7 Things Worth Your Attention This Week

Swiss type designer Fabian Harb built a variable font whose letterforms visibly swell and contract based on real-time humidity data from your location. Whether this is useful is debatable. Whether it's an interesting proof of concept about how environmental data could be woven into everyday interfaces is not. The GitHub repo has been starred 4,000 times in two weeks, which is either validation or a sign that developers will star anything that involves APIs and fonts.

TabMap is a browser extension that visualizes your open tabs as a node graph, showing which tabs you opened from which other tabs. It produces a surprisingly accurate portrait of how you actually think versus how you think you think. The creator built it to understand their own research patterns and published it partly as a productivity tool and partly as what they describe as "a way to make your cognitive mess visible." The mess, it turns out, has structure.

The subreddit dedicated to mechanical watchmaking has 340,000 members and a culture that's earnestly, unfashionably enthusiastic about its subject. No irony, no discourse, just people taking watches apart and explaining what they found. The comment threads on movement restoration posts regularly hit 500 replies. In a landscape where most hobbyist communities have been colonized by acquisition advice and brand discourse, this community's focus on making things is striking enough to be worth pointing out.

Marginalia is a search engine that deliberately favors non-commercial web content — personal sites, wikis, hobby pages, and amateur writing over SEO-optimized results. It indexes a small fraction of the web and is therefore genuinely not useful for most searches. For finding non-commercial perspectives on niche topics, it remains remarkable. The developer runs it alone on a server in their apartment and refuses to monetize it. Six years running. Worth bookmarking as a second search engine for when Google's first page is entirely useless.

The Rebooting is a media industry newsletter written by Brian Morrissey, and it covers the economics and culture of media without the reflexive self-importance that afflicts most media writing about media. This week's issue on why newsletters that started as passion projects fail when they try to become businesses contains a specific and useful analysis of the point at which audience-building logic corrupts editorial judgment. If you work in publishing or make anything online, it's worth the fifteen minutes.

Felienne Hermans published research in 2013 arguing that spreadsheets are a programming language, and that the people who build complex Excel workbooks are programmers who don't identify as such. Twelve years later, someone built a working game of Tetris inside Google Sheets using only cell formulas and conditional formatting — no scripting. The video demonstrating it has been circulating this week and it's a better argument for Hermans' thesis than most of the academic writing on the subject.

The Wikipedia article for "List of common misconceptions" has been edited 14,237 times since 2004, making it one of the most contested pages on the site. The talk page is longer than most academic papers on epistemology. What counts as a misconception, what counts as evidence that it's common, and what counts as a reliable source for its debunking are all genuinely difficult questions that the volunteer editors have been arguing about for two decades. It's an unexpectedly interesting window into how distributed communities construct and negotiate knowledge.