[rektide]

> Sorry, do you honestly think that Tweets are more complicated than real time video and voice

do you not? have you heard of metcalfe's law? the value of a network is proportional to the square of the nimber of users. so too is the difficulty in supporting & running the vastly more interconnected network.

video is easy as shit. yeah you need bandwidth/transit. but people join & leave infrequently. they are low impact consumers: them joining does not change the total load significantly, it's only a little more linear growth. whatever to that, irrelevant.

twitter is a vastly wider & more connected not-so-micro- cosm. if there are 50m followers, and they tweet, the burden of rippling that event out is huge. tracking & making visible every fav, retweet & reply further escallates the difficulty by more deep deep orders of magnitude.

it's laughable to me that you would try to say discord has a higher event rate than twitter. first, i think thats horseshit, totslly absurdly flat wrong in extreme if you count favs, retweets, et-cetera. but more so, event rate is pointless. it's event receievers that actually count, that matter. and the little intranet micro-world discords are, by any measure, totally pointless & irrelevant by compare versus the all-connected fabric that is twitter. discord has no scale, is itrelevant entirely, in how far it has to spread any given message/event. discord is a large number of the tiniest little fishbowls; it compares not at all to the sea.

[justahuman74]

> video is easy as shit

Agree. We used to just stick a giant metal pole in the air and ask others to do the same with a smaller pole.

[rektide]

It's just data, is my point. Events are harder, require actual processing.

[bluefirebrand]

I'm really curious what you think is so hard about Twitter.

As far as I can tell it is basically a CRUD app with some real time analytics to surface some trends.

You said some stuff about Events bubbling, like if I Retweet something you seem to think there's a lot of complexity there. But as far as I can figure all that does is create a new row in a database somewhere and when someone who follows me fetches their feed, my new retweet gets included in there.

Aside from the scale, what's hard about it? Am I completely missing something?

[fragmede]

"Aside from the hard part, this seems easy".

[bluefirebrand]

Is this even the hard part anymore? In 2006 maybe, but nowadays it seems like scaling is pretty solved. I always feel like scaling is more just expensive than hard.

[fragmede]

I mean, you can make things easy for you and just choose GCP and use Spanner instead of trying to build your own from scratch, but that doesn't make the problem itself easy.