Bitsy Perlman

A line curve showing number of awards for fiscal year 2026 compared to fiscal years 2021-2025 across NSF. The fiscal year 2026 curve lies well below curves for other fiscal years.

NSF Update

Funding curve overall. A little bit of progress in the past week, but only a little bit.

Now by Directorate...

1/11

Is anyone else bothered by the way mint flavor spreads? If e.g. any other flavor of chocolates are stored in a bag with mint chocolates, they all become mint chocolates.

I don't dislike mint chocolate; I just don't like it enough for all chocolate (or cupcakes or etc) to be all of that thing.

Our new research shows that hundreds of federal surveys removed questions about sexual orientation and gender identity since the start of the second Trump administration. Read the full report at: tinyurl.com/SOGIdata

America's Richest People Are Not its Most Generous

Fascinating chart with one outlier: Warren Buffett.

At the low end, giving 0.06% of one's wealth is equivalent to:

Net worth -> Lifetime Giving
50K->$30
100K->$60
500K->$300
$1M->$600

Most folks give far more by % in a *single year*.

www.forbes.com/sites/forbes...

I'd do freezing=0, boiling=200 if offered

I switched my phone to Celsius for a long time so I got used to temperatures in it too (at least, in the normal range for Boston), and I ended up frustrated that a degree of change in Celsius is too large, and felt one really needed to use half degrees to express things well

Having randomly looked this up, English Wikipedia says the original attempt was to scale the human body to 90° 🤷‍♂️

Of course, Fahrenheit was an attempt to use scientific principles as well, they were just a bit... messier however they have the advantage of being scaled around human experience

en.wikipedia.org/wiki/Fahrenh...

::taps the sign::

Evaluation of an AI model for the Automation of In-Field Enumeration Reinterview Clerical Matching in 2026 Census Test Kexin Zhang, U.S. Census Bureau Mike shaw, U.S. Census Bureau Kristine Roinestad, U.S. Census Bureau Robert Long, U.S. Census Bureau For the 2026 Census Test In-Field Enumeration Reinterview program, production interviews will be selected for reinterview. Some of these interview pairs will undergo a clerical matching of their household information. This process is critical for determining if the interviews were conducted at the correct household while following proper procedures. Traditionally, this process relies on human expertise and manual review, which is time-consuming, subjective, and prone to human error. This paper presents a methodology for evaluating the performance of a pre-trained Large Language Model (LLM), prompted to automate this clerical matching process. The evaluation framework focuses on a comparative analysis of the AI model’s classification outcome against that assigned by experienced subject matter experts. The evaluation process starts with a manual matching task performed by a panel of subject matter experts to determine a reliable ground truth classification. The consistency of their collective decisions is quantified using Fleiss' Kappa. A high Fleiss' Kappa value validates the quality of this human-generated gold standard, ensuring a dependable benchmark for model evaluation. The AI model's performance is rigorously assessed by comparing its predictions to this gold standard using inter-rater reliability, confusion matrix and other key metrics. The AI model demonstrated strong agreement with the clerks’ decisions, measured by the value of Cohen’s Kappa, with precision and recall exceeding [x] percent in a small test using household cases from the 2020 Census. Human-in-the-loop feedback can help AI model identify data patterns that mislead it...

NEW: The Census Bureau has been looking into using AI to determine whether follow-up, in-person census interviews are conducted at the correct households, according to the preliminary program for the Federal Committee on Statistical Methodology’s conference
static1.squarespace.com/static/5d0bb...

How AI slop is causing a crisis in computer science Preprint repositories and conference organizers are having to counter a tide of ‘AI slop’ submissions.

My latest @nature.com story on what computer science can do about the deluge of AI-generated/assisted papers.

Charge to submit? Use AI to review? Have AI-only outlets? Switch to a journal, rather than conference model? Reprimand repeat offenders? 🧪

More ideas?

www.nature.com/articles/d41...

This "AI will destroy all white collar jobs" stuff is stressful so let me (re)tell you a story.

Every month these homes are delayed, the number of houses should be doubled

But what if you find ways to build institutions so you can extract the surplus?

The pace at which US wealth concentration is rising is simply staggering

The concentration of AI wealth into the hands of a few tech barons + plutocratic capture ==> unchartered territory

Someone has sure already made this observation but the fact they can convert all those empty warehouses into prison camps means they could have converted them into housing, community centers, job training centers or, hell, libraries or schools all along. It’s always a matter of will not resources.

Your regular reminder that we (the people of Massachusetts) used to execute Quakers on the Common and celebrating Christmas was considered highly suspect

'Getting quick emergency care after a heart attack reduces your risk of dying by 30%!'

'Don't bother - emergency care is useless. The rate of heart attacks is so low. The absolute reduction in mortality is only 0.006%!'

When the first dinosaur bones were found, artists didn't really know how to render them realistically, so they do the artistic equivalent of stretching skin around them as tightly as possible, without considering where muscle and fat must have been.

Massachusetts has spent months delaying & denying @boltsmag.org’s public records requests on how many people it has transferred ICE during Trump as part of its formal contract with ICE.

A supervisor of records ruled last week that the state’s denials are improper; we hope to get records soon.

It is unacceptable that we're forcing more school budget cuts even as the Select Board weighs whether and how big of an override to present to voters. We should be incorporating these positions into the override ask and give voters an option to fund these positions.

Two pedestrians injured in separate Brookline car crashes on Thursday - Brookline.News One crash was outside the Driscoll School, and the other in Coolidge Corner.

I was looking into an incident of a pedestrian struck and injured while walking in a Brookline crosswalk yesterday, near Driscoll School. By the time I had enough information to publish a story, another pedestrian had been struck and injured in a different crosswalk.

brookline.news/two-pedestri...

"Moderna's CEO announced the company will no longer invest in new Phase 3 vaccine trials for infectious diseases: 'You cannot make a return on investment if you don't have access to the U.S. market. Vaccines for Epstein-Barr virus, herpes, and shingles have been shelved.” 9:12 AM Feb 12, 2026 604.8K Views

In terms of what cures are being lost:

- Epstein-Barr virus is perhaps the major trigger for multiple sclerosis
- herpes simplex virus causes cold sores, genital herpes, infections in babies, deadly meningitis
- shingles virus causes an intensely painful disease

The IRS slips on ICE The tax agency just shared thousands of unsuspecting people's private tax data with immigration enforcement.

On the newsletter -- How the IRS managed to spill thousands of the "wrong" people's confidential tax data to ICE.

Subscribe (for free!), if you will. Let's keep in touch.

We all know what's going to resist right wing oligarchs: middle to upper class people policing who enters their communities, and if the building near them are sufficiently attractive

Today's adventure in local journalism: Being called a "pawn for right wing oligarchs" because I wrote that zoning can be a factor that restricts development

PolyamoryResearchers groups.io Group This group is for people involved or interested in the academic research of polyamory. Its primary purposes are to: provide a forum in which to discuss research approaches and results; share academic...

Check out groups.io/g/PolyamoryR...

Happy birthday to one of my favourite haters, Charles Darwin

Person who was fired here - you should still trust BLS data. The agency is being run by the same dedicated career staff who were running it while I was awaiting confirmation from the Senate. And the staff have made it clear that they are blowing a loud whistle if there is interference.