There’s a lot of debate about whether this reduces model accuracy, but this is basically Chinese grammar and Chinese vibe coding seems to work fine while (supposedly) using 30-40% less tokens

It's like googling... if you have skillz/experience you can google almost anything with 3-4 words...

It’s commonly “they”

For example: “Meta released a new product, they (Meta) are calling it …”

Both sports betting and prediction markets are effectively the same as reading tea leaves anyway. You can determine the real odds in roulette or blackjack because those are closed games that have simple probabilities, but any odds given in sports betting or prediction markets are just fancy guesses

The odds in betting are the payout ratio. To hinder manipulation, the payout ratio would have to be hidden and then the platform can just always say you won less money than you did.

Let me know when it runs DOOM

A valid use case I can think of is if npmjs.com is blackholed by your company to prevent supply chain attacks

> The blocker wasn't capability—it was liability.

Yikes (regarding the AI patterns in the comment)

Isn’t that why you should like it then? It saves your time because you’d get the page earlier

Or Rspack (https://rspack.rs/) which is intended to be a drop-in Webpack replacement

how is this possible? isn't everyone using the same node version?

and checking in lockfile changes

I think they're referring to the spiders that are deep in the web, since the midges presumably don't make it that deep