Transcript for #363 — Knowledge Work

Sam, it's always a pleasure to talk with you.

I mean, you're welcome. I enjoyed and found fascinating the reaction. to you leaving Twitter, I don't know if this is how you experienced it, but to me, it's what I imagine it's like when your buddy at the bar stops drinking. Because people got mad that you left, I can't imagine being mad about someone stopping doing something. But people is as if they took it personally as far as I could tell, what do you mean you're leaving Twitter? What's wrong? I thought that was as instructive as hearing about what your experience has been like after you left.

Well, you know, I did some writing about this more recently. And I had to remind a little bit when I was thinking about this. I wrote this New Yorker essay, late last summer, early fall. And one of the big questions I had in there is, why did we come to believe that the right way to use the internet to have discussions, the sort of surface ideas, the spread news? Why did we think the right way to do this was to try to get everyone to use the same global platform? And this is, we take this for granted as like a course, this is what you should do. We should have everyone use the same global conversation platform because if we're all on the platform, we all can see each other. But of course, the reality and I sort of laid this out of the article is when you have 500 million tweets being generated per day. And the average person is going to see 100 in their feed. What you have to do is just incredibly aggressive curation, right? Because there's just you can't have 500 million people using the same platform and yet have a sort of globalized centralized zeitgeist defeat where you're trying to surface a small number of trends for everyone to see this requires an incredible curation this sort of amazing cybernetic part algorithm part human part network theory powered curation that Twitter uses and is of course that incredibly aggressive curation that makes this a platform that just fuels outrage that fuels the darker side of people and one of the arguments I made in that article is here's what works better small communities that have weak-type connections between them. You have lots of small communities online and they have overlapping membership. We know from network theory, really interesting ideas can spread to these network of networks. Really important news will spread to these network of networks, but most of your interaction in this sort of digital vision is going to be with a small number of other people that have a sort of emergent shared sort of community standard. It's much richer, much more personal. That's really what the internet envisioned. Everyone is going to have the possibility to be connected to everyone else, not everyone actually needs to be directly connected to everyone else on the same platform. So I've really been thinking about the folly of global conversation platform as one of the key folly of the sort of 2010s internet.

I still feel strong about that hypothesis. I mean, I think we see for various reasons, but we see the deep thrown in of Twitter as having that same central cultural status it had before. TikTok, we see, for example, like my argument was about TikTok, is without this entrenched advantage of my social graph is in there. I've already spent years trying to set up follower networks that I said the connection the TikTok would be very superficial and we can we are seeing that I think last year for example there was double digit drops and TikTok users among the sort of 20 and especially the upper 20s and in the 30s that sort of demographic is sort of young adult demographics at enough. And they really had no problem leaving because what's actually connecting there. It's a very zeitgeistly platform that people can take or leave. I mean, Instagram is holding it there, but I don't think it has that same against centrality that it might have had two or three years ago. And the alternatives independent media, so podcasting, email newsletters, these really are ascendant in the last year, year and a half since we've talked. And that is the opposite of the centralized platform model. because this is an independently produced information. People discover these things almost entirely through point to point curated trust. Someone I know, Ford, me and email newsletter, I signed up. Someone I know told me about this podcast. I started listening to this podcast. They mentioned another podcast. Now I listen to that one. That distributed trust model of curation as opposed to a centralized algorithm model. I think really works well and we're seeing that, right? So I remain bearish on the idea of a small number of social platforms that dominate internet culture.

Yeah. Do you find it surprising the number, I'm thinking journalists in particular who very much dislike Elon Musk, right? So they have a sort of moral personal ethical commitment to stop using his platform and they'll still talk about it. I don't like what's happening on X. I don't like Elon Musk and yet almost none of them have left. I find that interesting. I think there's something interesting in that where it would for a lot of people who are writing like a tech journalist, business journalist, really dislike, clearly dislike Elon Musk. Still can't bring themselves to leaving the platform. which, again, I think speaks to something interesting about the way the platform plays, especially people with some sort of public profile, the way it plays and how they understand their influence on the world or their impact. But that's to me, that's been the more interesting observation of the post-Musk period is actually how rare you still are, which is, you know, I thought to be a lot more Sam Harris's, a lot more people's, and yeah, I'm just leaving. People are having a hard time.

Yeah. Well, I mean, I think this is one of the more engaging human psychological experiences is this idea of most people were thinking this, but then me and my group figured out that that is true, right? That inversion, the inversion of the whatever, the empiricism structure is incredibly engaging. And it occurs every once in a while in reality. Like everyone was thinking this and then we realized whatever DNA is a double helix. And it's like amazing. Online culture, especially algorithmic driven online culture, has given a way to basically commodify that and spread that. Like we can create, you can build an entire epistemology around everyone thinks this is true, but it's really that. Everyone thinks this is bad for you, it's good. Everyone thinks like this about this disease, but it's that. Everyone thinks that, and you can build an entire epistemology around that. Like your entire world can be that, and there can be a whole audience that's just going to reward that. The algorithm is going to reward that. So I have definitely seen that as well. There's a well-known effect among professors. called the sort of academic and exile or academic and the wilds effect, which is if you if you take an academic and then they leave they leave academia you know they go they go independent eight times out of ten They go to some really conspiratorial places. And partially what's going on here is, well, first of all, they're smart. So, like, it completely makes sense to them that I could figure something out that other people didn't understand. Because I'm very smart. But one of the purposes and services academia, plays as there's this checking mechanism. Everyone else is smart too. And so when you're like, hey, I think, look, here's this whole new way of seeing it. The Earth is hollow. You have all these other smart people being like, here's why you're dumb. And they take you down, right? But when you leave, you have this academic and exile. And this is sort of like Linus Pauline with vitamin mega dose. It's much, much higher rate now in the age of social media. Because now when you leave, you can immediately algorithmically have constructed an audience that cheers you on. And so now I think the severity of academic and exile effect is much more pronounced and much more ubiquitous than it used to be that when you sort of leave academia to start your podcast, it's not too long until there's world changing conspiracies that you're uncovering and it could be medical and it could be governmental and it doesn't matter and so I think that effect is like one of the more interesting effects has been happening is that you can get a cheering section And the algorithmically constructed sharing section of people that's you're being rewarded. For a saying, I think this is the way it really happens. And you see this all the time, I think, I like podcasts to stand as a term. This idea of there can be a hundred studies saying something. But if there's one study saying something different, the way you perceive that is, well, everyone knows now that things not true. It's this interesting sort of sampling of evidence, this sort of destabilization of Bayesian priors, that is amplified and supported in sort of the world of algorithmically discriminated or algorithmically disseminated information.

Yeah. Well, I think the town square is a town square analogy that's causing the problem here. Right. The concept of a town square, the sort of central gathering place where people can democratically discuss, depends on scale being reasonable. Right? I mean, we call it the town square. We don't call it the city or the state square, right? Because it's a place where the demos and Athens, we sort of a relatively constrained group of people who all know each other and have other ties to each other. You have other social trust ties to each other. We live in the same town. I run the hardware store that you come to to buy your nails, them to come together. There's this free speech notion of what we don't want to buy fiat in advance, say, here are topics that are off limits, because how are we going to work together to advance what we understand? Twitter is not a town square, right? When you have 500 million users, that's not a town square. It's an entertainment product. We have 500 million users who are inputting lots of different possible bits of content that could be interesting. We're going to run them through this cybernetic curation algorithms. This is what I mean by that is there's algorithms involved, but it's also the expansion properties of the underlying follower graph means that individual decisions to retweet or not retweet. So these are human decisions interacting with these digital networks can create these cascades of information spreading. That's a lot about how trends arise. It's a really powerful, actual curation mechanism. It's unlike TikTok, which is purely algorithmic on Twitter, it's cybernetic. And so you have these digital networks with good expansion properties. And 100 million people making individual decisions with enough to click Retweet or not, whether it is a quote tweet or not, which is a whole interesting sort of computer science question. All of this aimed towards how can we take this giant pool of potential content and choose to sort of small number of streams of content that are going to be relatively globalized or interesting and engaging? That's not a town square. It's an entertainment product. It's why, in a New Yorker piece, I wrote right after Musk took over Twitter, as I said, it's not the town square as much more of the Coliseum. That's much more of the better idea. It's tons of people watching carefully curated entertainment. In that context, of course you have all sorts of thumbs on the scale. Like the whole point is we're trying to put on this show and I think of like the trending topics of the day as the show on the Coliseum floor with all the huge crowds is watching and chiming in to see the blood sport between hey Sam today is having a war with you know whoever right this is the entertainment for today it's an entertainment product of course the thumb is on the scale because you're trying to find something that like just pushes the buttons right maybe there's some outrage but not too much Maybe it's absurd, but not in a eight-chan, like completely over the top. Lowell's type absurd. You're trying to program a television station. You're trying to program entertainment in a Coliseum. It's not a town square. Now, if you have actual digital town squares, like here's a place where like a small number of people who actually other ties to each other are gathering to think things through and talk things through, I mean, we've seen examples of those can have a wide variety of different community standards, including standards where like almost anything is going to go on this discussion group, but it works because we're all whatever. lumberjacks from this part of the country, and we sort of have other ties to each other. And so I think it's that town square metaphor that threw us off. As we took this entertainment product and somehow tried to make it seem like this was the Roman Senate. Like this is where just like this reasonable, reasonable scaled group of people were getting together to hash things out. And that's never what they were actually trying to do there.

What ends up happening on Twitter is that more local interactions just get swamped out by the the ultra amplified content right and increasingly their feed is driven by this so there's some stuff in there from people you straight follow but the feed is algorithmically sorted and one of the major criteria on which things are sorted for your feed is their engagement across the network so so really what's happened is it's that it's the content that has really gained this big boosting effect, this sort of cascade of retweets leading the retweets. That's what's being programmed for. And there's some conceptual regionalization. So if you follow a lot of a certain type of people, you might be seeing what's really being amplified in that subgroup. That's true. There's some of that going on as well. But it's a really large scale at which a lot of this is happening. And especially when it comes to the most-town-squarey piece of this, which is discussions of politics, discussions of policy, discussions of world events, the sort of the stuff we think of as the grist of civic discussion, those are incredibly large sub-networks in which information is spreading. And that's where you really have the policy in effect. It's people competing to get their turn on the Coliseum floor, they have their trident ready. And that's like that most civic-minded aspect that we associate with Twitter, I think is the most entertainment center.

Yeah. No, and I'm with you on this, right? It does that. It does that. And you're auditioning. You're auditioning when you comment on someone's tweet. You're auditioning for the algorithm. There's this sort of cybernetic amplification effect. So everyone tries to one up each other. I mean, I hear a lot from people, you know, what they're really, they're worried about mainly is not the contents from the other side, but that the in-group policing, I think it's had a massive impact. we see it, you see it in journalism, you see it in academia, you see it in sort of theoretical frameworks, you certainly see it, the some degree in politics, those interesting politicians, they're so used to that. It's almost as if they, they're the one group that at least sort of understands the social dynamics of something like Twitter, like this is sort of their lives is, you know, In group policing, out group contempt, being okay with like these people are upset, who am I going to hit my wagon to putting their finger to the wind? But for most other people aren't used to that. And so it's a it's definitely an effect. It makes you either power or makes you conform. Look, I only get some taste of this like when I'm doing book tours, right? Because I'm not on social media. So I'm not subject to people talking about me and in these sort of contexts when you have a book out you people do talk you get reviews and people come out to talk about your book or whatever and I hate it and I couldn't imagine if that was just all the time but if like all the time that was the world I was in is like every week I have to get like the two or three people that are like taking their swing at me or whatever I mean that would drastically affect I'm sure what I write about how I went through my life, just like the subjective well-being. Yes, I really, I was really hoping for Elon to destabilize Twitter so much that it essentially collapsed. It would actually be like a great civic duty that he would have done. But it seems to be holding on to some degree, at least as far as I can tell, which is I think unfortunate.

Well, I agree with your characterization of the questions I think about. I would add something else to it, which is, as a computer scientist, as a digital theorist, I care in particular how technology intersects with that story as well. So in most of my writing of the last decade, There's usually an unintended consequence of a technological development that gets us out of touch or becomes an obstacle to living some life that's going to be deeper, more meaningful, and we have to grapple with that and understand that technology, the opportunities, the parallels, and sort of navigate around it. So even this book, like slow productivity, the problem that I'm solving, there's actually a techno story behind it. I mean, so there's a, an easier way of summarizing it, which is like, there's a key question that a lot of people are in mainly, as you say, the sort of knowledge work world. These are people who are doing pretty well, right? You have a job in which you look at a computer screen in your an air conditioner, right? So a lucky place to be. But that group of people, and these are the people, these are big group of my readers. A big question a lot of them have is, okay, how do I do my work? Well, how do I produce things on proud of, have impact, you know, support my family, and yet not let work just take over all of my life. And how do I avoid sort of falling in to burn out? I don't want to be like pre-Twitter Elon Musk and just don't sleep and have seven businesses and just get after it. Like how do I still do stuff I'm proud of but also spend time with my kids right there's this big question and in the book looks at that question from both the perspective of some like creatives that have a huge amount of autonomy and also from the perspective of sort of a standard office worker who has less autonomy how can they still sort of navigate that that knife sense but there's a techno story behind how we got to a place where that question became more and more relevant and and the story that I'm trying to the tell in the first part of the book is a knowledge work itself when that emerged as a major sector, which is really like mid-20th century. Have this issue of not really knowing how to define productivity. Because we had industrial productivity was a quantitative concept. It's a ratio, it's model T's per paid labor hour. Agricultural productivity is a quantitative idea. It's bushels a corn per acre of land. You could measure this. Industrial manufacturing agriculture, you had well-defined production systems. So you could tweak something very specific and see how it impacted that number. Knowledge work comes along under that works. Knowledge work is more have hazard and autonomous and ambiguous. I might be working on seven things that are different than the eight things you're working on. There's not one thing we're producing. There's no well-defined production systems we use for our work as well. Organizing labor is very independent and individualized in knowledge work. And so in response to that, the knowledge sector came up with this idea of we will use visible activity as a proxy. for you doing something useful. So we'll all gather in the same building like we would a factory will work factory shift hours. And like if I see you here doing stuff, I'm assuming that's useful stuff. And if we need to be more productive come early, stay late, right? I call that pseudo productivity. The techno story is that worked okay. Not great, but worked okay until the front office IT revolution of the late 90s and early 2000s. And then once we threw network computers and then later mobile computing into the sort of office works fear, pseudo productivity spun off the rails, right? Because the personal computer came in and now suddenly the amount of different things you could work on. You know, quadrupled. No longer do we have specialists, the type, and specialists to handle communication. Like the amount of work you could do quadrupled, low friction communication networks made it really easy to ask people to do things. So workload skyrocketed, email, chat really changed the game when it came to demonstrating visible effort. Now you could be doing this in an incredible fine granularity at a frenetic pace. How quickly I respond to a message might be really important signal in trying to show how pseudo-productive I actually am. So my argument was The front office IT revolution plus this older idea of suitor productivity they didn't mix and it led to this increasing exhaustion of knowledge workers as the amount of stuff they're working on increased the amount of their day dedicated to talking about work instead of actually doing work increased. the freneticism and speed of their work increased. At the same time, they're getting more nihilistic. Like, what am I actually doing here? I'm just doing all these visible signs of productivity. I'm sending emails and meetings, but I'm not actually writing the marketing report. I'm not actually programming the computer. And it led to the burnout epidemic. And so there's that underlying techno story of technology plus that crude metric didn't work well. So we have to reassess What does it mean to produce really good stuff? Can we do that in a way that doesn't make work really exhausting? So I have that sort of personal angle. And then there's also this sort of deeper techno social economic story. And they're both circling, I think, the same issues.

I think the beginning of COVID pushed this increasing issue people were having with the unstainability of suitor productivity, it pushed it over the edge. Because a couple things happen when knowledge workers had to go remote right away. When we were shifting remote, we were already for the most part at our max capacity for the amount of work that we could have on our plate at the same time and have any chance of not drowning. And the reason why we're at max capacity is because in knowledge work, we leave workload management up to the individual. For the most part, that's up to you. The figure out how to manage what you're doing and what you say yes to. There's a lot of autonomy and ambiguity in knowledge work. So how a lot of people began to manage their workloads is they would wait until the stress of their workload got high enough that it out, outweighed their concern about the negative social costs of saying no to new Like this became the primary governor mechanism by which so many knowledge workers managed to work loads. So of course, this keeps you at a state of having a stressful, stressfully large amount of work. If you have to be stressed before you begin to say no, everyone has a stressful amount of work. Then the pandemic hit, which automatically gave us like 25% more tasks overnight because we had to adjust unexpectedly to, oh, we got to run our company the web down your groceries. Yeah, and just work itself, like how do we deliver our services, you know, if I remember this going on at the New Yorker, like how do we move our whole production process to like a digital pipeline? It was just new work came out of nowhere. The collaboration then also became less efficient because we lost all of the, oh, I see your office door is open. So I'll book my head in. And like, hey, what are we going to do about this client in a text two minutes? Instead, we had to start scheduling Zoom meetings, but the smallest granularity of these meetings was a half hour because it's hard to drag anything smaller on your calendar. And so now we were making the collaboration, the overhead related to work, that became a lot less efficient. And then finally, the compensate for the fact that I can't see you doing visible activity. We just move this over to being even more phonetic with digital communication. Like I really now is important for me not to reply quickly. Like I have to reply very quickly to Slack because like this is maybe the only way I have right now to demonstrate that I'm being suitor productive, right? So all these things happened overnight and people just said enough is enough. And I actually argue this is a piece I wrote a couple months ago I argued that many of these sort of spasmatic emergent grassroots revolution reform movements and work that we saw throughout the pandemic period were in part people responding in a primal way to this I've been pushed over the edge with what's going on in knowledge work. So I think the knowledge work component of the great resignation was a response to the people saying enough is enough. I think quiet quitting was a response for the younger generation who couldn't resign or switched down the half hours and moved to a cabin. So quiet quitting was a response. I think much of the fury around the remote work wars that really picked up steam in 2021 and 2022 was also a sort of misguided response to this deeper primal rejection of work has just become sort of intolerably frenetic and overloaded and I'm getting actually almost no real work done. Objectively, the remote work wars didn't make a ton of sense. Right, you had this thing that did not exist 16 months earlier. And now you had worker groups talking about it like it was a Geneva convention. It was just like fundamental right that a course work is supposed to happen at home. How could we ever take this away? This didn't even exist 16 months earlier. I think part of that was just a generalized zeal for reform. because when we took this already sort of exhausting tempo and this nihilism of all we do is talk about work and rarely get worked on. When we pushed that over the edge in the pandemic, it broke a lot of people. And a lot of the unrest we saw in the knowledge work sector throughout the pandemic, I think with people just responding to enough as enough, but they didn't really know exactly what they were responding to and I think a lot of those efforts were sort of misguided energy. And we missed a lot of the opportunities we had to really make better reform here That's the context in which I was thinking about slow productivity.

Yeah, I mean, and the reason, I mean, because it was a big decision. But the reason I chose to use as the primary case studies, these sort of historical figures, there's really two aspects that I think is important to sort of set the stage. One, I was wary of the uncanny valley effect. I've seen this a lot when talking about contemporary work. There's an uncanny valley effect. If you say, look, I'm going to tell you about a company that exists right now that's doing things differently, or a specific employee at a marketing company, and here's how they do things differently. Their job is so similar to yours that the difference has really begun to matter, right? And people have a hard time getting past like, oh, but that's a client service firm and our client. We have a time sheet firm and it's actually a, I found it is difficult for people to get past that uncanny valley. It's too close. If it's too close to what they do, but not exactly what they do, it becomes an unbridgeable or just orienting gap. So I said, why don't I look at what I call traditional knowledge workers who are actually defined by this could be the critique, but I'm twisting the critique to be a benefit. They're defined by all the freedom and autonomy they had to experiment with what's the best way to create value using my brain. I said, this is why these people are important because they had all the freedom in the world to figure out what works. So if we look at what they settled on, they're probably uncovering some useful universal principles about the best way of creating valuable things using the human brain. Now, what we could then do is once we isolate those principles, I can do the hard work of, okay, so how could we make that relevant to someone who works in a cubicle? How could we make that relevant to an entrepreneur in 2024? Yeah, there's a lot of work then to translate those principles to tactical things that makes sense to people today. But I thought that was the right way the right way to do it. So those three principles you mentioned. where the three big things that came up, if you study historically, people who were good with creating valuable things with their brain, they didn't work on too many things at the same time. They really avoided overload. Their pace was varied, hard periods, less hard periods, also they would measure productivity on very large time scales, so many of the most productive people in history. If you go back and look at a random month in their life, they seem incredibly non-productive, right? Because they didn't think about productivity in terms of like today needs to be productive. They thought about it like the next 10 years I want to produce something that matters. And then finally, they cared a lot about craft. Right, that was a sort of antidote to the appeal of busyness was to instead reorient their interest towards, I want to do something really well and I want to keep better at what I'm doing. All three of those principles I argue can be first adapted to modern knowledge workers with a lot of freedom and then can be further adapted even to knowledge workers who are in a situation where they have less freedom. We can get ideas from there that transmute into interesting tangible advice for people in various situations in our current and our current moment.