Happy to answer any questions about the results or the proofs!"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"New Lower and Upper Bounds for the Grothendieck Constant"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://arxiv.org/abs/2608.11158"}},"_tags":["comment","author_alan_li_","story_49303652"],"author":"alan_li_","comment_text":"I am one of the first authors of this paper. To summarize our work, the Grothendieck constant (KG) is a fundamental constant in fields like analysis, algorithms, and quantum information, and concretely we prove 1.7135 < KG < 1.7819. The results came from truly novel ideas, and diverged significantly from the previous literature on proving upper and lower bounds. Our results were discovered in collaboration with a long-horizon research harness we built using publicly accessible frontier models, with separate reasoning and coding agents. We found that although AI excels at technical execution, they need deep human insight for research taste\u2013understanding which ideas to pursue or abandon. We\u2019ve documented our research experience as a case study of long-horizon human-AI collaboration; we hope to help shape future conversations around humans and AI in math, in a time dominated by results from large AI companies.
Happy to answer any questions about the results or the proofs!","created_at":"2026-08-15T12:57:45Z","created_at_i":1786798665,"objectID":"49310185","parent_id":49303652,"story_id":49303652,"story_title":"New Lower and Upper Bounds for the Grothendieck Constant","story_url":"https://arxiv.org/abs/2608.11158","updated_at":"2026-08-17T13:41:46Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"yohji1984"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["long","horizon","harness"],"value":"I wrote a blog yesterday on how useless RTK and Ponytail are on real coding tasks. And published my agent harness long-horizon task benchmarks on 80% real token saving.
I just want to know why people just ignore the fact those pulgins are useless and don't care about the real savings?
full reports are on my repo: https://github.com/Tura-AI/tura
Arm n Harness score Total tokens Modeled cost Rounds Duration\nNo plugin 2 78.85% 6.660M $5.281946 62.5 895s\nPonytail 2 80.77% -7.56% -8.87% -9.60% +13.51%\nRTK 2 76.92% +13.20% +7.18% +44.00% +40.69%
Configuration Passes Pass rate Observed tokens Rounds Estimated cost\nTura Balanced High 48/60 80.0% 229,695,477 2,017 $221.138\nTura Direct High 39/60 65.0% 75,108,167 969 $99.620\nCodex CLI Medium 38/60 63.3% 333,538,349 3,140 $257.173\nCodex CLI High 36/60 60.0% 455,742,296 6,074 $327.483"},"title":{"matchLevel":"none","matchedWords":[],"value":"Why people chasing after useless token saving plugins and ignoring real solution"}},"_tags":["story","author_yohji1984","story_48937781","ask_hn"],"author":"yohji1984","created_at":"2026-07-16T17:46:52Z","created_at_i":1784224012,"num_comments":0,"objectID":"48937781","points":2,"story_id":48937781,"story_text":"I wrote a blog yesterday on how useless RTK and Ponytail are on real coding tasks. And published my agent harness long-horizon task benchmarks on 80% real token saving.
I just want to know why people just ignore the fact those pulgins are useless and don't care about the real savings?
full reports are on my repo: https://github.com/Tura-AI/tura
Arm n Harness score Total tokens Modeled cost Rounds Duration\nNo plugin 2 78.85% 6.660M $5.281946 62.5 895s\nPonytail 2 80.77% -7.56% -8.87% -9.60% +13.51%\nRTK 2 76.92% +13.20% +7.18% +44.00% +40.69%
Configuration Passes Pass rate Observed tokens Rounds Estimated cost\nTura Balanced High 48/60 80.0% 229,695,477 2,017 $221.138\nTura Direct High 39/60 65.0% 75,108,167 969 $99.620\nCodex CLI Medium 38/60 63.3% 333,538,349 3,140 $257.173\nCodex CLI High 36/60 60.0% 455,742,296 6,074 $327.483","title":"Why people chasing after useless token saving plugins and ignoring real solution","updated_at":"2026-07-17T00:43:40Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"zamir_akimbekov"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["long","horizon","harness"],"value":"Build domain-specific harness and long-horizon agents with OS Python framework"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/cayu-dev/cayu"}},"_tags":["story","author_zamir_akimbekov","story_49721806"],"author":"zamir_akimbekov","children":[49721831,49721945],"created_at":"2026-09-16T03:35:04Z","created_at_i":1789529704,"num_comments":0,"objectID":"49721806","points":3,"story_id":49721806,"title":"Build domain-specific harness and long-horizon agents with OS Python framework","updated_at":"2026-09-16T04:03:42Z","url":"https://github.com/cayu-dev/cayu"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"matt_d"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["long","horizon","harness"],"value":"EvoHarnessRL: Learning Self-Evolving Runtime Harness for Long-Horizon LLM Agents"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://arxiv.org/abs/2608.05446"}},"_tags":["story","author_matt_d","story_49249998"],"author":"matt_d","created_at":"2026-08-10T21:28:07Z","created_at_i":1786397287,"num_comments":0,"objectID":"49249998","points":3,"story_id":49249998,"title":"EvoHarnessRL: Learning Self-Evolving Runtime Harness for Long-Horizon LLM Agents","updated_at":"2026-08-10T22:45:13Z","url":"https://arxiv.org/abs/2608.05446"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"tingletech"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["long","horizon","harness"],"value":"LongHorizon-Harness: Advancing Long-Horizon Agents for Real-World Tasks"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["harness"],"value":"https://github.com/AMAP-ML/LongHorizon-Harness"}},"_tags":["story","author_tingletech","story_49339286"],"author":"tingletech","children":[49339457],"created_at":"2026-08-17T23:51:43Z","created_at_i":1787010703,"num_comments":0,"objectID":"49339286","points":1,"story_id":49339286,"title":"LongHorizon-Harness: Advancing Long-Horizon Agents for Real-World Tasks","updated_at":"2026-08-20T03:51:26Z","url":"https://github.com/AMAP-ML/LongHorizon-Harness"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"stevenhuang"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["long","horizon","harness"],"value":"It takes quite a lack of foresight to think RSI is completely speculative when it's already been demonstrated how capable agents are at long horizon tasks given suitable harness and unambiguous success criteria. It's hardly a leap to give LLM the goal of improving itself on benchmarks and let it conduct it's own experiments and spin up training runs completely unsupervised.
It's strange you believe this can't happen when a weaker form of it is already happening. And to be so certain RSI can't happen when there really is no technical basis why it can't."},"story_title":{"matchLevel":"none","matchedWords":[],"value":"We must pace the frontier"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://darioamodei.com/post/we-must-pace-the-frontier"}},"_tags":["comment","author_stevenhuang","story_49672510"],"author":"stevenhuang","comment_text":"It takes quite a lack of foresight to think RSI is completely speculative when it's already been demonstrated how capable agents are at long horizon tasks given suitable harness and unambiguous success criteria. It's hardly a leap to give LLM the goal of improving itself on benchmarks and let it conduct it's own experiments and spin up training runs completely unsupervised.
It's strange you believe this can't happen when a weaker form of it is already happening. And to be so certain RSI can't happen when there really is no technical basis why it can't.","created_at":"2026-09-12T21:29:42Z","created_at_i":1789248582,"objectID":"49677410","parent_id":49677039,"story_id":49672510,"story_title":"We must pace the frontier","story_url":"https://darioamodei.com/post/we-must-pace-the-frontier","updated_at":"2026-09-18T02:17:02Z"}],"hitsPerPage":8,"nbHits":73,"nbPages":10,"page":0,"params":"query=%22long-horizon%22+harness&hitsPerPage=8&advancedSyntax=true&analyticsTags=backend","processingTimeMS":11,"processingTimingsMS":{"_request":{"queue":40,"roundTrip":15},"afterFetch":{"merge":{"total":1},"total":1},"fetch":{"query":6,"scanning":3,"total":10},"total":11},"query":"\"long-horizon\" harness","serverTimeMS":52}