Work posts

Deduplicating 40k SNAP news articles with MinHash LSH

I needed to identify and tag duplicate news articles from a Google Alerts RSS feed which filters the news through keywords while working on topic modeling of news coverage on the Supplemental Nutrition Assistance Program since summer 2024. I also want to in the future group these on the surfaced SNAP News site I put together for these articles. After reading it for around a year, I consider duplicates / near duplicates to be an overall rare, but noticeable phenomenon. This post quantifies that feeling, explores why it occurs, and discusses some implications for topic modeling. ...

February 10, 2026 · 23 min

Visualizing SNAP Performance Indicators

A slow pace, to be sure, but I have been adding to a dashboard of SNAP Performance Indicators over the past year, based on the SNAP data I’ve been cleaning and compiling over the past 2 years, and built with the Observable Framework library. You can view the dashboard here. No extremely tricky visualizations, but that is because I have found that these datasets that are simple tables by state each year mask the complexity of federal and state policy around implementing SNAP benefits. The dashboards become more and more text heavy, I realize, as I absentmindedly pull in context that I believe is valuable to know alongside each chart. Additionally, the recent changes to SNAP in 2025 have complicated any forward-looking projections. ...

December 22, 2025 · 5 min

Quickly visualizing some low-effort content on YouTube

Recently on YouTube I searched “population apportionment” to find that many of the recommended videos across different channels looked the same and appeared to be AI-generated, heavily automated, and extremely low effort. Of the 18 Recently Uploaded videos for the search, 10 of theme were all from this same “family” of channels: 2 to 3 minute, political-themed questions with an AI-generated man or women in the thumbnail, leading to an animated PowerPoint presentation. ...

December 17, 2025 · 4 min

How political are shutdown alerts?

If you visited a .gov website during the 2025 government shutdown, you may have noticed alert banners on the top of the page that indicate a suspension or interruption of normal operations for the agency. Some agencies vary from the standard, however, and instead use this opportunity to deliver a political message. From a cursory analysis, of the 248 shutdown banners I could find from about 1,150 agency websites, 92 were political to some extent (37%).1 ...

November 17, 2025 · 7 min

Well, let's just not measure food insecurity then

In September 2025, the Trump administration said it will no longer release the Household Food Security Reports. The USDA also put relevant employees on indefinite leave. This post provides context to the published press release based on the USDA’s own 2025 Information Collection Request renewal for the Food Security Supplement to the Current Population Survey, hosted on Reginfo.gov.1 These redundant, costly, politicized, and extraneous studies do nothing more than fear monger. Is the survey redundant? ...

November 3, 2025 · 15 min

Debugging public data - county codes in Texas

If you sort a list of Texas counties by county name, the counties will not be in order by county number. Texas Center for Healthcare Statistics Every county in the United States has a FIPS code. These standards codes are usful for mapping and joining additional datasets. Texas state agencies do not bother with these FIPS codes. Instead, each state agency publishes data using their own county codes. The county codes of one state agency may not match other agencies’ county codes. I recently spent a few hours extremely confused by this, because I used the formula published by the Texas Department of State Health Services on data published by the Texas Health and Human Services, resulting in incorrect data. ...

May 16, 2025 · 9 min

The Trump Admin's data visualizations need some work

The Trump Administration has a famously loose approach to data visualizations. If the data doesn’t match up with what you want, grab a marker and draw over the visualization until it does. Source: New York Times article showing hand-drawn modifications to official NOAA weather map The USDA released an announcement, USDA Invests Up To $1 Billion to Combat Avian Flu and Reduce Egg Prices, with two charts that made some questionable decisions worth discussing. ...

March 2, 2025 · 9 min

Questions on “How Congress Can Fix Food Stamps”

The National Review in July published a short opinion piece in July ‘24 entitled How Congress Can Fix Food Stamps. Sam Adolphsen, the author, argues that the cost of SNAP benefits is too high. He comments that “about 17.6 million food-stamp recipients are able-bodied adults, and about two-thirds don’t work,” which leads him to list a series of Republican-led, state-level policies that require SNAP recipients to participate in job training or employment programs. These policy decisions, Adolphsen argues throughout the article, are “common-sense”, “wildly popular among voters”, “turning a handout into the hand up that [SNAP benefits were] always supposed to be,” and are “protecting taxpayers while promoting work over welfare”. ...

September 22, 2024 · 16 min

Lessons learned on dashboard design for early-stage apps

What follows are a few takeaways from a recent app launch, where I had some successes and some things I had to fix. I also maintain dashboards for a wide set of clients on a different project that will be overhauled soon, and would like to apply these learnings during the redesign. The three guiding business questions for initial design are: How are we getting users? What our users are finding value in when using the application? ...

September 7, 2024 · 17 min

Daily SNAP News Aggregator launched

Site is accessible here, and the accompanying GitHub repository can be accessed through that site as well. It is based on a Google Alerts RSS feed, built every ~6 hours with Jekyll and GitHub Actions. Monthly commentary will be included on that site in the future. Immediate next steps are to add in more news aggregators as sources, to try and get articles that are missed. I would like to integrate Google Scholar as well, but it does not appear they offer an RSS feed or any simple method. Article display filtering is currently done on a keyword basis, which has been working fine, but would like to move to a classifier model. ...

June 11, 2024 · 1 min