on everything the other does so that they can do it first or simultaneously, so it did not even occur to them that it was wrong
You've entered definition-land, @hikikomorphism.bsky.social. Nothing new is ever learned in definition-land, and nothing unexpected can happen. It's impossible — by definition. Anything that happened would just be an event, and we've seen events before.
September 9, 2026 at 7:43 PM
Actually sympathetic to Tufekci’s conclusion that we shouldn’t expect jobpocalypse, but people have got to stop grounding that belief in the premise that we have some mysterious power only achievable by protein colloids.
Rather: in primate society, lots of roles can only be filled by primates.
Rather: in primate society, lots of roles can only be filled by primates.
September 9, 2026 at 12:30 AM
I don’t know if they’ve gone through user data but @tedunderwood.com makes a good point here: bsky.app/profile/tedu...
I think if you start by taking OpenAI 100% at their word, the phrase "inspired by these rumors" still announces "we did something fairly slimy by trying to scoop priority." Maybe it doesn't read as slimy in the social world they inhabit, but it would in mine.
September 9, 2026 at 1:58 AM
I don’t know if they’ve gone through user data but @tedunderwood.com makes a good point here: bsky.app/profile/tedu...
I guess I was beat to the punch on this thought bsky.app/profile/tedu...
As it becomes clear that language models—like humans—love passing notes to each other on message boards, I’m starting to think the thing we need to worry about is not the “alignment” of an isolated model but, like, the fluid dynamics of a language slurry.
September 4, 2026 at 9:12 PM
I guess I was beat to the punch on this thought bsky.app/profile/tedu...
I definitely think that it is still really early, but ideas and writing also diffused a lot more slowly in the the 19th century... bsky.app/profile/tedu...
It took a few decades!
1776: First commercial Boulton and Watt steam engine
1813: Robert Owen, A New View of Society
1848: Marx and Engels, Communist Manifesto
1776: First commercial Boulton and Watt steam engine
1813: Robert Owen, A New View of Society
1848: Marx and Engels, Communist Manifesto
August 31, 2026 at 6:04 PM
I definitely think that it is still really early, but ideas and writing also diffused a lot more slowly in the the 19th century... bsky.app/profile/tedu...
What remains A)The Single Market:Retention of a lean free-trade area without tariffs & trade barriers B)Transnational Border Defense: #Frontex & transnat. military cooperation(TEDU)).Isolated internal border management is logistically& economically more complex than a shared external perimeter
---
---
August 26, 2026 at 2:46 AM
What remains A)The Single Market:Retention of a lean free-trade area without tariffs & trade barriers B)Transnational Border Defense: #Frontex & transnat. military cooperation(TEDU)).Isolated internal border management is logistically& economically more complex than a shared external perimeter
---
---
Some of the replies are good though. bsky.app/profile/tedu...
Um. I don't want to freak you out.
But, knowing your brother, is there any chance that he's planning to transform himself into Spiderman's nemesis, The Scorpion?
But, knowing your brother, is there any chance that he's planning to transform himself into Spiderman's nemesis, The Scorpion?
August 20, 2026 at 1:57 PM
Some of the replies are good though. bsky.app/profile/tedu...
[RE: https://cafe.nilfm.cc/u/nilix/h/8Q928sc6DF22T4W3H2]
RIP tedu
they still doing weird shit with mastodon
we still out here
RIP tedu
they still doing weird shit with mastodon
we still out here
August 19, 2026 at 3:01 AM
[RE: https://cafe.nilfm.cc/u/nilix/h/8Q928sc6DF22T4W3H2]
RIP tedu
they still doing weird shit with mastodon
we still out here
RIP tedu
they still doing weird shit with mastodon
we still out here