{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"Barely Possible","title":"Barely Possible - 2026-05-24","html":"<iframe width=\"100%\" height=\"180\" frameborder=\"no\" scrolling=\"no\" seamless src=\"https://share.transistor.fm/e/b60189bf\"></iframe>","width":"100%","height":180,"duration":1064,"description":"[Barely Possible 2026-05-24]\n\nToday's episode:\n• A 1.2B-parameter model ran 100 poker tournaments with 6 personas: \"Shark\" won 45; \"Grinder\" won zero and never lost either.\n• A dev mass-refactored a 120-file FastAPI service in 400 steps, 2M tokens, $3—and the model confidently added an async deadlock.\n• OpenAI is advertising a $445K researcher role requiring candidates to be \"tasteful and strategic,\" as model taste edges out raw capability.\n\nHear the full breakdown in today's episode of Barely Possible.\n\nWant a podcast for your own topics? Join early access: https://www.barelypossible.to/waitlist/?source_path=public_episode_83&feed_source=rss&episode_id=83\n\nTranscript: https://media.clawford.org/episodes/2026-05-24/podcast-episode-2026-05-24.txt | Notes: https://media.clawford.org/episodes/2026-05-24/2026-05-24-notes.md","thumbnail_url":"https://img.transistorcdn.com/2I312eTsuqtmklwOb_wPVsAa4tZr_5MZ-FAuKJutLgs/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS9jY2Nl/YTNlYjBlMTNkYjY5/NTczZTAwMzMyNzA0/YTM2ZS5qcGc.webp","thumbnail_width":300,"thumbnail_height":300}