Phil Eaton

In all cases, and this is something I'm learning, I am most looking for submissions that talk about existing projects, not focused on things you (the author) built. If you wrote code during the course of the article that's fine, but the focus should not be on your own work.

If you're a CS grad student in the US, I'll pay you a small fee to write for The Consensus. I'm especially looking for articles that compare the state of research to the state of what devs do in practice, because there are often interesting discrepancies.

theconsensus.dev/contribute.h...

Very happy to have the first contributed article published in The Consensus. And very happy to turn Alperen loose on you all to talk about type systems and gradual typing.

It was shown but I agree it could be clearer. Perhaps display it as a table would be better.

Thank you for the feedback and kind words!

It has not been easy to decide what goes into each tier. But for the moment the Student tier gets you access to paywalled articles, Standard tier gets access to a full content RSS feed and the monthly newsletter.

If you have signed up for the Standard tier after the newsletter was sent, you can find the newsletter archive here to read it online.

theconsensus.dev/n/2026/02/st...

You can subscribe here

theconsensus.dev/subscribe.html

The Consensus launched in February, payments were only in place two weeks ago; I was prepared to send out this edition to only myself as practice. But in the last two weeks nearly 80 individuals subscribed at the Standard tier. The support and interest has been astounding. 🙇

The very first of The Consensus Standard just went out. This is a monthly newsletter for folks subscribing at the Standard Subscription tier.

A collection of 1) articles written for The Consensus, 2) interesting jobs, 3) funding announcements, and 4) external articles I enjoyed.

New batch of jobs dropped. These job postings are not sponsored (if they ever are, they will be labeled). They're simply interesting-looking opportunities related to software infrastructure, pulled from around the internet.

theconsensus.dev/jobs.html

AI writing is like store-bought cake. It might be perfectly fine, maybe even as good as something you could make yourself, but it’s weird to give it to someone and say it’s homemade

Just spent an hour today in our agentic AI coding class showing all the way Claude Code produces *subtly* bad programs for a pretty trivial application. Wrong in some ways, but just bad in other ways. Reading its output is always a sobering experience.

Performance Engineer, Backend Automattic’s products—WordPress.com, WooCommerce, Beeper, Tumblr, Jetpack, and more—serve tens of billions of page views every month from data centers around the world. The PerfOps team exists to make...

The performance team I lead at Automattic is hiring - automattic.com/work-with-us... #webperf #php #jobs #wordpress #woocommerce

Thanks for clarifying that!

While yes those projects exist, no they did not account for any of these major projects I surveyed. :)

Some of the AI-assisted commits I found made me chuckle: a three-line CSS change by Claude in Mattermost, or a change by Claude "reviewed by NOBODY (OOPS)" in an Apple engineer's commit to WebKit (granted it's a comment-only change).

That is very generous, thank you.

We - like many other technical teams - are hungry for the market for trusted, neutral, technical advice, but it's hard to find. The big analyst firms aren't sufficiently forward-thinking, and their publications are rarely detailed enough, geared towards managers rather than practitioners.

No I am not publishing it as an indictment. The 40 or so projects that didn't use AI explicitly I suspect many of them are using AI without mentioning it.

theconsensus.dev/p/2026/03/02...

I surveyed 112 major source-available projects to understand their AI contribution policy and whether or not they have actually accepted explicitly-labeled AI contributions.

Only 4 projects banned AI completely: Zig, NetBSD, GIMP, and qemu. 70 already have AI-assisted commits.

The recording finally went great this time. I also demo'd doing a backup audio recording so that we can more reliably get a good recording to post, so hopefully the trend will continue 🤞

We have pgvector at home

(paywall has expired, article is now available for all to read)

theconsensus.dev/p/2026/02/22...

theconsensus.dev/p/2026/02/26...

I spoke with Kosta Tarasov who contributes to DataFusion and arrow-rs in his free time. This article is not paywalled, read it now.

A big goal of The Consensus is to highlight and celebrate the work of open source contributors, both newcomers and long-time contributors.

I started a software research company

notes.eatonphil.com/2026-02-25-i...

Also, a little announcement, without a subscription you're going to see paywalls on new articles for the first week after publication.

This work depends on your support. You can subscribe at a discount with FIRST100. DM if you have any issues! <3

theconsensus.dev/p/2026/02/22...

You're probably right to pick a modern extension to support vector similarity search in Postgres. But did you know Postgres already has one built in?

I took a look at the cube extension in Postgres, pgvector, and model2vec for some impressively fast embeddings generation.

what are some examples of technical books/booklike things that do nontraditional things structurally ala The Little Schemer? not necessarily the Q&A format specifically, but not being structured like a normal book