Reversion to the Mean

Tech companies are firing a bunch of people but in context it doesn’t seem like a large proportion of how many they hired in the past two to three years.

Hackers are making some terrible people unhappy, and I am here for it.

Chinese tech companies are growing again as regulators lift their boots off the sectors’ collective necks. That gives them enough room to look up at the swords dangling above them, I guess.

A bunch of people are suing generative AI companies to find out what copyright even means today. Or perhaps what it means in 2037 when the lawsuits finally get decided.

You can also marvel at, and learn how to use, fonts from the future-past.

My column this week is on reversion/regression to the mean and what it says about the human condition.

Things to note

Activist investor Elliott Management has purportedly invested in Salesforce. This usually means Elliott thinks it can bully the target company into doing its bidding so the share price goes up a bit and they can sell at a profit. One to watch.

Amazon is firing about 1% of its workforce, about 11,000 people.. Most of those to be shoved out are in retail and, ironically, HR. To put this in perspective, Amazon hired 310,000 people in just 2021.

Microsoft is also firing about 10,000 people it apparently hired by mistake, none of whom appear to be the managers who made the hiring mistakes. This comes less than two weeks after Microsoft made paid time off ‘unlimited’ i.e. no longer recorded as a liability on their balance sheet because you don’t get paid out leave owing when you get fired. Diedre Chambers, what a coincidence! But again, to provide some perspective, Microsoft hired 40,000 people in just the first half of 2022.

Google HR has entered the chat. Google is bringing the same copying-competitors-late-and-badly energy to its HR decisions and is firing 12,000 people. CEO Sundar Pichai says he takes “full responsibility for the decisions that led us here” which is why he’s not getting fired but all you peons are. I am unclear on what less-than-full responsibility would involve.

Gartner reckons global IT spending in 2022 dropped 0.2% instead of going up a predicted 0.8%. Gartner now reckons growth in 2023 will be 2.4%, not the 5.1% it predicted just three months ago. “Economic uncertainties” is how you say “fucked if I know” in polite enterprise analyst language.

ODIN Intelligence, a tech firm that makes apps for cops, got hacked and a bunch of data got leaked. Stuff like confidential police reports, detailed tactical plans for raids, AWS private keys, etc. Nothing major or important. Last year ODIN were trying to sell a facial surveillance system for identifying homeless people so forgive me if I have a bit of a chuckle at their misfortune.

In more What Could Possibly Go Wrong news, the 2019 US No Fly list was left online and unsecured.

Chinese regulators have apparently lifted a ban on taxi-company-with-an-app Didi registering new users. It’s been a year since the ban began.

The line is happy that Chinese authorities appear to be willing to let big tech firms like Tencent and Alibaba grow again. Tencent’s stock has nearly doubled since October 2022. To get volatility like that you usually have to invest in shady penny stocks, cryptocurrencies, or Tesla.

Chip exports from Taiwan grew again in 2022, up 18.4% YoY. This indicates that lots of people are still keen to torture sand into doing maths, so shortages are probably due to issues elsewhere in the supply chain.

Ant Group has partnered with a credit firm (“Buy Now Pay Later” is the fancy new term for this ancient concept) called Splitit to let Alipay users on Aliexpress in Europe pay for goods after delivery. It’s apparently to boost sales in Europe because sales in China are slowing, and it may well succeed, though I wonder what the default rates will be.

In inevitable lawsuit news, three artists are suing Stability AI and Midjourney for using their work without permission. It’s the same law firm that is suing Microsoft/GitHub over CoPilot, as covered in The Crux #17 and several other issues.

Getty Images is also planning to sue Stability AI for copyright violations. It’ll take years before these lawsuits get decided, so who knows what the world will look like by the time we find out if copyright exists any more or not. My money is on “yes, but only for big companies” same as it is now.

In news that will no-doubt shock you, it looks like Tesla lied about its self-driving capabilities. I wonder what the shorthand name for the Therac-25/Triangle Shirtwaist equivalent event for ‘self-driving’ cars will be?

I learned this week (via Mastodon, have you joined us yet?) that there is a physical book version of the website Typeset in the Future. The site is all about the typography and design used in classic Sci-Fi movies. Learn how to make your stuff look like it’s from the future!

Longer reads

It’s not long, but this quote from Daniel Kahneman about the human condition is worth reading.

“This was a joyous moment, in which I understood an important truth about the world: because we tend to reward others when they do well and punish them when they do badly, and because there is regression to the mean, it is part of the human condition that we are statistically punished for rewarding others and rewarded for punishing them.”

For a longer read, try this paper from 1956 by Herbert A. Simon Rational Choice and the Structure of the Environment.

It’s about how people make decisions here in the real world, as distinct from the theoretically perfect decisions of rational choice theory that exist only inside mathematical models.

He coined the term satisficing, which is basically about figuring out how much is ’enough’ to be happy with a decision, as distinct with trying to find an ‘optimal’ solution.

Weekly tip: Reversion to the mean

There has been lots of coverage of the layoffs in tech recently, but as I look into the details I’m unable to discern any particular reason for it other than maybe a bunch of firms deciding they want to and mostly copying what everyone else is doing. It could be just “we panicked a bit and over-hired, so now we’re firing the people we shouldn’t have hired in the first place” which is just reversion to the mean.

It doesn’t seem to be reflecting any wider trend of a rise in unemployment. Not in the broader labour market, and not in the tech sector itself, either. A quick look over the data from the US Bureau of Labor Statistics non-farm payrolls data by industry and industry sector doesn’t seem to show a sector-wide shrinking in November-December. There’s a 6.4% reduction in General merchandise stores, including warehouse clubs and supercenters that might cover Amazon’s warehouses, which is about 7,000 people altogether.

It means that, yes, a few tech firms may well be firing a bunch of people, but most of them will go and get other jobs. Most of them in other tech firms. The net loss to the sector doesn’t appear to be that large.

The firings also look large in comparison to the relatively high number of staff because of massive hirings over the past couple of years. If these big companies had hired (and fired) at their longer-term average rate in the past couple of years, the news would now be full of articles about the thousands of people being hired at big tech firms. Or there’d be no news at all, because it was just normal. This is news because it looks different to what’s been happening recently.

Overall, I’m sceptical of any attempt to justify or reason about what’s happening. I don’t think there is any reason behind it, not really. It’s just some stuff that’s happening. The reason, if you want to call it that, is statistics. It’s not personal, though it feels like it when you get fired. My sympathies are with those who got caught up in it.

It’s nothing to do with their skill or abilities or enthusiasm or anything like that. It’s just some executive decided they needed to adjust a number in a spreadsheet. The executive doesn’t even really know why the number needs to be different, they just tell themselves stories about how The Line feels today to justify their actions.

The stories about why are just that: stories. They’re post-factual justifications for things that people decided to do, for reasons they don’t really understand and don’t want to look too closely at lest they discover just how little control they have over their own lives most of the time. It’s what we humans do.

Here’s another story you might like to consider: the tech companies are firing a lot of people now because they chose not to fire as many people as they normally would have, because a brand new and scary pandemic was happening, and people would have lost their healthcare. Sure, it sucks that they lose their job now, but maybe it would have been worse if they’d lost their job 12-24 months ago?

That story is just as likely to be true, given the data I’ve seen so far. It sounds nicer, so it’d be useful for making you feel a particular way about what’s happening. Who is telling the stories here and how do they want you to feel? That, I think, is the more interesting question.

And then we could ask ourselves: what evidence would we need to find to prove our story is the wrong one? Is it even possible to have a single, true story?

Maybe what’s going on is messy and there are multiple true stories about it.

Maybe sometimes stuff just happens for no reason, because the universe is mostly a cold, empty, and uncaring place. Maybe the stories are more important than the reasons, because the stories are what make us human.

Well that and ruining everything since forever.