In the Brothers Karamazov, The Grand Inquisitor was irritated that Jesus did not give people what they want. Jesus annoyingly insists that he loves them and thus cannot give them what they want, but instead something more. I think this is a similar issue, though I often read the Grand Inquisitor chapter as an argument against the sort of optimization-minded EA folks (and find it compelling)
The problem with this take is that good and bad are subjective measures. And since we don't all have the Yglesias Values Engine installed, it's not practical nor feasible to enshrine universal values from the bottom-up.
***The Catholic Church has entered the comments section...***
Compared to a history book, Slow Boring is a bag of Doritos – or maybe, in a fairer appraisal of its quality, those chips that come to your table at a great Mexican restaurant, fresh and hot from the kitchen. When I was an academic, my book reading was easily an order of magnitude larger than my journalistic reading. But like many others, I now spend too much time reading about current events in newspapers, magazines, and substacks. And the day still has only 24 hours. In 2022, I'm having only partial success going back to my old ways, but 2023 will be better.
I could have spent so much time today explaining in this comment thread how useful Facebook is, but I was too busy preparing to start a job which I might not have if it weren't for Facebook keeping me connected with old professional acquaintances, so whatever.
A more correct statement would be "...[social media]’s just showing people the subset of what *the platform owners allow people to post and what the platform owners think is interesting to people*, and if the people are interested in garbage, that’s the people's fault."
What people see is already moderated by what people actually post, followed by what's allowed to get in the door, followed by what the platform chooses to show you at any given time. Users do not have an opportunity to choose "healthier" options, only to avoid the place altogether.
At least in a grocery store, you have clear options (even if you are being manipulated before, during, and after your shopping experience).
Social media platforms could easily do a better job of filtering out "bad" content. They easily filter content when they have to or choose to.
Social media is cigarettes for the mind. Assuming we survive, we will look back on this time and wonder how we allowed children access to it, and how it took us so long to understand the danger.
Social media has also infected traditional media, which now has to resort to clickbait headlines, comments sections, and other horrible compromises.
I sometimes tell people, "be the boss of your phone, don't let it be the boss of you". It's hard though, and I struggle with it myself. So many apps are designed to be addictive, and they are so good it at that all you have to do is try it and you're hooked within minutes. Turning off unnecessary notifications helps, but it doesn't fix everything.
> So far nothing catastrophic has happened because of Meta’s recommendation engines.
Maybe true if you're in the US, but there are almost a million Rohinyga refugees who lost their homes (and thousands who were brutally murdered) due to ethnic cleansing in large part incited by Facebook. https://www.nytimes.com/2018/11/06/technology/myanmar-facebook.html
Facebook's algorithms gave the perpetrators "what they wanted" - a call to genocide, secretly driven by the military.
This ties into Milan’s Sunday Thread on declining Oscar viewership. The public wants its comic book movies and doesn’t care which indie film wins an award.
This touches on an idea I've wanted to write up for some time. I think the notion of 'preference' as a single, unitary thing has been particularly pernicious, and is demonstrably false. We do not have a single, discernible preference for one thing over another in all circumstances, and that preference is very certainly not revealed by our behavior.
This becomes obvious with a straightforward thought experiment: Consider the obese person who would like to be thin. They are presented with a cookie. They claim they want to be thin, but they eat the cookie anyway. The naive economist says "Well, their revealed preference is to eat the cookie, so obviously they prefer consumption to thinness".
However, what if there existed a pill that would disable the craving for the cookie. This very same person would probably take that pill and *choose* to switch off their craving for the cookie, and then make a different choice the next time they were in that situation. What this reveals is that this isn't an issue of *preference*, at least not in the naive sense. It is an issue of choice architecture, in this case mediated by the available technology.
Given the current state of cookie-desire technology, cookie-related choices have to be made on an individual basis, in the moment. And choices made on an individual basis, in the moment, give more strength to the "lower" (metaphorically, and maybe also literally) parts of our brains. However, that very same person's "higher" executive decision making apparatus may very well make the exact opposite choice.
I think the question we need to start asking of technology: digital, pharmaceutical, and otherwise is: How can we create a world in which we can shift more of our everyday decision making to our executive apparatus and away from our lizard brain? This is a pathway to a better world, for everyone. And the first step on that path, I think, is understanding that preference is not one thing, and it is definitely not something that can be simplistically revealed by behavior.
For whatever an outsider’s point of view is worth: social media didn’t drive you all insane. Social media on your phones drove you all insane. You were all fine when you had to go home and boot up your computers to fuck around on Facebook. I don’t know that there’s a way to improve your lives by improving the algorithm unless it also takes the phones out of your pockets.
I think the food-scientist analogy is a good one, but for the opposite reasons that Matt seems to think it is. It's obvious to me that a food scientist at Impossible Foods is doing more "good" for the world than one at Frito-Lay, but it's very very revealing to me that Matt would never bother writing multiple columns going out of his way to shame the people working at Frito-Lay. It's just not something that would ever cross his mind to do. For whatever reason software engineers are required to have this moral responsibility that Matt would never seriously apply to other professions.
Worth noting here the role Facebook algorithms played in the anti-Rohingya ethnic cleansing in Myanmar - AFAICT there seems to be pretty broad agreement that viral content on Facebook helped incite people to participate in anti-Rohingya violence. I'm sure it's true that lurid anti-Rohingya Facebook posts got lots of clicks and engagement, and the algorithm rewarded them accordingly. But this goes to show that optimizing for lots of clicks can be bad!
I would love to hear you and Ben do a podcast
This whole thing feels pretty silly when the average American spends 5 hours a day watching TV.
In the Brothers Karamazov, The Grand Inquisitor was irritated that Jesus did not give people what they want. Jesus annoyingly insists that he loves them and thus cannot give them what they want, but instead something more. I think this is a similar issue, though I often read the Grand Inquisitor chapter as an argument against the sort of optimization-minded EA folks (and find it compelling)
The problem with this take is that good and bad are subjective measures. And since we don't all have the Yglesias Values Engine installed, it's not practical nor feasible to enshrine universal values from the bottom-up.
***The Catholic Church has entered the comments section...***
Compared to a history book, Slow Boring is a bag of Doritos – or maybe, in a fairer appraisal of its quality, those chips that come to your table at a great Mexican restaurant, fresh and hot from the kitchen. When I was an academic, my book reading was easily an order of magnitude larger than my journalistic reading. But like many others, I now spend too much time reading about current events in newspapers, magazines, and substacks. And the day still has only 24 hours. In 2022, I'm having only partial success going back to my old ways, but 2023 will be better.
I could have spent so much time today explaining in this comment thread how useful Facebook is, but I was too busy preparing to start a job which I might not have if it weren't for Facebook keeping me connected with old professional acquaintances, so whatever.
You should make fun of Coogler for working on Space Jam: A New Legacy. Much worse than Black Panther
A more correct statement would be "...[social media]’s just showing people the subset of what *the platform owners allow people to post and what the platform owners think is interesting to people*, and if the people are interested in garbage, that’s the people's fault."
What people see is already moderated by what people actually post, followed by what's allowed to get in the door, followed by what the platform chooses to show you at any given time. Users do not have an opportunity to choose "healthier" options, only to avoid the place altogether.
At least in a grocery store, you have clear options (even if you are being manipulated before, during, and after your shopping experience).
Social media platforms could easily do a better job of filtering out "bad" content. They easily filter content when they have to or choose to.
Social media is cigarettes for the mind. Assuming we survive, we will look back on this time and wonder how we allowed children access to it, and how it took us so long to understand the danger.
Social media has also infected traditional media, which now has to resort to clickbait headlines, comments sections, and other horrible compromises.
I sometimes tell people, "be the boss of your phone, don't let it be the boss of you". It's hard though, and I struggle with it myself. So many apps are designed to be addictive, and they are so good it at that all you have to do is try it and you're hooked within minutes. Turning off unnecessary notifications helps, but it doesn't fix everything.
> So far nothing catastrophic has happened because of Meta’s recommendation engines.
Maybe true if you're in the US, but there are almost a million Rohinyga refugees who lost their homes (and thousands who were brutally murdered) due to ethnic cleansing in large part incited by Facebook. https://www.nytimes.com/2018/11/06/technology/myanmar-facebook.html
Facebook's algorithms gave the perpetrators "what they wanted" - a call to genocide, secretly driven by the military.
This ties into Milan’s Sunday Thread on declining Oscar viewership. The public wants its comic book movies and doesn’t care which indie film wins an award.
This touches on an idea I've wanted to write up for some time. I think the notion of 'preference' as a single, unitary thing has been particularly pernicious, and is demonstrably false. We do not have a single, discernible preference for one thing over another in all circumstances, and that preference is very certainly not revealed by our behavior.
This becomes obvious with a straightforward thought experiment: Consider the obese person who would like to be thin. They are presented with a cookie. They claim they want to be thin, but they eat the cookie anyway. The naive economist says "Well, their revealed preference is to eat the cookie, so obviously they prefer consumption to thinness".
However, what if there existed a pill that would disable the craving for the cookie. This very same person would probably take that pill and *choose* to switch off their craving for the cookie, and then make a different choice the next time they were in that situation. What this reveals is that this isn't an issue of *preference*, at least not in the naive sense. It is an issue of choice architecture, in this case mediated by the available technology.
Given the current state of cookie-desire technology, cookie-related choices have to be made on an individual basis, in the moment. And choices made on an individual basis, in the moment, give more strength to the "lower" (metaphorically, and maybe also literally) parts of our brains. However, that very same person's "higher" executive decision making apparatus may very well make the exact opposite choice.
I think the question we need to start asking of technology: digital, pharmaceutical, and otherwise is: How can we create a world in which we can shift more of our everyday decision making to our executive apparatus and away from our lizard brain? This is a pathway to a better world, for everyone. And the first step on that path, I think, is understanding that preference is not one thing, and it is definitely not something that can be simplistically revealed by behavior.
For whatever an outsider’s point of view is worth: social media didn’t drive you all insane. Social media on your phones drove you all insane. You were all fine when you had to go home and boot up your computers to fuck around on Facebook. I don’t know that there’s a way to improve your lives by improving the algorithm unless it also takes the phones out of your pockets.
I think the food-scientist analogy is a good one, but for the opposite reasons that Matt seems to think it is. It's obvious to me that a food scientist at Impossible Foods is doing more "good" for the world than one at Frito-Lay, but it's very very revealing to me that Matt would never bother writing multiple columns going out of his way to shame the people working at Frito-Lay. It's just not something that would ever cross his mind to do. For whatever reason software engineers are required to have this moral responsibility that Matt would never seriously apply to other professions.
At least we can count on Joe Rogan to hold them accountable.....
Worth noting here the role Facebook algorithms played in the anti-Rohingya ethnic cleansing in Myanmar - AFAICT there seems to be pretty broad agreement that viral content on Facebook helped incite people to participate in anti-Rohingya violence. I'm sure it's true that lurid anti-Rohingya Facebook posts got lots of clicks and engagement, and the algorithm rewarded them accordingly. But this goes to show that optimizing for lots of clicks can be bad!