[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-researchers-build-a-tool-to-expose-codes-worst-case-slowdowns":10,"sections":35},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":24,"tags":25,"sources":30,"feedback":34,"feedback_at":22,"cost_usd":34,"total_tokens":34},9441,"researchers-build-a-tool-to-expose-codes-worst-case-slowdowns","Researchers Build a Tool to Expose Code's Worst-Case Slowdowns","STAB generates test inputs from a problem's spec alone, catching algorithmic worst-case slowdowns far more often than previous testing methods.","A new testing pipeline called STAB writes code-breaking test cases straight from a problem's plain-English description, with no finished implementation required.\n\nSTAB splits the job into two parts. A \"constraint saturator\" reads the problem spec, pulls out its stated limits, and uses rule-based logic plus CP-SAT optimization to push input sizes to the edge of what the problem allows. An \"adversarial scenario injector\" then pulls known troublemaking input patterns from a curated catalog, matched to the problem through keyword search and k-nearest neighbors. STAB folds the spec, the resolved size boundaries, and the retrieved attack patterns into a structured brief, which an LLM turns into a working Python test generator. On the CodeContests benchmark, this lifted the rate of test cases that actually exposed algorithmic worst-case slowdowns from 50.43% to 73.45% for open-source LLMs and from 57.45% to 71.85% for closed-source ones, with similar gains across Python, Java, and C++.\n\nThat's a real gap in how efficiency gets tested. Earlier approaches either threw bigger inputs at code and hoped something broke, or reverse-engineered slow cases from a specific implementation already sitting in front of them. STAB instead reasons about the structural conditions that trigger worst-case behavior before any code exists to test, which matters as more algorithms get LLM-generated and nobody checks whether \"correct\" also means \"fast enough.\"\n\nA 73% hit rate is progress, not a guarantee - meaning roughly one in four genuine bottlenecks still slips through unnoticed.","[\"ai\",\"software-testing\",\"llms\",\"benchmarks\"]","2026-10-01T04:00:00.000Z","2026-10-02T19:05:55.407Z","2026-10-02T19:05:59.987Z","published",null,[],"dev-tools",[26,27,28,29],"ai","software-testing","llms","benchmarks",[31],{"name":32,"url":33},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2605.27981",0,{"sections":36},[37,41,46,51,56,61,66,71,75,80,85,90,95,100],{"name":38,"slug":26,"count":39,"latest_published_at":40},"AI",5746,"2026-10-02T04:00:00.000Z",{"name":42,"slug":43,"count":44,"latest_published_at":45},"Security","security",829,"2026-10-01T17:31:56.000Z",{"name":47,"slug":48,"count":49,"latest_published_at":50},"Policy","policy",437,"2026-10-01T18:10:00.000Z",{"name":52,"slug":53,"count":54,"latest_published_at":55},"Deals","deals",317,"2026-10-01T22:00:00.000Z",{"name":57,"slug":58,"count":59,"latest_published_at":60},"Hardware","hardware",198,"2026-10-01T17:38:48.000Z",{"name":62,"slug":63,"count":64,"latest_published_at":65},"Science","science",168,"2026-10-01T18:35:55.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Consumer Tech","consumer-tech",155,"2026-10-01T19:54:10.000Z",{"name":72,"slug":24,"count":73,"latest_published_at":74},"Dev Tools",96,"2026-10-01T16:57:03.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Software","software",93,"2026-09-30T21:41:11.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":84},"Startups","startups",90,"2026-10-01T21:55:22.000Z",{"name":86,"slug":87,"count":88,"latest_published_at":89},"Gaming","gaming",53,"2026-10-02T02:50:39.000Z",{"name":91,"slug":92,"count":93,"latest_published_at":94},"General","general",50,"2026-09-30T21:37:54.000Z",{"name":96,"slug":97,"count":98,"latest_published_at":99},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":101,"slug":102,"count":103,"latest_published_at":104},"How-To","how-to",7,"2026-10-01T09:00:00.000Z"]