Here's how a Twitter engineer says it will break in the coming weeks | Hacker News Reader

Here's how a Twitter engineer says it will break in the coming weeks

technologyreview.com

29 pointsilamont22 comments

Twitter uses Scala heavily. It is one of my favorite languages. I love Scala, but it's also sprawling and complex, and requires great care to code for readability without getting too cute.

I would expect there is a lot of code at Twitter written by Scala show-offs and acrobats that the remaining engineers are going to have to read and understand now.

Ooofff.

I used to think that Elon would make Twitter better.

Now I'm not sure. His erratic behavior is showing up more than ever.

He cut half of the engineering staff, realized Twitter needed some fired engineers, while trying to cram untested monetization features into production. He's flip flopping on ideas within hours.

If advertisers continue to pull out, he's going to have continue to sell his TSLA shares.

In Feb 13, exactly one year before Whatsapp sold to Facebook for 19 billion dollars, they had about 200 million active users and only 50 staff members.

Firing half of the 7500 Twitter employees (most at $100k+ comp', lots at $300k+ comp') does not seem that erratic to me.

Facebook, at one point, had 50 employees too. Doesn't mean Facebook today can run on 50 employees.

Twitter has 396.5 million users globally today. That's about twice what Whatsapp had at the time when they had 50 employees. Yet catering to twice the users takes Twitter 150 times as many employees. There's a huge gap spanning two orders of magnitude there.

> Facebook, at one point, had 50 employees too. Doesn't mean Facebook today can run on 50 employees.

Indeed Facebook had 50 employees at one time, and indeed Facebook most probably can't run on 50 employees toay given they have 2.5 billion users. But I'm not sure what your point is?

My point is that you're comparing apples and oranges.

My last company had 5,000 users and hundreds of employees. Does that mean we should reduce the employee count to 1 employee given that WhatsApp was able to sustain 200m users with 50 employees?

How how about the fact that WhatsApp generated $10m in revenue when it had 50 employees or 200k/employee while Twitter in 2021 generated $5b or $677k/employee?

Your logic is so bad that I didn't even want to reply.

I think people are disagreeing with you because you show a fundamental misunderstanding of how larger companies work. There is not a simple "queries per employee" metric that makes sense as companies grow larger. Large sites with a lot of surface area have to solve different problems- and structure their business, which includes larger teams working on different problems- and if you got rid of them, you'd run a risk of serious problems with your infrastructure.

At Facebook, if you fired all the UX designers, the site could stay up for years. If you fired all the SREs, the site would be down in two weeks.

While there is something to be said for tech bloat and obsession with pouring money into insanely complicated distributed systems, WhatsApp does not have to have a moderation system covering the whole planet and a confusing maze of government laws and regulations based on locale. Outside of encryption, it's by default non-controversial. Twitter is.

I wonder what will happen to the Twitter’s third party clients relying solely on API. Nobody seems to noticed them.

I'd be shaking now. Building a business upon someone else's API...

Yes I am using Tweetbot and I’m worried for them.

isn't it likely that they won't do anything with the api, if it works currently why would it break? Seems like twitter would have redundancy built in. Barring someone hacking into systems and breaking them what could go wrong in the near future.

Keeping an API stable is a surprisingly large amount of work. I've never worked with a codebase as large as Twitter's, but I have seen that even with testing in place it is easy for seemingly unrelated changes to break the API. If Twitter has reduced engineering staff, API functionality would be one of the first things to suffer in my estimation.

This doesn't manifest as someone purposely breaking the API, but incidental changes breaking it. For example, a minor change in something like time display code that does the "posted X minutes ago" could easily end up silently corrupting API calls in odd ways. And if the engineering staff is reduced, fixing the API that broke a 3rd party vendors product will likely be a very low priority.

> if it works currently why would it break?

Because they are planning much more rapid iteration with relatively little planning on core product features with a butchered staff and quite likely not prioritizing maintaining API stability for third-party tools while doing it and, its very overtly a “move fast and break things” mindset.

Plus, the new high priority Blue-with-verification rollout has been quite visibly broken on the web and first party apps, with people getting inconsistent reeults about availability when trying to upgrade to it and having inconsistent results as to whether (and to whom, and in what contexts) the checks display after they do.

Do you think keeping the API working for third-party integrations has a higher priority?

Yes I agree, I have been following Twitter’s API for years and the progress is just so slow. I even followed their Postman API.

Considering that Musk is on a roll making decisions that will outright murder the platform (making everyone pay for the Blue Check or your tweets are treated like garbage), who's to say the new boss will not come up with some knee-capping payment regime for API clients?

No one knows. The guy is increasingly an unstable wildcard.

Reply on news.ycombinator.com