An unsolicited redesign of arXiv's listing page

arXiv is the open-access preprint repository run by Cornell Tech that most of computer science publishes through before peer review. It's an institution worth liking: free, fast, and it's why a paper posted this morning can already be read anywhere in the world. This page is a studio redesign of one corner of it — the recent listing for cs.HC, the Human-Computer Interaction category.

What the original gets wrong

The real listing page is a single long, plain-text column: title, authors and subjects run together in one small block per paper, in strict chronological order, with no visual hierarchy to help a reader's eye find the title first. There's no way to filter by subject tag or search by author without leaving the page, so browsing means scrolling past dozens of unrelated papers to find the handful you actually care about.

What changes here

The same real papers — pulled from arXiv's own public API, not screenshotted or copied — are laid out as a card grid with a clear typographic hierarchy: title first, then authors, then subject tags, then a trimmed abstract. A search box and subject-tag filters let you narrow the list in place, and a sort control makes the ordering explicit rather than implicit. Every card still links out to the real abstract and PDF on arXiv, because the redesign is about how you find a paper, not a replacement for reading it there.