People Matter. Even in Tech
So Space Karen might have really broken Twitter this time.
Over the holiday weekend, Twitter started behaving even worse than it usually has under Musk. First, Twitter became unavailable to anyone not logged in. Then, people began experiencing errors telling them that had been rate limited (usually a message indicating that you have hit a limit related to an application programming interface, not a web page or mobile app native to the company itself). Eventually, Musk tweeted that because of companies scrapping Twitter for data (presumably for AI training) users would be severely limited in the amount of tweets they could see. The restrictions were supposed to be temporary, but are still in place as I type this.
This was terrible timing. It happened in the first day of the NHL free agent period. How was I supposed to know who overpaid for Alex Killorn or signed Ryan Reaves to be their tough guy from the press box in the playoffs? (Oh. Right. I could wait an hour for journalists to write up the articles.)
It didn’t take long, though, for people to notice remember that Twitter’s contract with Google— the one it did not pay and then did pay — ran out in June. And that there were at least some reports at the time that they needed to move some services off the could and that they were running behind. The problems started at the end of the month. It could be coincidence, but there are other indications that this is a problem of Twitter’s own making.
There is evidence that the twitter home page attempts to load data tens or hundreds of times a second for users who are not logged in, certainly causing load issues for the service. In other words, Musk introduced a “feature”, but his engineers were either not give the time to do it properly or were not capable of doing it properly. My money is on some combination of the two plus Twitter not having paid their hosting bill and thus being limited by their provider, but it certainly appears that the issues twitter is having are at least partially self-inflicted.
All of this was predictable and predicted. Musk gutted the staff of Twitter, cutting by close to eighty percent across all departments and treating the engineering staff with disdain and contempt. He thought that he could run a company without people, and it turns out that, well, you can’t.
This is not a Musk only delusion, however. The CEO of Reddit explicitly name-checked Musk during the user revolt over the API changes. That revolt shows an interesting wrinkle — all of Reddit’s value comes from its unpaid users generating content and its unpaid content moderators ensuring the site does not become a complete hellhole. When they revolted over Reddit removing tools that were better than the inhouse ones, they were at first relatively calm. But when traffic, and thus revenue, were impacted, they came down with a heavy fist and forced the rebels to open their subreddits back up.
Over and over, companies want to do their work without having to pay people to, well, do their work. AI is the ultimate expression of this. No one with an ounce of intelligence thinks AI can do most work, but the sincere hope is that it can do enough it that they can force employees to clean it up at a much lower pay scale, despite the fact that AI still depends on humans to actually generate any value. But Musk likely shows the problems with that for most companies.
There is a concept in tech, and likely most industries, called keeping the lights on. It refers to giving a product just enough resources to keep it up and running in its current state. You do this when improving the product doesn’t justify the cost, but winding it down cost too much in time, money and goodwill. However, you make this decision with the understanding that you will never improve the product and that if it breaks you likely won’t fix it quickly and you may not fix it at all. Musk kept just enough people around to keep Twitter’s lights on, and we are seeing the results now. And any company that attempts the same is likely going to eventually run into the same problem.
No matter how much we automate, human beings are still more flexible, still more intelligent, still best at dealing with unpredictability, still best at handling other people. People are still the most valuable economic drivers. Companies that forgot that are doomed to rate limit themselves into oblivion.