Hacker News Reader

Hacker News

Comment by omega3 | Hacker News Reader

Parent Full thread

omega3

Twitter doesn't need "best people" - it's not innovating and hasn't been for a very long time.

View on HN

"Innovating" is not the only nor primary thing that requires "best people".

I agree, just keeping things running at that scale requiere people with some impressive skills

We'll see, my bet is that nothing catastrophic will happen as a result of layoffs and hiring freeze at Twitter.

Revenue-wise I think a fake checkmarked Nintendo being up with a photo of mario doing the middle finger for hours does a lot, and spells not great stuff. My understanding is that advertisers aren't even locking in contracts for next year or similar.

I bet advertisers LOVE this situation. They probably weren't excited about how centralized systems are, which robs them of some of their say and power in the ecosystem. This is an opportunity for them to show what control and influence they have and basically a warning to others not to toy with their demands.

Twitter has already been having reliability issues for weeks or more, that's not going to get better if you lay off a bunch of your ops and sweng personnel

innovating is a meaningless term in ad tech companies. optimizing I think is a better word.

Well it's currently operating at a loss, so if they don't innovate somehow the company is doomed.

It's amazing how many techies seem to be unable to understand this as if there was still infinite VC money and cash raising via stock sale for every company that can't figure out how to make money.

If we get a general tech correction as part of this downturn things are going to be very different for companies that have been burning money for years.

Yeah but if they want to do anything more than struggle to fight fires 24/7 they need at least average people, and management that trusts them, and buy in for the work needed to make a robust system.

Have you any idea what it takes simply to maintain a machine such as twitter, let alone make even minor alterations? There was a recent telling interview by someone who used to work there who thought it wouldn't be long until the whole thing collapses on its own because there were huge teams dedicated to simply keeping the thing running.

Back of the envelope calculation:

A billion users, each tweeting once an hour with 140 characters. That's 38.147 Megabytes/s. My laptop could handle that raw volume. Increase it by an order of magnitude for all the network nonsense and it can still run on my 4 year old desktop.

Twitter is not some hypertech company, it shouldn't need more than a hundred engineers to run. I imagine that's the bet Musk is making too.

What a terribly dishonest and overly-simplistic way of modeling of a distributed system much less a simple web service. found the engineer who, in their own words, “couldn’t code their way out of paper bag.”

If you're serving 40mb a second you don't _need_ a distributed system.

Twitter isn't Netflix.

In the same vein, look at how Plenty of Fish has a huge customer base, and runs on very skimpy hardware. Back in 2006 it had 45M visitors a month, served up over 1B page views a month, all running off three database servers and two load balanced webservers. Guess how many employees? One, Markus Frind[1].

1. http://highscalability.com/plentyoffish-architecture

Of course things have changed, money will do that.

Twitter has about 10x that monthly visitor number just in mDAU. And pof has scaled 100x! (To 100 employees — that seems pretty insane relative the traffic they have going by this weird metric of “amount of data served should roughly equal the number of employees by some ratio”). Comparison also seems a bit lacking given the difference in magnitude also the engineering problems involved (e.g. moderation, botting etc.) Guessing also that creating a dating site is not an exercise in needing a lot of skilled engineering work given it’s been a solved set of problems since the late 90s. Hey Verizon has 132,000 employees — I guess they should only need a fraction of that right since consumer cellular has 2,400?

It’s pretty laughable you believe your own “math.” I guess even serving an actual front end doesn’t factor into your calculations. Hey go build something and you might find out what it actually takes to build/maintain a system of any real consequence instead of doing leet code exercises and smelling your own brain farts.

I guess you're right, you need 10,000 JS engineers to change a light bulb.

Reply on news.ycombinator.com