2026 Indiana Fever
127 · 2026-06-25 · medium
One of the biggest issues with artificial intelligence is that it can't tell what day it is. HTML5, which ensures that web pages look and behave similarly in all browsers, was hammered out over a four-year period in the 2000s. In early drafts, there was a standardized "pubdate" attribute, to record the date and time a web page is published. This didn't make it into the final specification; Google, Apple and Mozilla were focusing on structure and speed, not metadata. Two decades later, the World Wide Web is AI's superprimary source of information, and the lack of reliable timestamps has turned into a big problem. If you want an illustration of why, try subscribing to a Google News feed these days. It used to be that every new URL would be logged and tagged in a gigantic database; if that URL popped up again on a subsequent crawl it would be skipped. Now Google News is run by Gemini, so you'll randomly get a bunch of news in your news feed from years and decades ago. I recently got Obama signs 9/11 health bill – despite the January 2, 2011 publication date right there in the goddamn text. It makes me want to inject a virus of miniature digital Flavor Flavs into every single one of their AI systems. Anyway, basketball. If you watch as much women's hoops as I do, you are fully aware that Amazon Web Services is the official AI partner of the WNBA, because they have enough money left over from building gigaclusters to buy thousands of advertising impressions. AWS imagineers created a new metric called Gravity, which quantifies defensive attention with a 25fps "optical tracking system" that uses 3D pose detection. This sounds like great data if you were, say, I don't know, developing a video game. And someone really should. NBA 2K26 is the first edition that lets you Crossplay online with WNBA teams, but the majority of the females play like nerfed G Leaguers with boobs. ("The real Gamergate is that 2K Sports is blatantly sexist," says the inventor of Shirtdoku, who plays more video games than me.) There are fewer than a handful of exceptions: A'ja Wilson has a 98 OVR, and Napheesa Collier is at 97. Caitlin Clark has the same rating as Victor Wembanyama (96), despite only playing 13 games last season. Which brings me to another one of the biggest issues with artificial intelligence: the algorithm values quantity over quality, every single time. These days, when she's not getting the shit kicked out of her by opposing players, you can usually find Caitlin Clark on the disabled list. I asked Google Gemini why she gets injured so much, and it responded with Something went wrong (1099). Another AI, from a different country, responded with three bullet points: a.) the physical toll of the pro game, b.) cumulative soft tissue injuries, and c.) mental/physical fatigue. All of its cited resources were web pages from nytimes dot com slash athletic – which, as I noted last summer, was publishing more stories about Caitlin Clark on the DL than about Wilson and Collier combined. With the future of AI accuracy on the line, do we need more sportswriters who aren't straight and Caucasian, or at least sportswriters who are aware that the last two generations of top WNBA stars have played year-round for overseas paychecks? Well, ummm, yeah. Duh. While we're all waiting for that to happen, it's Hoop Girl Summer, and everyone wants to know the answer to the $16 million question. Will Caitlin Clark be healthy enough to play at the FIBA Women's World Cup in September? Yeahhh, boyyyyyyyy! (1076).