High-scale software is complex, sometimes there are edge cases where weird unexpected stuff happens. This isn’t a situation they would normally run into.

It’s crazy how easy this recent drama has made leaving Reddit for me. Saw all the user-hostile changes and just deleted my Reddit apps and have only been visiting it via google searches on very domain-specific knowledge.

Likewise, it was feeling a little dry here today and I finally figured out the same thing you did. Being able to subscribe across many servers is wicked sick, and having an instance sitting "in front" of them the way we're using it makes it slick as heck when those other instances are unavailable or spotty.

It would be great if the instance kept a pulse on how federation to other instances is going and showed a health check in the app sidebar and near instance names to temper user expectations.

Probably not, .ml is free (provided by the same registrar as .tk) on a year-by-year basis and was much more popular around the time lemmy started.

Massgrave is a tool that can create legit (oem) keys for windows and office out of thin air*

• it’s not literally creating them from nothing, it’s using a system Ms themselves run to get working keys. Evidently they don’t have a huge problem with it.

Mastodon is written in RoR, whereas Lemmy's backend is in Rust. It's an order of magnitude faster just by being a compiled language with lighter-weight middleware.

I haven't used Ruby/RoR in half a decade but even in the early 2010's it was memingly slow compared to many alternatives.

Reddit has attempted, multiple times, to reach out to Andrew Tate to partner with him on advertisements, including twice while he was imprisoned.

I know this because their numbskull marketing folks can’t proofread the email address they’re sending outreach to, and I received the outreach emails as a result. Thanks, Reddit!

Outright telling active partners that they won't help them identify inefficient interactions with the first-party API was a great look.

Good thing most instances are run by other people, then.

Likes/upvotes are “Yeah!”s and there are miis associated with each post

No poop July is so hype

A flatpak of the snap, running in a docker container inside a vm for maximum security.

This absolutely is not true today, they create links that are absolute and refer to the host of the community in question.

Likely experience and knowledge improving the quality of deployment. Most instances are likely underspecced, are on hosts that aren't easy to scale up with, or are maxed out in their current offering tier (lemmy.ml comes to mind there)

This can happen if federation breaks for a while, I think. For instance, if a lemmy instance goes down and can’t receive activity for a time, I don’t think there’s any mechanism to backfill that activity

Checking ip ownership is a moving target more likely to result in outcomes these sites don’t want (accidentally blocking google bots and preventing results from appearing on google).

Checking useragent is cheap, easier, unlikely to break (for this purpose, anyway) and the percentage of folks who know how to bypass this check is relatively slim, with a pretty small financial impact.

Hey! I resemble that statement!

Thank you MargotRobbie, bringer of positivity and Barbie

nope, their pricing is ridiculous lol.

I used Apollo and Relay extensively and not having those makes it so hard to even try for me.