Dj Ripley
LightNews LightNews
djripley.bsky.social
Dj Ripley
@djripley.bsky.social
2.7K followers 990 following 1.5K posts
Scholar, teacher, DJ, radical, junglist.
Former streetmedic, current assoc. prof. More at http://larisa-mann.com
She/her.
I wrote : RUDE CITIZENSHIP: copyright, Jamaican popular music & reverberations of colonial power https://tinyurl.com/RudeCBook
Posts Replies Media Videos
Pinned
djripley.bsky.social
Dj Ripley @djripley.bsky.social · Nov 20
Hi new followers 👋🏻 FYI I am a luddite (laudatory), an anti-zionist secular Jew (like my grandparents) for Palestinian liberation, against white supremacy. 27 years as a dj &event organizer = praxis & care work. Related: I help care for my elderly mother. Against cop shit (in education & everywhere).
Reposted by Dj Ripley
broadwaybabyto.bsky.social
A Google software engineer won his battle against MetLife to receive disability benefits for Long Covid.

This is huge

Long Covid has a lot of stigma attached to it and it can be hard to get disability.

You’re told to “try harder”.

It’s “not that bad”.

You can “still work”.

Not everyone can
MetLife Owes Disability Pay to Software Engineer With Long Covid
A Google LLC software engineer suffering from long Covid won his lawsuit seeking long-term disability benefits from Metropolitan Life Insurance Co.
news.bloomberglaw.com
November 28, 2025 at 8:45 PM
Reposted by Dj Ripley
fightforthefuture.org
Tell your lawmakers to oppose the Kids Online Safety Act- there are better ways to protect kids online without throwing other vulnerable people under the bus.

www.stopkosa.com
November 28, 2025 at 5:54 PM
Reposted by Dj Ripley
moviessilently.bsky.social
This is absolutely true. I worked at a small indie store and if a book that wasn't from a major author sold within a month of being brought in, we would reorder. The margin for success is tiny in small bookstores.
alyssaharad.bsky.social Alyssa Harad @alyssaharad.bsky.social · 3d
I once asked a bookseller at a large indie store how many people would have to buy a book for it to get the attention of the store buyer and cause an additional order and they said: Three.
blipstress.bsky.social violet allen @blipstress.bsky.social · 5d
I see some book piracy discourse, and, to make a positive argument in favor of buying books, your marginal ability to influence what books get published and support the careers of writers you like is massive compared to most other forms of media.
November 28, 2025 at 3:26 PM
Reposted by Dj Ripley
coreybrickley.bsky.social
artists and designers using ai like
cartoon of a cute little monkey using a saw to cut off a tree branch that it is standing on. The monkey is saying "Wow technology is amazing"
November 27, 2025 at 12:35 AM
Reposted by Dj Ripley
pastures.myatproto.social
Here's one for the people who are new here:

wikileaks.org/clinton-emai...
From: Jake Sullivan To: Hillary Clinton Date: 2012-02-12 04:01 Subject: SPOT REPORT 02/12/II UNCLASSIFIED U.S. Department of State Case No. F- 2014-20439 Doc No. C05789138 Date: 10/30/2015 RELEASE IN FULL From: Sullivan, JacobJ <SullivanJJ@state.gov> Sent: Sunday, February 12, 2012 4:01 PM To: Subject: Fw:SPOTREPORT02/12/11(SBU) See last item - AQ is on our side in Syria. Otherwise, things have basically turned out as expected.
November 28, 2025 at 7:52 PM
Reposted by Dj Ripley
fnwave.bsky.social
simple really 💅
Jean Luc Godard saying the solution is to stop making films for imperialism.
November 26, 2025 at 2:03 AM
Reposted by Dj Ripley
blackamazon.bsky.social
The first legislators arrested for inspecting these spaces were Black legislators from New Jersey
laurenunderwood.bsky.social Lauren Underwood @laurenunderwood.bsky.social · 4d
Today, after months of requests, I toured the Broadview ICE Processing Facility. I want to share what I saw.

The facility has no food vendor, no medical care, and detainees have to use toilets in the middle of shared cells. These are in no way suitable conditions to be holding anyone — period.
November 25, 2025 at 9:10 PM
Reposted by Dj Ripley
matthewcort.land
Robert F Kennedy Jr, whose body count goes back at least as far 80 dead human beings in Samoa, has appointed a murderer by the name of Ralph Abraham as the principal deputy director of the U.S. Centers for Disease Control and Prevention, we learned earlier today.

www.patreon.com/posts/144415...
One Murderer (RFK Jr) Appoints Another Murderer As #2 At MAGA's CDC | Matthew Cortland
Get more from Matthew Cortland on Patreon
www.patreon.com
November 26, 2025 at 1:55 AM
Reposted by Dj Ripley
kendrawrites.com
From the new HBO documentary Thoughts & Prayers that looks into the impact of the $3 billion active shooter preparedness industry on schools and communities.

If you're not American/don't live in the US, I'm asking you to please not comment.
November 26, 2025 at 1:18 AM
Reposted by Dj Ripley
anamariecox.bsky.social
Scientists have a point!
Is ChatGPT Conscious? Many users feel they're talking to a real person. Scientists say it's time (edited) for them to get out more.
November 25, 2025 at 10:14 PM
Reposted by Dj Ripley
clancyny.bsky.social
Quite a pair. One was arrested. For journalism. One shot in the face with a less lethal munition for ... journalism.

The rest of the country would know far less about Chicago's Operation Midway Blitz and the Border Patrol without them.

If you can, pls throw @unraveledpress.com bucks to keep going.
unraveledpress.com unraveled @unraveledpress.com · 4d
~1 yr ago we sent a Giving Tuesday email to our tiny mailing list shortly after starting our own publication.

In the time since, we have been overwhelmed by how many of you find our work worth supporting.

No email this time around. But if you'd like to support indie media today, here's how:
Support Unraveled
unraveledpress.com
November 25, 2025 at 7:01 PM
Reposted by Dj Ripley
leavittalone.bsky.social
My grand unified theory of suburbia is that camera doorbells are like SUVs, they provide an illusion of security while making you more paranoid
brooklynspoke.bsky.social Doug Gordon @brooklynspoke.bsky.social · 4d
“Camera doorbells are social cancer as a technology, breaking down social solidarity and further encouraging people to see strangers with suspicion.”
parismarx.com Paris Marx @parismarx.com · 4d
Christmas is coming, and the Black Friday sales signal we’re in peak gift-buying season.

When you’re picking out gifts for friends and family this year, there’s a whole slew of tech products that are better to avoid. I put together a guide to help you through it.
November 25, 2025 at 5:40 PM
Reposted by Dj Ripley
davidjbier.bsky.social
ICE deported a woman who owned a home in the US for 30 years. A judge banned ICE from deporting her to Sierra Leone where she had been tortured, so it deported her to Ghana, which kidnapped her and sent her back Sierra Leone anyway. ICE is part of this illegal conspiracy.
November 25, 2025 at 4:40 PM
Reposted by Dj Ripley
ketanjoshi.co
Is there no simple gut instinct at work here? "This thing lies by design, and unchangeably based on its fundamental technological construction, and is therefore unsuited to replacing human science communication"

Nope. Just gets shrugged off as a minor whoopsie
November 25, 2025 at 7:11 PM
Reposted by Dj Ripley
ketanjoshi.co
The more I think about this, the more it really, really, REALLY bothers me

We've reached a point where a text-based reference fabrication engine - comfortably the most effective machine liar we've ever invented - is being pushed by an ultra-polluting tech company and accepted by climate scientists
ketanjoshi.co Ketan Joshi @ketanjoshi.co · 5d
Deeply absurd. This Google PDF published on a blog (arxiv, not peer reviewed) claims an LLM is "PhD level" but in most cases the MAJORITY of reference URLs were invalid or inaccessible.

A PhD sitting down and just fabricating >50% of sources = career ending

arxiv.org/abs/2511.11597
CLINB: A Climate Intelligence Benchmark for Foundational Models Michelle Chen Huebscher1 , Katharine Mach2 , Aleksandar Stanić1 , Markus Leippold1,3, Ben Gaiarin1 , Zeke Hausfather4 , Elisa Rawat , Erich Fischer5 , Massimiliano Ciaramita1 , Joeri Rogelj6 , Christian Buck1 , Lierni Sestorain Saralegui1 and Reto Knutti5 1Google DeepMind, 2University of Miami, 3University of Zurich, 4Stripe, 5ETH Zurich, 6 Imperial College London Evaluating how Large Language Models (LLMs) handle complex, specialized knowledge remains a critical challenge. We address this through the lens of climate change by introducing CLINB, a benchmark that assesses models on open-ended, grounded, multimodal question answering tasks with clear requirements for knowledge quality and evidential support. CLINB relies on a dataset of real users’ questions and evaluation rubrics curated by leading climate scientists. We implement and validate a model-based evaluation process and evaluate several frontier models. Our findings reveal a critical dichotomy. Frontier models demonstrate remarkable knowledge synthesis capabilities, often exhibiting PhD-level understanding and presentation quality. They outperform “hybrid" answers curated by domain experts assisted by weaker models. However, this performance is countered by failures in grounding. The quality of evidence varies, with substantial hallucination rates for references and images. We argue that bridging this gap between knowledge synthesis and verifiable attribution is essential for the deployment of AI in scientific workflows and that reliable, interpretable benchmarks like CLINB are needed to progress towards building trustworthy AI systems. Total Reference URLs Generated claude-opus-4-1 claude-sonnet-4 gpt-5 hybrid gemini-2.5-pro gemini-2.5-flash o3 0.0 0.2 0.4 0.6 0.8 1.0 Proportion Reference URL Status hybrid gemini-2.5-pro claude-opus-4-1 o3 gemini-2.5-flash claude-sonnet-4 0 200 400 600 800 1000 Count of URLs Total Image URLs Generated hybrid claude-opus-4-1 gemini-2.5-flash gemini-2.5-pro claude-sonnet-4 o3 0.0 0.2 0.4 0.6 0.8 1.0 Proportion Image URL Status Status OK INACCESSIBLE_CONTENT INVALID_URL ERROR Figure 3 | Number of reference (top), and image (bottom), URLs and their status. Ablations We perform several ablation studies with the autorater (Table 4). Notably, removing the question-specific rubrics from the prompt changes the results only in the bottom half, with the Hybrid answers overtaken by Gemini 2.5 Flash and Claude Sonnet 4. This suggests that the additional resolution provided by the rubrics applies primarily to the kind of responses used to develop the rubrics. Or, in other words, that rubrics are far from complete. Hence, it is important that rubrics adapt to new data as better models become availab A New Expert-Grounded Benchmark for Scientific AI We introduce CLINB, a benchmark for modelbased evaluation of frontier models on complex, multimodal scientific communication. Its core is a new dataset of real-world climate questions paired with data-driven, question-specific evaluation rubrics, curated and validated by leading climate scientists through a novel three-phase, human-in-the-loop process.2 PhD-Level Synthesis vs. Attribution Failures Frontier models demonstrate remarkable knowledge synthesis, often exhibiting a PhD-level understanding. However, this performance masks a critical inadequacy in grounding. We report substantial hallucination rates for references (10% to 25%) and even more failures for images (50% to 80% in certain settings), exposing a major gap between synthesis and verifiable attribution. Insights into Human-AI Collaboration Dynamics Autonomous frontier models surpass ’hybrid’ answers (curated by experts using weaker AI assistance), revealing the assisting model’s capability—not human oversight—as the primary bottleneck. Counter-intuitively, highly motivated non-specialists (our ’Advocates’) who deeply engage with AI tools can produce higher-quality answers than domain experts who engage less with AI during answer curation. A Validated Methodology for Scalable Oversight We validate a rigorous, rubric-based autorater. Ablation studies demonstrate that structured prompts and automated evidence-checking are essential for mitigating inherent LLM judge biases. This process is hampered by inaccessible sources (up to 50%). Furthermore, we identify evaluation challenges, including model familiarity bias in human raters and the limitations of rubrics to generalize across models.
November 25, 2025 at 7:01 PM
Reposted by Dj Ripley
ketanjoshi.co
Deeply absurd. This Google PDF published on a blog (arxiv, not peer reviewed) claims an LLM is "PhD level" but in most cases the MAJORITY of reference URLs were invalid or inaccessible.

A PhD sitting down and just fabricating >50% of sources = career ending

arxiv.org/abs/2511.11597
CLINB: A Climate Intelligence Benchmark for Foundational Models Michelle Chen Huebscher1 , Katharine Mach2 , Aleksandar Stanić1 , Markus Leippold1,3, Ben Gaiarin1 , Zeke Hausfather4 , Elisa Rawat , Erich Fischer5 , Massimiliano Ciaramita1 , Joeri Rogelj6 , Christian Buck1 , Lierni Sestorain Saralegui1 and Reto Knutti5 1Google DeepMind, 2University of Miami, 3University of Zurich, 4Stripe, 5ETH Zurich, 6 Imperial College London Evaluating how Large Language Models (LLMs) handle complex, specialized knowledge remains a critical challenge. We address this through the lens of climate change by introducing CLINB, a benchmark that assesses models on open-ended, grounded, multimodal question answering tasks with clear requirements for knowledge quality and evidential support. CLINB relies on a dataset of real users’ questions and evaluation rubrics curated by leading climate scientists. We implement and validate a model-based evaluation process and evaluate several frontier models. Our findings reveal a critical dichotomy. Frontier models demonstrate remarkable knowledge synthesis capabilities, often exhibiting PhD-level understanding and presentation quality. They outperform “hybrid" answers curated by domain experts assisted by weaker models. However, this performance is countered by failures in grounding. The quality of evidence varies, with substantial hallucination rates for references and images. We argue that bridging this gap between knowledge synthesis and verifiable attribution is essential for the deployment of AI in scientific workflows and that reliable, interpretable benchmarks like CLINB are needed to progress towards building trustworthy AI systems. Total Reference URLs Generated claude-opus-4-1 claude-sonnet-4 gpt-5 hybrid gemini-2.5-pro gemini-2.5-flash o3 0.0 0.2 0.4 0.6 0.8 1.0 Proportion Reference URL Status hybrid gemini-2.5-pro claude-opus-4-1 o3 gemini-2.5-flash claude-sonnet-4 0 200 400 600 800 1000 Count of URLs Total Image URLs Generated hybrid claude-opus-4-1 gemini-2.5-flash gemini-2.5-pro claude-sonnet-4 o3 0.0 0.2 0.4 0.6 0.8 1.0 Proportion Image URL Status Status OK INACCESSIBLE_CONTENT INVALID_URL ERROR Figure 3 | Number of reference (top), and image (bottom), URLs and their status. Ablations We perform several ablation studies with the autorater (Table 4). Notably, removing the question-specific rubrics from the prompt changes the results only in the bottom half, with the Hybrid answers overtaken by Gemini 2.5 Flash and Claude Sonnet 4. This suggests that the additional resolution provided by the rubrics applies primarily to the kind of responses used to develop the rubrics. Or, in other words, that rubrics are far from complete. Hence, it is important that rubrics adapt to new data as better models become availab A New Expert-Grounded Benchmark for Scientific AI We introduce CLINB, a benchmark for modelbased evaluation of frontier models on complex, multimodal scientific communication. Its core is a new dataset of real-world climate questions paired with data-driven, question-specific evaluation rubrics, curated and validated by leading climate scientists through a novel three-phase, human-in-the-loop process.2 PhD-Level Synthesis vs. Attribution Failures Frontier models demonstrate remarkable knowledge synthesis, often exhibiting a PhD-level understanding. However, this performance masks a critical inadequacy in grounding. We report substantial hallucination rates for references (10% to 25%) and even more failures for images (50% to 80% in certain settings), exposing a major gap between synthesis and verifiable attribution. Insights into Human-AI Collaboration Dynamics Autonomous frontier models surpass ’hybrid’ answers (curated by experts using weaker AI assistance), revealing the assisting model’s capability—not human oversight—as the primary bottleneck. Counter-intuitively, highly motivated non-specialists (our ’Advocates’) who deeply engage with AI tools can produce higher-quality answers than domain experts who engage less with AI during answer curation. A Validated Methodology for Scalable Oversight We validate a rigorous, rubric-based autorater. Ablation studies demonstrate that structured prompts and automated evidence-checking are essential for mitigating inherent LLM judge biases. This process is hampered by inaccessible sources (up to 50%). Furthermore, we identify evaluation challenges, including model familiarity bias in human raters and the limitations of rubrics to generalize across models.
November 24, 2025 at 7:36 PM
Reposted by Dj Ripley
publichealthguy1.bsky.social
nirav shah not mincing words re: the new principle deputy director of CDC. good report from @apoorvanyt.bsky.social

🎁 www.nytimes.com/2025/11/25/h...
Dr. Nirav Shah, who served as the C.D.C!s principal deputy director for two years before resigning this year, called Dr. Abraham "unqualified" for the post and said when he learned that he had been appointed, "My jaw hit the ground." "A large part of the principal deputy's portfolio is emergency response," Dr. Shah said. "Delayed notifying of the public of at least two pertussis deaths is not just unacceptable, it's shameful," he added.
November 25, 2025 at 6:50 PM
Reposted by Dj Ripley
abeba.bsky.social
Large language mistake

Cutting-edge research shows language is not the same as intelligence. The entire AI bubble is built on ignoring it.

www.theverge.com/ai-artificia... great read from @benjaminjriley.bsky.social
Our cognition improves because of language — but it’s not created or defined by it. Take away our ability to speak, and we can still think, reason, form beliefs, fall in love, and move about the world; our range of what we can experience and think about remains vast. But take away language from a large language model, and you are left with literally nothing at all.
November 25, 2025 at 7:07 PM
Reposted by Dj Ripley
prisonculture.bsky.social
They also remembered he's not white, most likely.
ms.now MS NOW @ms.now · 4d
MS NOW EXCLUSIVE: “President Trump is considering removing Kash Patel as FBI director in the coming months, as he and his top aides have grown increasingly frustrated by the unflattering headlines Patel has recently generated.”
www.ms.now/news/trump-k...
President Trump may have had enough of Kash Patel, according to sources
The president is weighing whether to oust the FBI director after scrutiny of Patel's stewardship of bureau resources, including his girlfriend’s security detail and use of a government jet.
www.ms.now
November 25, 2025 at 7:18 PM
djripley.bsky.social
goddammit. antivaxxers are killing babies, so many babies.
November 25, 2025 at 7:26 PM
djripley.bsky.social
just have to reiterate that every bit of this is art: TORTILLA TERROR
( an awesome band name waiting to happen)
djripley.bsky.social Dj Ripley @djripley.bsky.social · 5d
I will never let go of this one that welcomed me back to London after a few years away
Headline in English tabloid: Super-hot Dorito Nearly Killed Me
November 25, 2025 at 7:22 PM
Reposted by Dj Ripley
vortexegg.com
It’s funny that a bunch of people who clearly don’t believe in and are openly hostile to both the methods and outcomes of science keep talking about how they are going to keep putting more money than has ever been spent into “accelerating science”. Sure, Jan.
November 25, 2025 at 3:33 AM
Reposted by Dj Ripley
annameier.bsky.social
Horrid news out of Oklahoma about Dr. Vahid Abedini, a professor of Iranian Studies.

The university has refused to make a public comment on Abedini's detention. Folks with OU connections, you know what to do: call OU and demand they take a stand & support Abedini.
November 25, 2025 at 3:16 AM
Reposted by Dj Ripley
angelamli.bsky.social
👉 Our new paper uses daily mobility data to show that spatial isolation is much more common today among those living in advantaged neighborhoods than the converse.

👩🏻‍💻 Lots of massive data wrangling and careful assumptions about mobility data needed - but check it out here! doi.org/10.1177/0042...
Abstract for "The truly isolated: Spatial isolation of advantage in the United States" by Shannon Rieger, Angela Li, and Patrick Sharkey, published at Urban Studies
November 24, 2025 at 5:20 PM
Reposted by Dj Ripley
dmgreene.bsky.social
This is remarkable work
angelamli.bsky.social Angela Li @angelamli.bsky.social · 5d
👉 Our new paper uses daily mobility data to show that spatial isolation is much more common today among those living in advantaged neighborhoods than the converse.

👩🏻‍💻 Lots of massive data wrangling and careful assumptions about mobility data needed - but check it out here! doi.org/10.1177/0042...
Abstract for "The truly isolated: Spatial isolation of advantage in the United States" by Shannon Rieger, Angela Li, and Patrick Sharkey, published at Urban Studies
November 25, 2025 at 3:42 AM