Mini Posts Collection 2 - Nov 2023
"Nothing makes sense and that's okay", "Mourning Facebook Past", "Why Tech Companies have so many Employees", "How Music Became Meaningless", and "Accepting Pain".
This is my second in a series of “mini-post” posts. I use these to throw out multiple ideas that I’m not sure I have enough to say about to justify a full-length post (or justify sending you an email for just that thing). Some of these might eventually become full length posts.
Nothing Makes Sense and That’s Okay
We like the world to make sense, but sometimes it doesn’t. And that's okay.
There is value in having theories about how the world works. A good scientific theory helps one make predictions about what actions might lead to positive actions, and avoid the need to flail around at random. A good social theory can help a group of people come up with rules that help them cooperate and work towards a common goal.
But bad things happen if we care so much about wanting to understand the world that we try to fit a complex world into too simple a theory. We don’t want to follow the same social rules in all situations. We don’t want to write all code the same way. We don’t want all government policies to be governed by the same principles.
The awkward truth is that the world is too complex for a comprehensible theory to explain the right thing to do in all situations. A good theory helps us find things that might work, but in the end “do what works” should always trump “do what makes sense”.
Mourning Facebook Past
I used to love Facebook, and now I don’t.
When Facebook was fresh and new, the only people I was connected to were friends at my college. They were people I felt comfortable saying dumb things in front of, and so I did.
When Facebook was young, the prompt was “Rob is …” and so people would post about what they were doing or how they were feeling. This often revealed opportunities for me to help them or hang out with them, and so I did.
When Facebook was good, it had little support for article links, so it was rare for people to virtue signal by sharing links to angry political articles. There were also no “memes”, or “long opinion posts”, or even really “posts” at all, because status messages were visually minimal and felt more like messages.
When Facebook was cool, my friends logged in every day, and so I could reliably use it as a way of seeing which of my friends were interested in something.
I still use Facebook every few months, mostly to participate in one of the groups, or to look at a post someone has told me about, but it isn’t a place I want to hang out in any more.
When I worked at Facebook we tried an experiment to make Facebook more like it used to be, with the feed focused on posts about people’s lives, and article links moved to a separate tab. I thought it worked great, but the press said it was “Downright Orwellian”, so we turned it off.
I feel sad because there isn’t anything today that does what Facebook used to do well. Maybe sometimes someone will create one. Maybe it’s something that can only really work when you are at college, and I’m too old now to experience it again. Or maybe there is a reason something like that can never exist again.
Why Tech Companies have So Many Employees
Why does Meta have 84 thousand employees, when Mark Zuckerberg was able to write the first version of Facebook entirely by himself? Similarly, WhatsApp and Instagram had only 55 and 12 employees respectively at the point when Meta aquired them. If you look at the early versions of most successful tech products, they had most of the important functionality of the current versions but were built by a tiny fraction of the current employees. So what’s going on with the big head counts?
The answer is that if your product makes a lot of money, it’s worth hiring extra people to make it better, even if each person only makes it a tiny bit better. Maybe one person improved the support for some use case that only 0.01% of people care about. Maybe one person made it work better on a hardware platform only 0.01% of people use. Maybe one person jumped through the hoops to comply with the laws of a country with 0.01% of users. Maybe one person wrote code to support the needs of a niche advertiser who buys 0.01% of ads. Maybe one person improved the feed ranking algorithm in a way that increases user engagement by 0.01%. Maybe one person improved performance in a way that reduced server costs by 0.01%.
If your product doesn’t make much money then hiring someone to make your product 0.01% better is a waste of effort, but if your company makes $100bn of revenue (as Meta does), then increasing revenue by 0.01% means increasing revenue by $10m.
The downside of this is of course that if profit goes down, then a lot of those employees may no longer justify their salaries, even if the company is still profitable, and the employees are still improving the product by the same amount. This is part of the reason why tech company employment is so subject to economic fluctuations.
How Music Became Meaningless
Music used to have a very different social role to the one it does today. You’d buy albums one at a time, and curate a set of albums that reflected the way you thought about the world. When a friend came to visit, they would browse through your CD collection, and you’d play them some songs that you knew well and that really resonated with you. If an album no longer felt like “you” you’d trade it in at the used record store.
The best music was edgy. It said the things you believed about the world, accompanied by tunes that paired the words with the emotions you were supposed to feel while hearing them. You wanted these things to be edgy, because they communicated the things that you felt that made your way of thinking special. The best songs were essentially like hymns, but hymns to your own personal religion.
Nowadays people mostly subscribe to music, with services like Pandora or Amazon music. Rather than listen to a small, carefully selected set of albums, you listen to whatever the algorithm wants you to listen to. The advantage of it is that you get access to a much larger selection of music than if you had to buy it album by album. The downside is that it all becomes totally meaningless.
The music I listen to today doesn’t really say anything about what I believe or how I feel. It’s more like elevator music. Pleasant tunes to play in the background while I do something else. It’s entertaining, but it’s not meaningful.
Or maybe I’m just an old fogey and music is only really meaningful when you are young….
Accepting Pain
One of the perks of having been a lightweight crew rower in college is that I learned to be quite good at losing weight. I don’t like losing weight. I much prefer to eat as much as I like and gradually put on weight, but if I need to lose weight, I’m generally able to do it.
The way I would lose weight is to notice the physical discomfort of hunger, and say to myself “That’s hunger. That means you are losing weight. That’s a sensation you can ignore, or even enjoy a little”. It wasn’t easy, and I have to go through the process again any time I want to lose weight, but it generally works.
As an athlete you have to learn to have a similar attitude to physical pain or tiredness. The difference between an athlete is that when a normal person feels tired, they think “I’m tired, so I should stop”, but an athlete thinks “I’m tired. Let’s see if I can make myself pass out”.
I’ve had similar experiences with other sensations. I used to not like eating spicy food because I didn’t like the discomfort of the spicy flavor in my mouth, and so I would try to make it go away with water, or beer or yogurt, which rarely worked. But gradually I learned to tell myself “That’s spiciness. It’s okay for your body to feel like that”.
Sometimes this can be useful, but sometimes it’s a terrible idea. As a rower I inflicted long term damage to my back because I had taught myself to ignore pain. Some people who learn to enjoy hunger become anorexic, and do significant damage to themselves. Often when your body is telling you shouldn’t be eating something your body is right.
There is some truth to the idea that “pain is just information”, but sometimes that information is worth paying attention to.
On the large number of tech company employees, I am sure you understand, but maybe your readers don't, how stochastic those incremental improvements are and how that can make most employees at any one time look totally useless. An employee may do nothing value-added 9 years out of 10 but then make a great breakthrough in the 10th
Or 9 out of 10 hires who look great up front may end up adding little value but the 10th is a superstar, and there is no way to tell which one it will be before you let them all work for the company for 5 or more years. The gamble is similar to VC funding in that way-- the 0.01% improvements you mention are an average over a very spiky distribution.
My personal experience leads me to emphasize this point: in 15 years at Google, a large majority of all value I created for the company was in one six-month project. Nobody could have predicted in advance what that project would be, or whether I would have another hit like that (indeed I left in part because I sensed I wasn't on track to ever have another such hit). I "just happened" to be in the right place at the right time; and tech company human capital allocation is largely a process of trying, with very high failure rates, to put people in those right places.