I used to start every morning the same way: open six tabs. Business of Fashion. Dezeen. Highsnobiety. Fast Company. Hypebeast. Kottke. Scan headlines. Close tabs. Repeat tomorrow.
It was fine. It was also slow, scattered, and I never knew if I was missing something. The stories that mattered - the ones where fashion, tech, brands, and design were converging - got buried in the noise of everything else.
I wanted a Techmeme for culture. One place where the most important cultural stories rise to the top, scored by relevance, not by clicks. So I built CultureTerminal.
The CULT algorithm
Every article gets scored 0-100 across five dimensions. I called it the CULT score, partly because it's memorable and partly because the best cultural moments feel a bit like cults - intense, niche, magnetic.
Relevance (0-25) - how many cultural categories does this story touch? An article about Jony Ive designing a Ferrari interior hits design + tech + brand strategy = higher score. Single-category stories score lower. This is the key insight: the most interesting cultural stories live at intersections.
Freshness (0-25) - time decay. Less than 6 hours old gets full marks. Decays over 72 hours but never hits zero. Culture moves fast, but yesterday's insight isn't worthless.
Authority (0-20) - source tier. Business of Fashion and Stratechery are Tier 1. Highsnobiety and Dezeen are Tier 2. Niche voices are Tier 3. The gap is intentional - editorial weight matters, but small sources can still break through with the right story.
Brand (0-15) - does this story mention major brands? Not because brands are inherently important, but because brand stories signal commercial forces intersecting with culture. An article about "luxury fashion trends" scores lower than one about "LVMH's strategy shift."
Depth (0-15) - content length plus analytical keywords. Rewards long-form analysis over headlines and briefs. "Investigation," "deep dive," "strategy" all boost the score.
27 sources, curated ruthlessly
This is where the taste comes in. CultureTerminal doesn't pull from everywhere. It pulls from 27 specific sources that I've hand-selected across fashion, design, technology, advertising, art, and lifestyle.
The selection criteria isn't "who gets the most traffic." It's "who consistently publishes the stories I'd want to read if I only had 15 minutes." Business of Fashion for fashion-business intersection. Stratechery for tech-culture analysis. Kottke for pure internet curation. Dezeen for design. It's a reading list disguised as an algorithm.
I've tested and rejected dozens of sources. Campaign and The Drum? Too industry-inside. HBR? Too generic. The curation of inputs is as important as the scoring of outputs.
Eight cultural verticals
Every article auto-classifies into one of eight categories: Fashion & Style, Brand & Business, Culture & Ideas, Music & Entertainment, Design & Architecture, Art & Photography, Lifestyle & Taste, Tech & Digital.
The categorisation happens through keyword matching - 40-70 keywords per category, tuned over months. It's not perfect, but it's fast and consistent. And the imperfections are actually interesting: when an article hits multiple categories, that's usually a signal that it's a more important story.
Why it's not just an RSS reader
The obvious question: "Isn't this just an RSS reader with extra steps?"
No. And the difference matters.
An RSS reader shows you everything from your subscribed sources, in chronological order. CultureTerminal shows you the best stories from curated sources, ranked by cultural relevance. The experience is fundamentally different. You're not managing a feed; you're checking a radar.
RSS readers are consumption tools: "I want to read everything from X source." CultureTerminal is a signal tool: "What should I pay attention to this week?"
The automated pipeline
CultureTerminal runs on autopilot. Every morning at 6am UTC, a GitHub Actions workflow fires: fetch all 27 feeds, score every article, generate the day's ranking, deploy to Netlify. By the time I wake up, today's cultural intelligence is live.
800+ articles get scored weekly. The top stories surface. The noise drops away. I get a push notification when it's done, and I open the site over hot chocolate to see what's happening in culture today.
That automation was important to me. If CultureTerminal required daily manual effort, I'd have abandoned it within a week. The whole point is that it's a machine that runs on taste - my taste in sources, codified into an algorithm.
What it taught me
CultureTerminal taught me that algorithms and taste aren't opposites. The CULT score is an algorithm. But every decision inside it - which sources to include, how to weight authority vs freshness, what counts as "depth" - is a taste decision. The code executes my judgment at scale.
It also taught me that the most interesting stories live at intersections. Single-category news is fine. Multi-category news is where culture actually moves. That's why the Relevance dimension rewards cross-category stories. The best cultural moments are the ones that touch everything at once.
800 articles a week. 27 sources. One score. That's CultureTerminal.
