The GraphQL API came in around 2020, before that the front end used the same (documented) API as everyone else. The context here is what the API was written for - the current public API was written for their own front-end.

The mods go away, the site goes away.

How would you secure API access for a webapp that runs in a browser and doesn’t require a user login? Any public/rotating key or call-response algorithm in the client code is publicly exposed and liftable.

Yeah, I was going to mention this - APIs are almost always written for first party use first, and then if others benefit from it then that’s a pleasant side effect.

This is a worthwhile observation. I can see the wings being impractical at the launch party, and the spikes could have been a later addition for the photoshoot, but it’s also possible they’re edited in.

I was curious what you might be thinking of so I had a look down Amazon’s acquisitions list. None of the names jump out at me until I get down to MGM (2021), Curse (2016) and Twitch (2014) and Comixology (also 2014). Unrelated, they do seem to have picked up a few AI companies in the last few years.

Couldn’t the same be said about any hobby/pro thing, like gaming? It also has a professional aspect where the “why” is different and the main purpose isn’t the same as the casual scene. I don’t think most people would say you can’t call that gaming though.

I saw this earlier today on the subreddit, and I genuinely thought it was a render at first. Spectacular job, maybe one of the most detailed, accurate cosplays I’ve ever seen.

“Other reviews made me mad so I had a rant on this one even though this one didn’t make me mad” isn’t the bastion of rationality you seem to think it is.

It’s a single paragraph summary of the game that doesn’t reference the graphics at all. Are you sure this is all it takes to piss you off?

I’m still not sure I can take a game named “Lies of Pee” seriously.

The short answer is they had a big launch bank balance from preorders and Sony money, combined with extremely low operating costs (one office, very few staff) and Sean Murray treating it as a passion project, so revenue is mostly reinvested in development instead of profit.

Their risk evaluation for how many streamers they might lose from the platform entirely was done by an idiot. When the real risk started to present itself, they realised the change would cost them more than it would gain.

All good! I think we’re all on the same page that the request pricing they’ve set is obscene.

The bandwidth itself is relatively cheap. At an average 1kb payload per request, Apollo’s 7 billion monthly requests add up to about $500 on AWS (where, as far as I know, Reddit is hosted). The compute costs of retrieving and assembling the request data are much harder to estimate but they’re going to be the bigger…

I’m wondering if you expect all news media to work that way? If a newspaper rightly calls Putin a monster and then later reports on eg. a new tech breakthrough in Russia, would you accuse everyone at the newspaper of kissing Putin photos too?

Weird, this is how that thread looks for me - fresh page load, clicked nothing.

If it was, it doesn’t seem to be that way any more. I tested on a few larger posts side by side in old and new and the links were all in the same spots. There was (is?) a userscript that auto-expanded them, is it possible you have that installed?

The “continue this thread” and “load more comments” links happen on old.reddit too, at the same depth.

Yeah, I also prefer the new one (with “classic” style) to old.reddit. The expand button is worth it alone, but being able to click in the side margins to go back to the list without another page request is also pretty useful and substantially speeds up browsing.