7 More Corners of the Internet

A manually curated list of websites maintained by one person, updated when they find something worth adding, organized by category without any pretense of comprehensiveness. No RSS, no API, no engagement metrics. The site itself is static HTML. It's been running since 2019 and currently lists around 800 sites. The act of spending an afternoon browsing through it produces a feeling that's distinct from any algorithmically-driven discovery experience — something in the range of serendipity rather than relevance.

Explain Paper lets you highlight any sentence in an uploaded academic PDF and get a plain-language explanation of what it means. It handles jargon, notation, and cross-references reasonably well. Researchers have been using it for papers outside their specialty; undergraduates have been using it for papers that should have been written more clearly in the first place. Neither use case is wrong. The tool is free, the explanations are useful, and it's built by a small team with no apparent venture backing — make of that what you will.

VOGONS (Very Old Games On New Systems) is a forum community dedicated to making old DOS and early Windows games run on contemporary hardware. Founded in 2000, it has accumulated a remarkable archive of technical documentation, hardware guides, and emulation tweaks. The community skews older than most internet forums — people who actually played these games when they were new — but the technical knowledge being preserved there is genuinely irreplaceable, and the tone is refreshingly free of the hostility that characterizes most gaming communities.

Someone built a force-directed graph visualization of Reddit's community structure, showing which subreddits share the most users and how communities cluster around topics. Previous versions of this kind of map were interesting but illegible. This one uses a smarter layout algorithm and lets you search and filter. The clusters are largely unsurprising — gaming, sports, regional — but the connections between them reveal things about shared communities that platform-level analysis doesn't show, and the hobby and interest clusters at the periphery are worth exploring.

Most design education online is either too abstract (here are principles) or too tool-specific (here's how to use Figma). Uxcel sits in the middle: short interactive exercises about actual design decisions, graded against established heuristics. The feedback is specific rather than affirming. Working through a few lessons produces the unfamiliar sensation of having learned something rather than having been encouraged to believe you already know it. The free tier is more useful than most paid alternatives in adjacent spaces.

Kagi is a paid search engine — $10 per month, no advertising — and it has been running long enough to have an interesting track record. The search quality is demonstrably better for technical and non-commercial queries. The business model is the experiment: whether there's a segment of users willing to pay for search that doesn't depend on advertising revenue. Their published user growth numbers suggest there is, though the numbers involved are small relative to the market. It's worth watching as a case study in whether quality can beat free.

Ted Chiang published a short story this quarter that's been passed around in tech circles with the kind of urgency his work usually generates. Without summarizing it too reductively: it's about a person using an AI assistant and what happens to their sense of authorship over time. Chiang's arguments about AI tend to be specific where most writing on the subject is vague, and this story is better than his essays on the same topic because it shows rather than argues. It's in the New Yorker and not paywalled at the time of writing.