This page describes how articles on phpwebthings.org get written, fact-checked, and updated. It’s deliberately concrete. If I claim a process here, I want to be held to it.
Who writes this site
I do. Edwin Rice, senior DevOps engineer, fifteen-plus years of self-hosting, day job in B2B logistics SaaS. There are no guest authors, no syndicated content, no AI-generated articles published as my work. If a draft was generated by an LLM, edited by me, and shipped, I would say so on the article. So far, none have been; that may change for very specific use cases (transcribing a long config file from a screenshot, for example) and if it does I’ll make it a paragraph at the bottom of the article in question.
What I review and why
Every piece of software I write a review or comparison of is something I have installed, configured, and run for at least one week, on hardware I own or rent. For most articles it’s been weeks or months. The hardware itself is described in the post (mini PC model, vCPU count, RAM, network speed, the works). When I cite a benchmark, I describe the methodology in the same article so you can reproduce it.
Versions and dates
Every review names the specific version of the software I tested and the month it was tested. Self-hosted projects move quickly. NextCloud Hub 8 in May 2026 is not the same product as NextCloud 27 in 2023, and I’m not going to pretend my tests on the latter cover the former. When a major release lands, I revisit the relevant articles and add an update note dated to the day, rather than silently rewriting the original text.
Sources
For technical claims I prefer (in order):
- The project’s own documentation or release notes (linked).
- The project’s source code or commit history when documentation is silent.
- Maintainer or core-contributor statements (forum posts, mailing-list threads, GitHub issues).
- Reproducible third-party benchmarks (LWN.net, ServeTheHome, Phoronix where they touch the topic).
- My own benchmarks, with methodology disclosed.
I don’t quote vendors’ marketing copy as fact. If a project says “scales to thousands of users” without a benchmark behind it, I either find the benchmark or I describe the claim as a claim.
Corrections
If I publish something wrong I correct it. The article will get a “Corrected on YYYY-MM-DD” note at the bottom describing what was wrong and what it was changed to. I do not silently edit articles to remove embarrassments. If a correction materially changes the conclusion of a comparison, I’ll add a note to the top of the article as well.
Email me at edwinrice1977@gmail.com if you spot an error.
Conflicts of interest
I disclose:
- Any affiliate relationship with a vendor I review (see affiliate disclosure).
- Any commercial relationship between my employer and a vendor I review (none at the time of writing).
- Any free or discounted hardware or service I received for review purposes (the cost-of-ownership math has to still make sense for a reader who pays full price; I don’t write favorable reviews of expensive hardware just because I got a sample).
What this site is not
- Not a news site. I don’t break stories; I write longer-form analysis after the dust has settled a little.
- Not a sponsored-content marketplace. I don’t run “sponsored posts” or “branded write-ups” of any kind.
- Not a review-of-everything site. I write about software I actually run. There are projects I’d love to cover that I haven’t gotten to yet, and that’s the only reason they’re absent.
A note on AI content
This is its own paragraph because the question is the right one to ask in 2026. The content on this site is written by me, on my keyboard, drawing on my own experience running these tools. I use search and standard developer reference material the way every working engineer does. I do not run an article through an LLM and ship it as mine. Where I think LLM assistance might genuinely help readers (regenerating a long ASCII diagram cleanly, for instance), I’ll say so in the article and credit the model used.