For the last few months I've been doing something slightly obsessive: reading and scoring company engineering blogs, one by one, to build a filter for them. I'm past 800 now. Only around 157 made the cut.
Going in, I assumed I'd be rejecting blogs for bad writing. I was wrong almost every time. The writing is rarely the problem. Here's what actually separates the engineering blogs worth your time from the ones that quietly waste it.
1. The loudest blogs are rarely the best
Publishing frequency and quality are almost completely uncorrelated. Some of the highest-signal blogs I found post three or four times a year. Meanwhile, some of the most prolific "engineering blogs" turn out to be a content-marketing team wearing a Kubernetes lanyard — a steady stream of posts optimized to rank for a keyword, not to explain a decision.
If you're using cadence as a proxy for quality, you're optimizing for the wrong thing.
2. A blog's quality tracks a person, not a company
Over and over, a great engineering blog turns out to be one or two engineers who genuinely care about writing. When they change teams or leave, the blog doesn't get worse — it just stops. The quality was never institutional; it was personal.
The practical takeaway: follow bylines, not logos. A specific engineer with a track record is a far better signal than a company's brand.
3. The single most reliable quality signal: does the post admit a cost?
This is the one I'd keep if I could keep only one. Real engineering writeups have scar tissue. They tell you what was given up — the feature freeze during the migration, the approach they rolled back, the tradeoff they're still not sure about. Marketing-shaped content never bleeds: architecture in, architecture out, no friction, no regret.
"What did this cost them?" is a better filter than any scoring rubric I tried. If a post describes a big win with zero downside, it's either incomplete or it's an ad.
4. The launch post is the peak, not the start
A team writes their big "how we built X" post as a capstone — months of work compressed into one narrative, after the thing has stabilized. It does numbers. And then… silence, often for two quarters.
That silence says almost nothing about the team's engineering health and almost everything about who's assigned to write. Don't read a quiet blog as a declining team. Some of the best engineering happens at companies whose blogs are dormant for a year at a time.
5. Reputation and blog quality are different assets
The most surprising pattern: several companies with elite engineering reputations publish shockingly hollow blogs, and several companies you've never thought about publish consistently excellent ones. Brand prestige and writing quality are just… not the same thing. Which is exactly why "just follow the famous blogs" doesn't work.
The heuristic I ended up with
If you want a fast test for whether an engineering post is worth ten minutes, before you commit:
- Read for the tradeoff, not the headline. Does it name a real constraint and a real cost?
- Follow the byline, not the company.
- Ignore frequency and prestige as signals.
That's most of what a good filter does — and doing it by hand across 800 blogs is what convinced me it's worth automating.
That automation became a small Android app called Hexbrief — it applies these signals and surfaces six high-signal engineering reads a day, each with a quick problem → approach → result breakdown. Free, no account. But honestly, the heuristic above works whether or not you ever use it.
What's an underrated engineering blog you rate that nobody talks about? I'm always looking for ones that clear the bar — drop it in the comments.
Top comments (0)