jerbear4328

556 karmajoined 3년 전

comments

They want to sign up for the Amazon Brand Registry[1], which requires a valid US trademark, and these crazy names are approved faster because they're obviously unique. See NYT[2]

[1]: https://sell.amazon.com/brand-registry

[2]: https://archive.is/IQs4i / https://www.nytimes.com/2020/02/11/style/amazon-trademark-co...

jerbear4328

9개월 전discuss

It (hopefully) took exactly 30 seconds, the page delays every item until 30 seconds after its posted date. It doesn't poll HN's server, it opens a websocket to the official HN Firebase, and without the delay, items appear in large chunks. I'm pretty sure the HN server syncs with Firebase every 30 seconds, so this is as fast as it can go while still being accurate.

jerbear4328

9개월 전discuss

I tried this, but it required making a request for every comment and would probably call for a backend, wheras this can run just off of the Firebase websocket stream on a static HTML file.

jerbear4328

9개월 전discuss

Haha, my version makes a websocket connection to the official Firebase that the HN servers already send everything to, so it is zero extra load on HN

jerbear4328

10개월 전discuss

The instance https://cobalt.meowing.de is currently working for YouTube, which is very useful

jerbear4328

11개월 전discuss

https://altstore.io is the big one. You might want the AltStore fork SideStore (you can do the weekly reinstall without a computer, https://sidestore.io). Other tools exist, like https://sideloadly.io and https://appdb.to.

jerbear4328

2년 전discuss

Discord runs in the browser, same as HN. And the HTTP APIs they use aren't officially documented (except HN's read-only ones via Firebase and Algolia) (though both are easily reverse engineered and OSS frontends exist). There's not much difference.

jerbear4328

2년 전discuss

Why even use an LLM? A classifier is perfectly suited for this kind of thing, and they aren't new. As far as I can tell, this is what is often used in the real world, and is incredibly cheap compared to an LLM, so GitHub/$OTHER_PLATFORM could totally run it on everything posted. They could even use a classifier as a first filtering step, then run a smarter model on flagged comments. (Then let a human double-check, right? Right?)