@usr_bin_roygbiv I did get banned once using gemini-cli with opencode sdk, is it better now? Or does running something with less overhead like Pi put less meta data in the headers to trigger review?
@drewvolpe@pitdesi@cremieuxrecueil Yeah, at top schools, over the past decade CS has been the top earning major. At other schools CS is often just a soft landing in entry level IT work
@JoelStaub@EXM7777 I think CC is a more bloated harness, but in the past few months itโs added some features that make it worth it. Models have also gotten much better at tool use and long term thinking, so the larger tool selection and system prompts donโt pollute context as much
Claude started having Chinese in it's responses sometimes, has anyone else seen this? I've had it happen with MiniMax and GLM before, but never on Claude before Opus 4.8
@thdxr before cursor, when I was too lazy to look up the syntax I would just do this...
const step1 = doThingA(input);
const step2 = doThingB(step1);
const result = doThingC(step2);
To be honest, I still do. My brain likes named functions better than reduce in JS
@BayAreaNewLibs@loganb TN requires budget-neutral reappraisals every few years, which leads to a similar result. It keeps reassessments from being a way to silently increase tax rates
I've felt like the coding plans are gateways to get developers to use the claude-sdk integrated within the app, rather than just to build the app. If claude-sdk is integrated within the application stack, then the market opportunity is all digital services. claude-sdk could have databases or a full cloud sdk included. If it's already in your dev dependencies and billing already, then it lowers friction for more services
@_fat_ugly_rat_ Growing up in the red area, I knew a lot of people who felt like Obama stole the nomination from Hillary. It feels foreign now, but the mid south still had a base loyal to the Clintons and Gores. Several of my teachers had photos of them meeting Hillary in their classroom.
@rai__as@CollapseAnime For training AI models, you need the latest generation to keep up, but for running them you donโt. Itโs why X is using older chips to run Anthropicโs models, but newer chips to help Cursor train composer
@rai__as@CollapseAnime Samsung and TSMC only have so much capacity, so they focus on the latest generation. Chinese firms canโt produce latest gen chips, but they are getting really good at making last gen chips
@thdxr Maybe in a few months when Effect is in more training datasets and reasoning gets better they might be able to use Effect better in greenfield cases without examples
@thdxr Yeah, I think the best packages and languages for agents are still the ones theyโre trained on. Javascript and Python have benefits, but theyโre also the majority of the training data. They can make up for reasoning ability by basically auto completing
@Prototype27F@The_Devin_Green Imagine being a Betazoid when the Pheobian navy attacks. Only the nomadic betazoids still left on the great beta steppe are safe
@JupiterKunt@_fat_ugly_rat_ This is the kind of medieval construction project that would have caused someone to get the nickname โHarold the Builderโ or โWilhelm the Greatโ