Dean Eckles's Avatar

Dean Eckles

@eckles.bsky.social

networks, contagion, causality faculty at MIT

9,568 Followers  |  1,301 Following  |  357 Posts  |  Joined: 06.07.2023  |  2.0653

Latest posts by eckles.bsky.social on Bluesky

This is the second account - the first being in Chicago - of DHS agents mass detaining people and then *sorting them by race*.

12.02.2026 02:26 β€” πŸ‘ 15382    πŸ” 7581    πŸ’¬ 519    πŸ“Œ 289
Take any two measurable classes of things in the
universe of perceptions, physical, organic, social or
economic, and it is such a dot or scatter diagram, which
we reach with extended observations. In some cases the
dots are scattered all over the paper, there is no association
of A and B ; in other cases there is a broad belt, there is
only moderate relationship ; then the dots narrow down to
a " comet's tail," and we have close association. Yet the
whole series of diagrams is continuous ; nowhere can you
draw a distinction and say here correlation ceases and causation begins. Causation is solely the conceptual limit to
correlation, when the band gets so attenuated, that it looks
like a curve.

Take any two measurable classes of things in the universe of perceptions, physical, organic, social or economic, and it is such a dot or scatter diagram, which we reach with extended observations. In some cases the dots are scattered all over the paper, there is no association of A and B ; in other cases there is a broad belt, there is only moderate relationship ; then the dots narrow down to a " comet's tail," and we have close association. Yet the whole series of diagrams is continuous ; nowhere can you draw a distinction and say here correlation ceases and causation begins. Causation is solely the conceptual limit to correlation, when the band gets so attenuated, that it looks like a curve.

It's like Karl Pearson's old and weird perspective on causation, from his "Grammar of Science" (2nd edition) book: "causation" is when the correlation is perfect. Nowhere does he notice the diff btw prediction and intervention.

11.02.2026 06:59 β€” πŸ‘ 32    πŸ” 3    πŸ’¬ 4    πŸ“Œ 1

Large Language Models was first used (AFAICT) to describe a google translate n-gram model, though it did have 300B parameters. aclanthology.org/D07-1090.pdf

10.02.2026 21:13 β€” πŸ‘ 8    πŸ” 1    πŸ’¬ 3    πŸ“Œ 0

Derek Chauvin cost the people of Minneapolis $27M in direct payments to George Floyd's family (equivalent to 1.6% of the city budget), plus millions more in property damage and economic harm.

ICE are over here saying 'hold my beer.'

11.02.2026 03:12 β€” πŸ‘ 10    πŸ” 5    πŸ’¬ 0    πŸ“Œ 2
Preview
Fragmentation of a longitudinal population-scale social network: Decreasing structural social cohesion in the Netherlands Population-level dynamics of social cohesion and its underlying mechanisms remain difficult to study. In this paper, we propose a network approach to measure the evolution of social cohesion at the po...

Fascinating new working paper by @bokanyie.bsky.social and colleagues showing a gradual decrease in social closure in the Netherlands over 10 years, based on population networks arxiv.org/abs/2602.002...

09.02.2026 12:31 β€” πŸ‘ 3    πŸ” 2    πŸ’¬ 0    πŸ“Œ 0
Preview
First Proof To assess the ability of current AI systems to correctly answer research-level mathematics questions, we share a set of ten math questions which have arisen naturally in the research process of the au...

Ten math problems with proofs known the authors. Proofs are encrypted until Feb 13. For all problems, authors claim both AI-based literature searches and zero-shot attempts at proofs failed. If you want to take a crack, you have until next Friday (2/13)!

08.02.2026 15:59 β€” πŸ‘ 25    πŸ” 5    πŸ’¬ 0    πŸ“Œ 3

Here’s the situation in the Twin Cities:

Random Bovino-style tear-gassing: Way down

Relentless persecution of immigrants: Unchanged

Thuggish violence against observers: Way up

08.02.2026 21:30 β€” πŸ‘ 9926    πŸ” 4007    πŸ’¬ 172    πŸ“Œ 90

Yeah don't seem to be any credible outlets saying this

07.02.2026 18:11 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

The other outcome that I've seen - when there are so many As students at the very top can't distinguish themselves through their classwork. Going really deep on a class and writing a truly superb paper won't change grade --> increased incentive to distinguish through extra curriculars.

07.02.2026 16:09 β€” πŸ‘ 25    πŸ” 6    πŸ’¬ 2    πŸ“Œ 0

β€œThey’re basically torturing people into signing off on their own deportation,” said one attorney.

Superb, deeply reported story, via @charpentier.bsky.social

07.02.2026 15:03 β€” πŸ‘ 1312    πŸ” 716    πŸ’¬ 19    πŸ“Œ 21
Preview
Immigrant whose skull was broken in eight places during ICE arrest says beating was unprovoked A Mexican immigrant whose skull was broken during his arrest by immigration officers last month in Minnesota says the beating was unprovoked.

ICE officials said he ran into a wall, which doctors immediately rejected as inconsistent with the wounds. The immigrant's account - that he was beaten in the head by steel batons, contrary to DHS guidelines - is more plausible. But no accountability exists for him, or us.
apnews.com/article/immi...

07.02.2026 16:06 β€” πŸ‘ 848    πŸ” 396    πŸ’¬ 20    πŸ“Œ 17
Preview
Trump administration won't let student deported to Honduras return President Donald Trump's administration on Friday called a judge's proposal that the government issue a student visa to a college student it deported to Honduras in violation of a court order "unfeasi...

New: Trump's administration tonight told a Boston judge that his proposal that it issue a student visa to a Babson College student it deported to Honduras in violation of a court order is "unfeasible" and said ICE declines to facilitate her return. www.reuters.com/world/us/tru...

07.02.2026 01:26 β€” πŸ‘ 231    πŸ” 131    πŸ’¬ 16    πŸ“Œ 31

One shortcoming of Claude Code it that it can't render math nicely in the terminal, and of course Latex math quickly gets very hard to read. Any tips for dealing with this? Are alternatives better?

06.02.2026 19:47 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 8    πŸ“Œ 0
Post image Post image

Okay well the OpenAI CMO is hopping on now and it is obvious that Anthropic hurt them and they were deeply unprepared

05.02.2026 04:22 β€” πŸ‘ 355    πŸ” 28    πŸ’¬ 31    πŸ“Œ 31
Preview
Spotlighting The World Factbook as We Bid a Fond Farewell Somewhat devastating news today from the CIA: One of CIA’s oldest and most recognizable intelligence publications, The World Factbook, has sunset. There's not even a hint as to why they …

The CIA just stopped publishing their World Factbook and took every page, including the archived copies of previous versions!

This sucks. It was public domain, so I recovered the 2020 edition (the last one published as a zip file) and shared it to GitHub simonwillison.net/2026/Feb/5/t...

05.02.2026 00:25 β€” πŸ‘ 6056    πŸ” 2188    πŸ’¬ 178    πŸ“Œ 196

Theory of mind and variations on the "false belief task" as benchmarks for LLMs... Benchmarks they seem to do poorly on

04.02.2026 14:33 β€” πŸ‘ 7    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

Remember the incident where DHS claimed officers shot a guy because he attacked them with a snow shovel?

Yeah. It didn't happen. They shot him through the door of his own home. And then arrested him. And then lied about it. Again.

And they almost shot some kids in the process.

03.02.2026 23:33 β€” πŸ‘ 8771    πŸ” 3797    πŸ’¬ 68    πŸ“Œ 64

Talk to me lil buddy
:)

04.02.2026 01:59 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Post image

🧡 New version of our paper (@bcegerod.bsky.social) is finally online: "How Many is Enough? Sample Size in Staggered Difference-in-Differences Designs"
We show that even well-identified DiD studies are often underpowered; sample sizes needed are surprisingly large
Paper: osf.io/preprints/os... 1/6

03.02.2026 14:46 β€” πŸ‘ 91    πŸ” 40    πŸ’¬ 3    πŸ“Œ 5
Post image

en.wikipedia.org/wiki/Please_...

02.02.2026 12:56 β€” πŸ‘ 711    πŸ” 53    πŸ’¬ 5    πŸ“Œ 13

We’ve got another @sjshancoxli.liberalcurrents.com banger on our hands folks

02.02.2026 11:06 β€” πŸ‘ 288    πŸ” 91    πŸ’¬ 4    πŸ“Œ 0
screencap of WH tweet

screencap of WH tweet

the White House's official X account gloats over the arrest of Don Lemon with a chains emoji

30.01.2026 15:10 β€” πŸ‘ 9022    πŸ” 3290    πŸ’¬ 1562    πŸ“Œ 966

Microsoft Research NYC is hiringΒ a researcher in the space of AI and society!

29.01.2026 23:27 β€” πŸ‘ 62    πŸ” 39    πŸ’¬ 1    πŸ“Œ 2
Post image

If Bilbo had Chat GPT

27.01.2026 16:24 β€” πŸ‘ 3717    πŸ” 1063    πŸ’¬ 39    πŸ“Œ 40
Preview
FBI to investigate Minneapolis activists after far-right claim about Signal chats The FBI director, Kash Patel, said the inquiry followed a far-right influencer’s post about anti-ICE Signal chats

"Patrick Eddington, of the libertarian Cato Institute, said the FBI had no business investigating. β€œThe use of encryption is as American as apple pie. The founders used it before, during, and after the revolution,” he said." www.theguardian.com/us-news/2026...

28.01.2026 04:43 β€” πŸ‘ 30    πŸ” 16    πŸ’¬ 0    πŸ“Œ 0

Andrew Huberman, Dog Phrenologist

28.01.2026 01:50 β€” πŸ‘ 2501    πŸ” 418    πŸ’¬ 93    πŸ“Œ 21

hivemind, i'm curious how you are dealing with llm/vibecoding in recruiting undergraduate research and masters RAs ..... so far the range has been pretty wide in terms of way overconfident use vs. being able to leverage tools to move along at a fairly effective pace

28.01.2026 01:05 β€” πŸ‘ 21    πŸ” 4    πŸ’¬ 3    πŸ“Œ 1
TikTok v. Garland and the First Amendment Anticanon <p><span>This essay argues that last term’s decision in <i>TikTok v. Garland</i>, which unanimously upheld the federal law that sought to ban TikTok in the Unit

Apropos of the TikTok news last week, I have a piece coming out abt how the Supreme Court's decision to uphold the ban-or-sale law belongs in the First Amendment anticanon. It's not just wrong, but so wrong we should hold it up as an exceptional symbol of wrongness.

papers.ssrn.com/sol3/papers....

26.01.2026 15:58 β€” πŸ‘ 51    πŸ” 23    πŸ’¬ 2    πŸ“Œ 2
CONTRIBUTOR
GREG BROCKMAN
AMOUNT
ELECTION TO DAT
$25,000,000 $25,000,000

CONTRIBUTOR GREG BROCKMAN AMOUNT ELECTION TO DAT $25,000,000 $25,000,000

The largest Trump superPAC donor so far this cycle is the president of OpenAI

26.01.2026 00:55 β€” πŸ‘ 1806    πŸ” 699    πŸ’¬ 33    πŸ“Œ 94

The replies to this post surfaced lots of great 3rd party tools (thanks!), but I still think it's interesting that bsky, as a platform, has decided to not prioritize algorithmic triangle closing (or follow recommendations, generally). See this fig from Zignani et al. (2014), when FB introduced PYMK.

25.01.2026 14:38 β€” πŸ‘ 27    πŸ” 2    πŸ’¬ 3    πŸ“Œ 2

@eckles is following 20 prominent accounts