#data-matching
Happy 13th anniversary to this pair of matching news headlines.
February 10, 2026 at 7:26 AM
To me, this feels like an extension of DOGE. Big claims about fraud, using data-matching to drop people from coverage. Without actual data, or investigation, it is impossible to refute the claims, but such transparency is unlikely to be forthcoming.
It’s surreal to see federal officials take a victory lap today for cancelling hundreds of thousands of marketplace enrollments, while massive health coverage losses are underway.

www.nytimes.com/2026/09/22/h...
Trump Officials Eject 750,000 From Obamacare Markets, Claiming Fraud
Vice President JD Vance said the measure would save the government $2.2 billion.
www.nytimes.com
September 22, 2026 at 7:31 PM
Same-region player matching will soon be available!

Players will be able to match with others on the same data center. This also includes the Party Finder! 🎉
April 24, 2026 at 6:26 PM
“Other recent papers have similarly suggested that LLMs don't actually perform formal reasoning and instead mimic it with probabilistic pattern-matching of the closest similar data seen in their vast training sets.”

It’s not a suggestion! They are *not* doing reasoning, just pattern matching.
Apple study exposes deep cracks in LLMs’ “reasoning” capabilities

“Irrelevant red herrings lead to "catastrophic" failure of logical inference”

- Large Language Models (LLMs) are the computing technique behind ChatGPT, etc.

ArsTechnica story by Kyle Orland: arstechnica.com/ai/2024/10/l...
Apple study exposes deep cracks in LLMs’ “reasoning” capabilities
Irrelevant red herrings lead to “catastrophic” failure of logical inference.
arstechnica.com
October 15, 2024 at 6:08 AM
Right now, people are being kicked of Obamacare subsidies based on mystery data matching techniques with mailed follow-ups for verification purposes, and this seems very much like the kind of set up that will lead to a high error rate bsky.app/profile/donm...
Two points here.
1. Automation can be great, but this administration is making it harder for states to use automation to expand coverage.
2. Automation to detect fraud has a really bad track record in practice.
donmoynihan.substack.com/p/robodebt-w...
Robodebt: When automation fails
Lessons from Australia's use of algorithms to detect welfare fraud become more relevant in the age of AI
donmoynihan.substack.com
September 24, 2026 at 6:52 PM
Starting Tuesday, October 13, same-region player matching will be available for the Japan data center.

Read on for details on the upcoming changes to matchmaking, party recruitment, and more! 🌐 sqex.to/d3n0A
September 15, 2026 at 11:00 AM
The top priority of the next functional government has to be reforming the Privacy Act to prevent the use of personal data for reasons other than originally prescribed & to effectively ban database matching all but entirely.
CNN reports that the oligarch is building a "master database to speed-up immigration enforcement and deportations by combining sensitive data from across the federal government" with the help of Palantir, to create "targeting lists".

That all sounds very very bad.
edition.cnn.com/2025/04/25/p...
Elon Musk’s DOGE team is building a master database for immigration enforcement, sources say | CNN Politics
Staffers from Elon Musk’s Department of Government Efficiency are building a master database to speed-up immigration enforcement and deportations, by combining sensitive data from across the federal g...
edition.cnn.com
April 26, 2025 at 8:20 PM
The math isn't matching. Data doesn't lie. Checkout Election Truth Alliance. @electiontruth.bsky.social
May 20, 2025 at 11:47 AM
New blog up: solomonkurz.netlify.app/blog/2025-02...

This time I dip my toes into causal inference for quasi-experiments using matching methods, and my use case has missing data complications. Many thanks to @dingdingpeng.the100.ci and
@noahgreifer.bsky.social
for their peer review! #RStats
Matching, missing data, a quasi-experiment, and causal inference--Oh my! | A. Solomon Kurz
I'm finally dipping my does into causal inference for quasi-experiments, and my first use case has missing data. In this post we practice propensity score matching with multiply-imputed data sets, and...
solomonkurz.netlify.app
February 4, 2025 at 4:36 PM
Currently (left) each region has several data centers that host worlds, and matching is restricted to each data center. Soon that will no longer be an issue. This includes Party Finder.
April 24, 2026 at 6:20 PM
The unsigned decision overrules lower court decisions that blocked the data-matching across government agencies.
Supreme Court Lets Trump Admin Use Social Security Info To Screen For Noncitizen Voters
The unsigned decision overrules lower court decisions that blocked the data-matching across government agencies.
www.huffpost.com
September 25, 2026 at 3:55 PM
cross data center player matching, everything is fixed
April 24, 2026 at 6:19 PM
Microsoft said its recent layoffs impacting 6,000 employees (3% of staff) was due to flattening hierarchies by removing layers of managers. However data provided to the state of Washington shows most of those impacted were individual contributors.

The company’s words and actions aren’t matching up.
May 18, 2025 at 5:58 PM
1/ Imagine walking into your town meeting.

Twenty people are wearing matching shirts supporting the new data center.

You think: “Maybe I’m the minority.”

Now imagine discovering a $50 MILLION influence campaign is helping manufacture that support.

#DV1 #ProudBlue #Voices4Victory #USDemocracy
September 5, 2026 at 8:44 PM
Algorithms to serve up best-matching URLs is a Grand Canyon of difference from algorithms making guesses in summaries from its data gathering.
May 1, 2025 at 7:28 PM
CROSS DATA CENTER PARTY MATCHING OOOOH MY GOD YES
April 24, 2026 at 6:18 PM
Old maps with matching coordinate data yesss
Chicago map lovers (you know who you are and there are too many of you to tag), look what's just been released!
node.uchicago.edu/collection/m...
UChicago Node
University of Chicago Node
node.uchicago.edu
May 16, 2026 at 12:30 PM
The team says players have been forced to use the data center travel system to find others who want to complete raid objectives and they apologize for it. This update will allow for easy matching without having to switch data centers.

Coming in Patch 7.5x.
April 24, 2026 at 6:22 PM
Includes phone numbers and participation in government programs such as Medicaid.
A potential end goal here is data matching across different systems to chase after fraud. Sounds good on paper, but the risk of false positives is very high (see election administration efforts in this area).
New from 404 Media: a Musk ally has demanded administrative access to Notify.gov, a system that lets the government text the public. The access would give them large swathes of the publics' personal data too. A worker has resigned in protest rather than give access www.404media.co/musk-ally-de...
February 18, 2025 at 7:58 PM
What LLMs do is pattern matching. They can recognize patterns from their training data and create similar content or analysis.

This overlaps white collar jobs because they are often mostly pattern matching (e.g. writing marketing copy, editing articles, etc) but this isn’t general intelligence.
June 22, 2025 at 3:42 PM
Also, what is the database of illegal immigrant faces they are planning on matching camera data to???
October 5, 2025 at 10:26 AM
"quant data is fucking useless without a serious body of theory to inform the construction of measures and the interpretation of the resulting data" - matching tattoos at MPSA? :P
August 18, 2024 at 2:49 PM
my favourite part about my job working with financial data is that the data is literally never clean and that there are constant mistakes, so for the past week i have been chasing a ghost about why numbers that should match aren't matching and it's because for no clear reason a process was changed.
March 25, 2025 at 9:44 PM
learning about different string matching methodologies today - Levenshtein, q-gram, cosine, Jaro...

data people who wrangle text data for a living are truly the unsung heroes of our data community.
September 18, 2025 at 5:17 PM