The Freethinking Times has been running on Project Broadsheet for about a year. I built the framework specifically to power it, then released the framework because people kept asking me how it worked. This post is my retrospective: what I learned, what surprised me, and what I'd do differently.

1. Readers care about reader tools more than I expected

Before I put the site live, I thought the reader-side features were a "nice to have" that most people would ignore. I was wrong. The reading panel is the second-most-opened thing on the site after the articles themselves. The most-used options, in order: choosing a font, dark mode, the reading ruler, and save-for-later.

Three separate readers have emailed me about the reading ruler specifically. They have dyslexia or attention challenges, and a ruler that follows the cursor is genuinely useful for them. I almost didn't ship it.

Takeaway: readers with accessibility needs exist and are underserved. If your platform makes it trivial to add these tools, you should.

2. RSS still works, and the people who use it are the best readers

The Freethinking Times has about 400 RSS subscribers. They're disproportionately long-term readers, disproportionately the ones who reply to emails, and disproportionately the ones who support the publication financially.

RSS isn't popular. It is high-quality. Whatever effort you put into per-section feeds, author feeds, and a properly-formatted main feed pays back in reader loyalty.

3. Substack's discovery is real, and leaving it cost me readers

I moved off Substack before I shipped The Freethinking Times. Substack's recommendation network brings a steady trickle of new readers that I lost by leaving. About six months in, I could feel the difference.

The tradeoff is that every subscriber I have is mine. I control the list, I can migrate, I don't pay 10% of revenue. But I'd tell anyone starting out: Substack's network is real value, and leaving it early means slower early growth.

If you're starting a new publication with no audience, Substack + a planned migration to self-hosted at 1,000 subscribers is a reasonable strategy.

4. The editorial section structure matters more than the visual design

I spent weeks on typography and color. None of it mattered nearly as much as getting the section structure right. News, Opinion, Analysis, Reviews, Science & Tech: that taxonomy shaped what I write and what readers expect.

When I added Reviews as a first-class content type (with star ratings and structured metadata), traffic to review articles went up 3x. Readers want to browse reviews as reviews, not as articles with the word "review" in the title.

Spend the time on taxonomy. Design follows.

5. Writing for a publication is different from writing a blog

Project Broadsheet encourages you to think of yourself as running a publication, not a blog. That framing changed how I wrote. I started running features, not posts. I thought about issues, not articles. I published an Edition each quarter that grouped related pieces.

The work is harder. The readership is more engaged. I don't know if that's universal, but it was true for me.

What I'd tell you if you're starting

Don't spend months perfecting the setup. Publish. Edit later.

Use the default sections. You can rename them if they don't fit, but don't invent a new taxonomy from scratch on day one.

Write for RSS subscribers as if they were the most important readers. They are.

And honestly: back up your email list. Mine is safe in Buttondown; yours should be somewhere you control too.