Tessera

@hughgarner.bsky.social

Software engineering, data, visualization and systems. Occasional art things.

Henry Farrell @himself.bsky.social · 3mo ago

knightcolumbia.org/content/ai-a... Now up at @knightcolumbia.org - Cosma Shalizi and myself on "AI as Social Technology." It says that AI models are really a technology of social mediation that allow humans to connect with thoughts of other humans. Short-ish thread with major points follows.

AI as Social Technology

knightcolumbia.org

turgut keskintürk @tkeskinturk.bsky.social · 5mo ago

there is a fantastic visualization strategy coming from cultural evolution circles. I thought it'd be a perfect way to describe public opinion data, so I adapted it for aggregate survey data, created an R package, and wrote a post introducing it here: tkeskinturk.github.io/blog/publicopin

a plot showing % favoring laws against interracial marriage in the GSS.

Christopher Barrie @cbarrie.bsky.social · 6mo ago

🧵on my new paper "Synthetic personas distort the structure of human belief systems" w Roberto Cerina I'm v excited about... 🚨 Do synthetic samples look like human samples? We compare 28 LLMs to the 2024 General Social Survey (GSS) to find out + develop host of diagnostics...

Dan Davies @dsquareddigest.bsky.social · 6mo ago

in which I write many things which will look ironic when I am starving in the gutter begging for a few hours work tagging images for my robot overlords. backofmind.substack.com/p/finally-we...

finally we have created the silver bullet

from Fred Brooks' classic essay "No Silver Bullets"

backofmind.substack.com

Ted Underwood @tedunderwood.com · 6mo ago

The corollary to this is that figuring out how to do reliable evals and tests for tasks other than code — incl fuzzy, subjective tasks — has become a critical bottleneck and will receive a great deal of attention.

hikikomorphism@hikikomorphism.bsky.social · 6mo ago

AI agents are great for anything where there's a compiler in the loop, or a sandbox eval environment where you can do multiple zero-consequence test runs. Many problems lack these properties.

Robin Wigglesworth @robinwigglesworth.ft.com · 7mo ago

The FT got is architecture critic Edwin Heathcote to write about data centres and it's wonderful. www.ft.com/content/7692...

Simon Willison @simonwillison.net · 8mo ago

Let's build hyper-personalized AI-powered software that avoids the attention hijacking anti-patterns that defined so much of the last decade of software design - here's our manifesto with principles on how we can do that - more thoughts on my blog: simonwillison.net/2025/Dec/5/r...

The Resonant Computing Manifesto

Launched today at WIRED’s The Big Interview event, this manifesto (of which I'm a founding signatory) pushes for a positive framework for thinking about building hyper-personalized AI-powered software...

simonwillison.net

Alex Komoroske@komorama.bsky.social · 8mo ago

What if technology didn’t feel so… hollow? Some friends and I just released a manifesto about a world where tech leaves us feeling nourished (along with an evolving list of theses about how we can build it) resonantcomputing.org

Andrew Harrison @nndroid.bsky.social · 9mo ago

🌳 As a #NationalTrust member I’ve just voted for the recommended slate to keep out the bogus, right-wing entryists of “Restore Trust”. Anything this creep wants is bad for Britain and bad for our heritage. 🌳 If you’re a member, vote Recommended Slate now: www.nationaltrust.org.uk/who-we-are/a...

@teresapbn.bsky.social · 10mo ago

One vital reason to vote in the National Trust elections and for the NT endorsed candidates. We can't risk the Restore Trust candidates getting any influence on it.

Omar Khan @omaromalleykhan.bsky.social · 10mo ago

I know statistics are poorly understood and are misused. But facts matter Between the 1991, 2001, 2011 and 2021 Census *every* ethnic group in the UK has become *less* geographically segregated and *all* groups, majority and minorities, are more likely to interact with people not like them

John Doran @jahduran.bsky.social · 11mo ago

They only like you when you're... SEVENTEEN! Happy birthday to The Quietus! Please, please, please consider taking out a sub. We've introduced a new playlist perk. Sign up today and get 45 this month! Pls share & ask sugar mummies & daddies to dig deep. thequietus.com/news/the-qui...

The Quietus is 17! We're Celebrating With New Playlists for our Subscribers | The Quietus

It’s now five years since we launched our subscription platform with Steady in a last-ditch attempt to save The Quietus during the desperate Covid crisis. Thankfully, we raised enough to temporally keep the wolves from the door, and the website is still here to celebrate its seventeenth birthday. SEVENTEEN! We can barely believe that we’ve […]

thequietus.com

Julia M. Rohrer @dingdingpeng.the100.ci · 12mo ago

Ever stared at a table of regression coefficients & wondered what you're doing with your life? Very excited to share this gentle introduction to another way of making sense of statistical models (w @vincentab.bsky.social) Preprint: doi.org/10.31234/osf... Website: j-rohrer.github.io/marginal-psy...

Models as Prediction Machines: How to Convert Confusing Coefficients into Clear Quantities

Abstract
Psychological researchers usually make sense of regression models by interpreting coefficient estimates directly. This works well enough for simple linear models, but is more challenging for more complex models with, for example, categorical variables, interactions, non-linearities, and hierarchical structures. Here, we introduce an alternative approach to making sense of statistical models. The central idea is to abstract away from the mechanics of estimation, and to treat models as “counterfactual prediction machines,” which are subsequently queried to estimate quantities and conduct tests that matter substantively. This workflow is model-agnostic; it can be applied in a consistent fashion to draw causal or descriptive inference from a wide range of models. We illustrate how to implement this workflow with the marginaleffects package, which supports over 100 different classes of models in R and Python, and present two worked examples. These examples show how the workflow can be applied across designs (e.g., observational study, randomized experiment) to answer different research questions (e.g., associations, causal effects, effect heterogeneity) while facing various challenges (e.g., controlling for confounders in a flexible manner, modelling ordinal outcomes, and interpreting non-linear models).

Figure illustrating model predictions. On the X-axis the predictor, annual gross income in Euro. On the Y-axis the outcome, predicted life satisfaction. A solid line marks the curve of predictions on which individual data points are marked as model-implied outcomes at incomes of interest. Comparing two such predictions gives us a comparison. We can also fit a tangent to the line of predictions, which illustrates the slope at any given point of the curve.

A figure illustrating various ways to include age as a predictor in a model. On the x-axis age (predictor), on the y-axis the outcome (model-implied importance of friends, including confidence intervals).

Illustrated are
1. age as a categorical predictor, resultings in the predictions bouncing around a lot with wide confidence intervals
2. age as a linear predictor, which forces a straight line through the data points that has a very tight confidence band and
3. age splines, which lies somewhere in between as it smoothly follows the data but has more uncertainty than the straight line.

Ian Rennie @theangelremiel.bsky.social · 12mo ago

There are so many forms of art where it is no longer possible to make a living by being quite good, which is harmful to all forms of art. You should be able to make a living wage as a good enough piano player or a decent actor or a competent portrait painter because that's where people learn.

Leon English @leonenglish.bsky.social · last yr.

This excerpt is an exceptionally good look at what a lot of people seem to believe LLMs are or will become versus the reality.

Still, Wolfsthal has heard whispers of other concerning uses of LLMs in the heart of American power. "A number of people have said, 'Well, look, all I want to do is have an interactive computer available for the president so he can figure out what Putin or Xi will do and I can produce that dataset very reliably. I can get everything that Xi or Putin has ever said and written about anything and have a statistically high probability to reflect what Putin has said," he says.

"I was like, 'That's great. How do you know Putin believes what he's said or written?' It's not that the probability is wrong, it's just based on an assumption that can't be tested," Wolfsthal says.

Ned Boulting @nedboulting.bsky.social · last yr.

A massive thank you from Lizzie, David, Pete (who'll be there if he can be) and me. The response to this new venture has been overwhelming. Please sign up here (entirely for free) to make your support known. Let's get this up and running. neverstraysfar.com

Never Strays Far | Cycling Podcast

Hosted by Ned Boulting, David Millar and Pete Kennaugh. The three talk about lots of things, although they never stray far from the core issue, bike racing. Expect guests from the racing world and som...

neverstraysfar.com

Cycling Weekly@cyclingweekly.com · last yr.

There's no way we're not going to be there' – Ned Boulting, David Millar and Lizzie Deignan launch plans for free audio and video Tour de France coverage www.cyclingweekly.com/racing/there...

Cycling Weekly @cyclingweekly.com · last yr.

There's no way we're not going to be there' – Ned Boulting, David Millar and Lizzie Deignan launch plans for free audio and video Tour de France coverage www.cyclingweekly.com/racing/there...

'There's no way we're not going to be there' – Ned Boulting, David Millar and Lizzie Deignan launch plans for free audio and video Tour de France coverage

NSF Live in France to cover Tour in 2026 and beyond, replacing free-to-air coverage

cyclingweekly.com

John Doran @jahduran.bsky.social · last yr.

I met up with @maryannehobbs.bsky.social last week at the Tate to talk about her new radio show, but instead found myself on a lazy river considering free will, art, conflict and life on a NWOBHM bus... thequietus.com/interviews/m...

A Condition of the Space: Mary Anne Hobbs Interviewed | The Quietus

Ahead of the launch of her new radio show on 6 Music & an appearance at MIF, Mary Anne Hobbs would sooner talk about an old bus...

thequietus.com

Daniel Devine @danjdevine.bsky.social · last yr.

Plenty of European countries - including the UK - are doing this, though in more of a foot-on-the-oxygen-line than a bullet-to-the-head way.

Paul Graf@paulginva.bsky.social · last yr.

Man comes to the US from Lebanon. Starts out delivering pizzas, becomes a Nobel winning neuroscientist. Trump freezes his funding, he gets an email from China offering to move his lab “any city, any university I want" with guaranteed funding for 20 years. What are we doing?

Ardem Patapoutian's story is not just the American dream, it is the dream of American science.

He arrived in Los Angeles in 1986 at age 18 after fleeing war-torn Lebanon. He spent a year writing for an Armenian newspaper and delivering Domino's at night to become eligible for the University of California, where he earned his undergraduate degree and a postdoctoral fellowship in neuroscience.

He started a lab at Scripps Research in San Diego with a grant from the National Institutes of Health, discovered the way humans sense touch, and in 2021 won the Nobel Prize.

But with the Trump administration slashing spending on science, Dr. Patapoutian's federal grant to develop new approaches to treating pain has been frozen. In late February, he posted on Bluesky that such cuts would damage biomedical research and prompt an exodus of talent from the United States. Within hours, he had an email from China, offering to move his lab to "any city, any university I want," he said, with a guarantee of funding for the next 20 years.

Henry Farrell @himself.bsky.social · last yr.

"When we think of AI, we should think less of Terminator 2 and more of the TV show Severance, in which office workers search for “bad numbers” on the strength of vibes alone." www.theideasletter.org/essay/our-sp...

Our Spreadsheet Overlords - The Ideas Letter

Weatherby argues that the current discourse around AI, especially the buzz around “artificial general intelligence,” is a distraction from its actual impact: the expansion of bureaucracy through massi...

theideasletter.org

Simon Willison @simonwillison.net · last yr.

Once again, if your LLM system combines access to private data, exposure to malicious instructions and the ability to exfiltrate information (through tool use or through rendering links and images) you have a nasty security hole This time, GitLab: simonwillison.net/2025/May/23/...

Remote Prompt Injection in GitLab Duo Leads to Source Code Theft

Yet another example of the classic [Markdown image exfiltration attack](https://simonwillison.net/tags/exfiltration-attacks/), this time affecting GitLab Duo - GitLab's chatbot. Omer Mayraz reports on...

simonwillison.net

Jonathan Hopkin @jonathanhopkin.eurosky.social · last yr.

How does destroying the university sector help growth I wonder

Margot Finn@eicathomefinn.bsky.social · last yr.

'UK research council leaders are bracing for cuts to quality-related (QR) research funding of up to £100 million a year ahead of next week’s Spring Statement – a move likely to lead to a fresh round of redundancies at UK universities.' 1/3

Henry Farrell @himself.bsky.social · last yr.

1. @alisongopnik.bsky.social, Cosma Shalizi, James Evans and myself have a new piece in Science on "AI" Large Models, pushing back against much of the collective wisdom about what they can and can't do. Official below, unpaywalled at henryfarrell.net/large-ai-mod... . So why this now?

Large AI models are cultural and social technologies

Implications draw on the history of transformative information systems from the past

science.org

Luc Rocher @rocher.lc · 2y ago

Curious to see how they plan to anonymise our NHS data for training ML models. I personally wouldn’t know how to start doing that www.theguardian.com/politics/202...

The government plan features a potentially controversial scheme to unlock public data to help fuel the growth of AI businesses. This includes anonymised NHS data, which will be available for “researchers and innovators” to train their AI models. The government says there would be “strong privacy-preserving safeguards” and the data would never be owned by private companies.