source&pool
A daily wire of long-form journalism, video, and discourse — filed, tagged, and laid out flat.
VOL. I·NO. 01
WEDNESDAY, SEPTEMBER 16, 2026
Hacker News3752X 主题热门3645MacRumors79CNBC77YahooFinance649to5Mac61Kotaku44Verge43IGN349to5Google32Gematsu32aihot31NintendoLife31Engadget26TechCrunch26Eurogamer25BusinessInsider23Guardian20NBC16CNET15NPR15FoxBusiness14Polygon14Fortune13SeekingAlpha13bgr12Gizmodo12CBS11Wccftech11Investor'sBusinessDaily10Mashable10PushSquare10TechPowerUp10USAToday10WIRED10CNN8NintendoEverything8Notebookcheck8NewYorkPost8CrudeOilPricesToday8VideoGamesChronicle8ABC7ArsTechnica7Fox7GameInformer7WindowsCentral7BleepingComputer6PetaPixel6AppleInsider5Deadline5DigitalFoundry5GamesIndustry.biz5Variety5Yahoo5AlJazeera4AndroidPolice4DroidLife4MotleyFool4GameRant4Jalopnik4PureXbox4SamMobile4SlashGear4Hacker4AP3CTech3CanonRumors3ChromeUnboxed3CoinDesk3GameDeveloper3GSMArena3Motor13Blizzard3XBOXWire3PCMag3PCWorld3SeattleTimes3Space3Register3TweakTown3YGOrganization3ZDNET324/7WallSt.2Aftermath2AndroidCentral2AwfulAnnouncing2BleedingCool2BuzzFeed2DigitalCameraWorld2DualShockers2DW2EventHubs2Futurism2Hodinkee2Independent2Lifehacker2MassivelyOverpowered2MyNintendo2Nature2Newser2Newsweek2PaulKrugman2PokémonGOHub2RoadtoVR2RPGSite2Conversation2Intercept2NextWeb2Tom'sGuide2UploadVR2VideoCardz2WarhammerCommunity2WindowsLatest2YourTango2404Media143rumors1ABC111AboveLaw1ageofempires1AndroidHeadlines1AOL1AVClub1Benzinga1BikeRadar1Billboard1BloodyDisgusting1Borderlands1Bungie1Yahoo!FinanceCanada1CineD1CnEVPost1comicbook1CreativeBloq1CyberSecurityNews1DailyKos1DCRainmaker1derekthompson1Draftsim1CNN1Euronews1flatpanelshd1FratelloWatches1FrequentMiler1GAMINGbible1garymarcus.substack1GearPatrol1GeekWire1GeekyGadgets1Hackaday1HollywoodReporter1InsiderGaming1InterconnectsAI1InterestingEngineering1JapanTimes1KITCO1KrebsonSecurity1KSL1LosAngelesTimes1Lloyd'sList1WPLGLocal101Macworld1Maxroll1Mediaite1MiddleEastEye1MonochromeWatches1MPR1SemiAnalysis1Newsshooter1NoMan'sSky1nylon.com.sg1NYT1OregonLive1PCGamesN1PersonaCentral1Pokemon1politico.eu1PittsburghPost-Gazette1QuantaMagazine1qz1RockPaperShotgun1SammyGuru1ScienceAlert1ScientificAmerican1SouthChinaMorningPost1Semafor1SFGATE1YahooFinanceSingapore1YahooSingapore1SportsIllustrated1SimpleFlying1Sources1supercarblondie1Tedium1TelecomTalk1GameBusiness1TheGamer1Times1LongmontTimes-Call1TmoNews1TopGear1TwistedVoxel1YahooFinanceUK1UnHerd1vox1WhatHi-Fi?1WPBF1WRAL1
  1. 001Hacker NewsSEP · 16English

    Vercel Labs

    Vercel Labs is a developer tools initiative that releases experimental projects early, gathers user feedback, and invests in proven solutions. Projects are classified by stage: Labs products (maintained tools), active experiments (iterating projects), and past experiments (discontinued or evolved projects).

    By MehrdadKhnzd
  2. 002Hacker NewsSEP · 16English

    We cut CDN metadata lookup latency by 91%

    Vercel reduced CDN metadata lookup latency by 91% by grouping path metadata into shards instead of fetching individual paths. The optimization batches metadata fetches together while using indexed shards to avoid unnecessary parsing, balancing efficiency gains against data transfer costs.

    By Tim Caswell
  3. 003Hacker NewsSEP · 15English

    Gemini 3.8 Live and 3.8 Live Extended Thinking

    Google introduces Gemini 3.8 Live and 3.8 Live Extended Thinking, two new AI models designed for real-time voice conversations and complex reasoning tasks. The models deliver near real-time processing, multi-language support, and tool execution capabilities, with 3.8 Live Extended Thinking achieving top scores on Speech to Speech Quality benchmarks while maintaining competitive pricing.

    By Tom Ouyang; Malini Jaganathan
  4. 004Hacker NewsSEP · 14English

    Ask HN: The amount of AI bot traffic is out control?

    A web app operator reports severe bot traffic problems, with Facebook's crawler making 3 million requests in 2 days and AI scrapers using residential proxies to bypass security measures including Cloudflare Turnstile. Despite blocking efforts, costs have surged and the operator is considering switching to hCaptcha as a last resort.

    By flyingcoder
  5. 005Hacker NewsSEP · 14English

    We cut CDN metadata lookup latency by 91%

    Vercel reduced CDN metadata lookup latency by 91% by grouping path metadata into shards instead of fetching each path individually. The new approach uses JSONL layouts with inline indexes, allowing the CDN to warm cache for many paths per fetch while still parsing only the needed record.

    By Tim Caswell
  6. 006Hacker NewsSEP · 14English

    Show HN: Skillzero – save tokens by omitting skills from agent context

    Skillzero is a tool that optimizes agent context by categorizing skills into unmanaged, hidden, and collection types, reducing token consumption and preventing accidental skill triggers. It allows developers to hide less frequently used skills from agents while keeping them manually invokable, and bundle related skills into topic-based collections that load on demand.

    By Kurtextrem
  7. 007Hacker NewsSEP · 14English

    Show HN: I made an unlimited TikTok video generator to get views on the internet

    A developer created an unlimited TikTok video generator tool that allows users to paste viral video URLs and generate their own AI versions, built on Modal and hosted on Vercel to simplify content creation for app promotion.

    By phanahm
  8. 008Hacker NewsSEP · 11English

    How to Build an AI Software Factory: Agents That Open, Review, and Merge PRs

    An AI software factory is an automated system that queues coding tasks, provisions isolated workspaces for agents, verifies changes automatically, and routes decisions to humans at explicit merge gates. The key insight is that a single engineer with a coding agent can generate pull requests faster than a team can review them, making the bottleneck not code generation but absorption of output through proper infrastructure and gates.

    By Hiba Fathima; Firecrawl
  9. 009Hacker NewsSEP · 09English

    DeepSeek v4.1 Flash Beta on Vercel

    DeepSeek V4.1 Flash Beta is an experimental model available on Vercel's AI Gateway, priced at $0.22 per million input tokens and $0.66 per million output tokens, expiring September 10th 2026. It integrates with the AI SDK and supports OpenAI and Anthropic API formats with routing, reasoning, and provider options.

    By JonSchneider
  10. 010Hacker NewsSEP · 09English

    They are still training on your data

    AI companies continue training models on user data through synthetic data generation techniques, despite privacy claims. A Google DeepMind method called Generative Data Refinement converts private data into training material by rewriting sensitive examples while preserving useful information, allowing companies to claim they don't train on original data while benefiting from its content.

    By Joseph Perla