Posts

Disclosure: This analysis was researched and written with the assistance of Claude Code, Anthropic’s AI coding assistant. All candidate information was sourced from public reporting, campaign websites, editorial board interviews, and polling data. The persuasion scoring reflects the author’s editorial judgment applied through an AI-assisted workflow. Voters should verify claims independently before casting a ballot. I just finished scoring every candidate in every contested race on both the Republican and Democratic primary ballots for Dallas County. Same methodology for both: two independent 1–10 scales measuring how much each candidate relies on facts/data versus feelings/emotions. Now that both ballots are done, patterns jump out. The two primaries are operating in different universes — different emotional registers, different power dynamics, different relationships with money. Here’s what I found. ...

Disclosure: This analysis was researched and written with the assistance of Claude Code, Anthropic’s AI coding assistant. All candidate information was sourced from public reporting, campaign websites, editorial board interviews, and polling data. The persuasion scoring reflects the author’s editorial judgment applied through an AI-assisted workflow. Voters should verify claims independently before casting a ballot. The March 3, 2026 Texas Democratic Primary ballot for Dallas County is enormous — 31 contested races spanning federal, state, and local offices. I went through my sample ballot (Precinct 12), researched every candidate in every contested race, and scored them on how much they rely on facts and data versus feelings and emotions to make their case. Same methodology as the Republican guide: two independent 1–10 scales. A candidate can score high on both, low on both, or anywhere in between. Judicial races tend to cluster low on emotion, which is appropriate — you probably don’t want a judge who campaigns on rage. One important context note: Democrats haven’t won statewide in Texas since 1994. The primary winner in most Dallas County local races is effectively the general election winner, but every statewide nominee faces long odds in November. ...

Disclosure: This analysis was researched and written with the assistance of Claude Code, Anthropic’s AI coding assistant. All candidate information was sourced from public reporting, campaign websites, editorial board interviews, and polling data. The persuasion scoring reflects the author’s editorial judgment applied through an AI-assisted workflow. Voters should verify claims independently before casting a ballot. The March 3, 2026 Texas Republican Primary ballot for Dallas County is packed — 12 contested races with 47 candidates total. I went through my sample ballot (Precinct 166), researched every candidate in every contested race, and scored them on how much they rely on facts and data versus feelings and emotions to make their case. The methodology is simple: two independent 1–10 scales. A candidate can score high on both (a skilled communicator who uses data and stirs emotion) or low on both (a quiet campaign with little public presence). The scores come from reviewing campaign websites, public statements, news coverage, editorial board interviews, and advertising. ...

This is going to be short but I wanted to get this down because I did this once before and everything worked out great but then when I tried it again with another lambda project, I had a lot of problems getting credentials setup. First of all, DO NOT go hard coding anything in the lambda. Definitely, DO NOT hard code credential key and secret but do not hard code the region either, unless there is a very specific need for this like you’re running the lambda in one region but say accessing an s3 bucket that is setup in another region. ...

Okay, they’re not totally useless but they might as well be. I opted to use a shell script in a mise task to manually create the file with the information I needed. Specifically, what I figured was an obvious thing was to run the command something like, hugo new content/posts/1767247204-sample.md --tile "Sample Post" --date "2026-01-01 00:00:04", otherise Hugo will insert a titlized version of the filename and the current date for the new post. This probably seems like niche issue but it’s important to me and it’s such an obvious option if you needed to back post something or perhaps you wrote it out on paper on one day and wanted it posted as that day not the current day the markdown was created. ...

Assume we have monorepo at git@github.com/haydenk/monorepo.git ...

Absolutely nothing against Hugo, in fact I love Hugo as a blog framework. I found myself focused too much on having the right look and meddling with themes that I was not producing any posts. This one will be short, I am not a wordy person. I just wanted to get my thoughts down while they were on my mind. I will return to Hugo at some point or maybe something better than Hugo will have come around by the time I am ready to return. In the meantime, I will focus on creating posts within this repository.

When I started, I struggled with what I was going to use. I wanted to use Rails but I couldn’t find much information on how to use Rails with an embedded app in Shopify. That has since been debunked. Before it was debunked, I settled on using Remix because it was in the Shopify developer documentation and made it easy to get started. Really, this turned out to be a huge pain in the ass both in development and with deployment. ...

UPDATE 2025-02-27 Replaced asdf for mise. I had some oddities pop up on the latest asdf update that I didn’t have with mise and mise seems to work nicer. Sometimes I like to wipe my Macbook and start over but I forget some little things that make starting over frustrating. I am not going to give full detail instructions on how I install it. I usually just follow the steps from the project site. I will only include nuances that I may do different or in addition to the normal steps. ...