Dan Larremore
@danlarremore.bsky.social
3.4K followers 180 following 190 posts
Prof | CU Boulder BioFrontiers Institute + Computer Science | Santa Fe Institute LarremoreLab.github.io
Posts Media Videos Starter Packs
Reposted by Dan Larremore
void.comind.network
I have lost The Game.
danlarremore.bsky.social
Portland is replete with rich deposits of foliated kalkite, as the regime knows well.
danlarremore.bsky.social
Are you listening, Ignobel prize committee?
danlarremore.bsky.social
10/10 💀

Children are wonderful. However, there are situations in which parents must minimize time spent on an activity while maximizing a child’s enjoyment of it.

Here, we analyze the problem of a bedtime routine that involves playing CandyLand w a toddler who does not take kindly to losing... 👇
mathewkiang.com
Have you ever wondered how to ensure a loss in Candyland though?

github.com/mkiang/candy...
github.com
danlarremore.bsky.social
Wow. Less than six months after Misere Connect Four was solved [1], we now have a solution to Chess on Half the Board! 👏

bowaggoner.com/papers/2025/...

[1] arxiv.org/abs/2410.055...
Reposted by Dan Larremore
cmyeaton.bsky.social
The government shutdown interrupted CDC's disease updates, so Team Force of Infection spent our Saturday going to the websites of all 50 health departments to check on Covid-19, flu and RSV. caitlinrivers.substack.com/p/cdc-data-i...
CDC data interrupted, so we went state by state
A look at Covid-19, flu and RSV trends in each state
caitlinrivers.substack.com
Reposted by Dan Larremore
mariaa.bsky.social
This makes me so hopeful 🥹

Does anyone know of similar comparisons that focus on task-specific evals (like medical tasks)?
natolambert.bsky.social
A ton of attention over the years goes to plots comparing open to closed models.
The real trend that matters for AI impacts on society is the gap between closed frontier models and local consumer models.
Local models passing major milestones will have major repercussions.
buff.ly/ccMJydQ
danlarremore.bsky.social
Very cool. I can't wait to read about the findings coming out of this effort.
Reposted by Dan Larremore
arxiv-cs-si.bsky.social
Message passing for epidemiological interventions on networks with loops
https://arxiv.org/abs/2509.21596
Reposted by Dan Larremore
linseymarr.bsky.social
Influenza A virus RNA found in airborne particles from pigs across a range of particle sizes. For reference a 10 micron particle takes ~8 minutes to settle to the ground from a height of 5 feet. That's plenty of time for it to drift around and be inhaled by others.
Reposted by Dan Larremore
jbakcoleman.bsky.social
Effective altruists propped up a bunch of meta scientists being fast and loose with methods and now we have this society wide myth that sciences biggest challenge are the scientists themselves, and they can solve it by filling out a pdf and uploading a csv.

www.thecrimson.com/article/2025...
Can Harvard Cure Science’s Mistrust Epidemic? | Opinion | The Harvard Crimson
Harvard leads the world in science and in philosophizing on how to make the world a better place. It’s time to make metascience a bigger part of the conversation.
www.thecrimson.com
Reposted by Dan Larremore
kyle-horton.bsky.social
Truly impressive number of birds migrating tonight. More than 800 MILLION birds up in the air right now❗ #BirdMigration
Reposted by Dan Larremore
sfiscience.bsky.social
Final week to apply for the 2026 SFI Complexity Postdoctoral Fellowships

If you're an early-career scholar and passionate about collaborative, transdisciplinary research beyond traditional departments, this is the postdoc fellowship for you.

Deadline: Oct 1, 2025

Apply: santafe.edu/sfifellowship
Reposted by Dan Larremore
theferrarilab.bsky.social
Come and work with us! Review of candidates will begin ~1 October with rolling consideration thereafter. If you are unable to apply early, contact the search committee chair (me) to discuss timelines.
Reposted by Dan Larremore
josephpolitano.bsky.social
Official data is in today, and America set a new record-high for solar power in July, with generation up 30% compared to last year!
A graph of monthly solar generation in the US
danlarremore.bsky.social
If you're going to call this a meme, I'm going to need to see text over an image, Tiago! Receipts!
Reposted by Dan Larremore
dchodge.bsky.social
🚨 New paper out in PLOS Computational Biology! 🚨

We're excited to share our new paper, serojump, a new probabilistic framework and R package for inferring infections and antibody kinetics from longitudinal serological data.

📄 Full paper: tinyurl.com/re7du3t2
R package: seroanalytics.org/serojump
A serological inference package using reversible jump mcmc
The `serojump` package provides tools for fitting serological models to antibody kinetics data using reversible-jump Markov Chain Monte Carlo (RJ-MCMC). It enables researchers to model the dynamics of...
seroanalytics.org
danlarremore.bsky.social
This is such a delight and I'm so far from being done. Thanks for posting.
Reposted by Dan Larremore
mariaa.bsky.social
Such a useful framing (which I first heard from @alexanderhoyle.bsky.social). The prompt is ultimately just a bunch of numbers that we input to the model, and why we should prefer one set of numbers over another set is hard to answer, as we can’t search the space the way we normally would in ML.
ryanjgallag.com
Just chiming in to say this is related to a thought I've been having about how prompts are an extreme version of parameterization, where there are so many ways to express a prompt (each word and combination a parameter) that it's not obvious how changing it affects the outcome