The internet increases variance. Digitization allows situations to be taken to their logical conclusion, instantly, even when that digital logic doesn’t quite work in physical reality. This means things can flip from zero to one, without warning. An overnight success, ten seconds in the making. The only certainty is rising volatility.
First, the observation: over the last 20 years, we’ve gone from 30 minute sitcoms to 30 second clips and 30 episode Netflix binges. From a stable 9-5 job to a gig economy task or a crypto windfall. From a standard life script to 30 year olds living with their parents and 20 year old startup CEOs.
This is a very general phenomenon. You see it in the dashboard of every internet disruptor. With Uber, for example, relative to the time of a standard taxi ride, some Uber trips are much longer and others much shorter.
Why is this happening? Because the internet connects people peer-to-peer. It disintermediates. In doing this it removes the middleman, the mediator, the moderator, and the mediocrity. Of course, each of those words has a different connotation. People are happy to see the middleman and mediocrity go, but they don’t necessarily want to see the moderator and mediator disappear.
Nevertheless, at least at first, when the internet enters an arena, once the Network Leviathan rears its head, this is what happens. Nodes that had never met before, could never have met before, now connect peer-to-peer. They can form something terrible like a Twitter mob, or they can form something amazing like ETH Research. You get extreme downside and extreme upside.137
One analogy is to a centrifuge. If you take a sample of biological fluid from your body and centrifuge it, you’ll see a bunch of layers that were previously mixed together. Then they all get separated out. That’s what the internet is doing to society, to institutions. It’s just centrifuging it into its constituent parts, whether that be albums separated into songs or newspapers disaggregated into articles.
That’s the unbundling. Then comes the rebundling. The songs get grouped into playlists, the articles grouped into Twitter feeds. This step too is profitable; it’s not the same as what came before, it’s a v3, it’s a flexible bundle. It’s the helical theory of history, where from one standpoint we’ve come full circle (“rebundling into an album-like playlist”) but from another axis we’ve made amazing progress (“anyone can play any individual song and create whatever playlist they want”).
With that said, that rebundling is still higher variance than the pre-internet bundles that preceded them. There are millions more playlists than albums, millions more Twitter feeds than newspapers.
As the internet increases variance, we see more upside and more downside in everything.
Technologists focus on the upside, because the gain from the wins (like search engines, smartphones, social networks, and artificial intelligence) should compound while the losses should be one-offs. That is, once you find a winning formula, or a rebundling, you can cheaply scale it across the rest of the network relatively quickly. So, this probably should lead to more net upside over time, just like every past technological revolution has. I think we’re already way in the black with the internet (almost every piece of information ever in billions of people’s hands for free at any time, for starters), but it depends on your metric.
Conversely, the establishment can only see the downside outcomes. That is, the BlueAnons can only see the QAnons who are worse than median, not the SatoshiAnons who are far better than median. It’s a bit like Paul Graham’s concept of the Blub programmer. Just like the Blub programmer can look down to see incompetence, but can’t look up to see brilliance, the establishmentarian can’t comprehend the upwards deviations of the internet. They think it’s just weird.
Just like Hollywood once compared Netflix to the Albanian Army, the US establishment doesn’t yet understand how much better Satoshi Nakamoto or Vitalik Buterin is than every apparatchik they have in the Federal Reserve system. And they don’t understand that upward deviation is creating a more competent group of global leaders than the American establishment, a more meritocratically selected group than the nepotists of the East Coast.
Just as it allowed Satoshi to rise.
There are two particular ways that the internet increases variance worth noting: social media and digital currency.
Social media increases social volatility. You can go viral or get canceled, experiencing large overnight gains138 or losses in status.
Digital currency increases financial volatility. You can go to the moon or “get rekt,” experiencing large overnight gains or losses in financial status.
There’s a parallel in history for this: glasnost and perestroika. Mikhail Gorbachev, the last Soviet leader, thought he could reform Soviet society by allowing more free speech (via glasnost) and free markets (via perestroika). He didn’t quite understand what he was in for. The resultant instability helped bring down the Soviet Union.
Similarly, social media is like American glasnost and cryptocurrency is American perestroika. Just as Gorbachev unleashed free speech and free market reforms because he believed communism could be reformed, the US establishment actually bought their own narrative in the 1990s and 2000s about their ostensibly free and democratic society. Only now are they realizing that the many speech and thought controls that their predecessors had set up and hidden in plain sight - like stringent regulations and high capital requirements for broadcast content production - was actually the key to their continued power.
Now that it’s clear that the Internet is to the USA what the USA was to the USSR, that it’s truly free speech and free markets, they are trying to tamp down the American Spring they’ve unleashed, but it may be too late. Obama was in a sense arguably America’s Gorbachev, as he allowed technology to grow mostly unimpeded from 2008-2016, to billions of users, without fully realizing what would ensue.
Few institutions that predated the internet will survive the internet.
Why? Because the internet increases variance, it causes huge surges of digital pressure on older institutions that just weren’t built for it. They can’t handle the peak levels of social and financial stress that the internet can unleash. They’re like seaside towns that weren’t built for a thousand year flood. Michael Solana’s post JUMP is quite good on this topic.
Indeed, this is a good analogy, because one of the ways to think about the internet is as a carrier of massive information waves. Most normal waves propagate in physical space –– the standard partial differential equations (PDEs) are 1-, 2-, or 3-dimensional (e.g. longitudinal waves like a slinky, transverse like electromagnetic waves, or earthquake-style spherical waves). But these information waves propagate on highly dynamic social networks where the topology139 of connection & disconnection changes quite a bit.
The digital is primary and the physical is now secondary.
The digital transition happens in three phases: there’s the physical version, the intermediate form, and then the internet-native version. If you’re into electrical engineering, you can think of this as analog, to analog/digital, to natively digital.
One example is the transition from a piece of paper, to a scanner which scans that file into a digital version, to a natively digital text file which begins life on the computer and is only printed out when it needs to be.
Another example is the transition from face-to-face meetings, to Zoom video (which is a scanner of faces), to natively digital VR meetings.
Yet another example: physical cash, to something like PayPal or fintech (which is just a scan of the pre-existing banking system), to the truly native digital version of money: cryptocurrency.
Once you see this pattern you can see it everywhere, and you can look for those spaces where we’re still stuck at the v2, at the scanned version, where we’ve taken the offline experience and put it online, but not fundamentally innovated.
Newspapers are actually only partially digitized. In 1996, the primary version of The New York Times was the physical paper and the mirror was the website. Then, gradually, more and more weight got shifted to the digital version. Now it can fairly be said that the physical paper is just a printout of the website, a snapshot at a particular time. And there are online-only features like interactive graphics that are impossible to replicate in the physical paper. Most importantly, the comments section is really social media, particularly Twitter, where all the reporters are located.
But this is still really just a newspaper, put online. Most of it can be printed out. What’s the next step in this evolution? What does natively digital news look like? There are at least two concepts of interest here: morning dashboards replacing the morning newspaper, and cryptographically verifiable event feeds replacing tweets of unverifiable content.
Dashboards > newspapers. If you are in tech, the first thing you look at each day may be a personal or company dashboard, like your fitbit or your sales. This is good. The first thing you look at each day shouldn’t be random stories someone else picked. Should be carefully selected metrics you want to improve. This is a good vector of attack to definitionally disrupt newspapers.
If we think about it from Clayton Christensen’s “jobs to be done” perspective, newspapers have this incredible pride-of-place — first thing you look at in the morning! — but typically do not add enough value to deserve that position.
On-chain event feeds > Twitter > newspapers. One key observation is that just as many sports articles are digest of box scores, and many financial articles are summaries of the day’s stock action, so too are many political and tech articles merely wrappers around tweets.
Because news breaks on Twitter. So, eventually, the next kind of newspaper will look something like a cryptographically verified version of Twitter. The first draft of history will be the raw on-chain event feed, written directly to the ledger of record by billions of writers and sensors around the world.
In other words, truly digital newspapers will be on-chain event feeds. Digitally signed crypto oracles, not corporations.
My friend Daniel Gross remarked that 2020 will be seen by future historians as the year when the internet age truly began.
The lasting impact of COVID-19 is that it flipped the world from physical to digital first. Because the internet in 2000 or 2010 couldn’t bear the load of the entire physical world. But by 2020, it kind of could. Now it’s not just about remote work, but remote life.
During the pandemic, every sector that had previously been socially resistant to the internet (healthcare, education, law, finance, government itself) capitulated. Those aspects of society that had been very gradually changing with technology shifted overnight. For example, the convention of politeness shifted: now it was rude to ask for an in-person business meeting, as you’d do it remote if at all possible.
With vaccination, many of these things have flipped back, but they won’t come back all the way. Digitization was permanently accelerated.
It used to be that the physical world was primary, and the internet was the mirror. Now that has flipped. The digital world is primary and the physical world is just the mirror. We’re still physical beings, of course. But important events happen on the internet first and then materialize in the physical world later, if ever.
All value eventually becomes digital, because we are generalizing the concept of “printing” from inking a piece of paper to actually materializing digital things in the physical world. This is counterintuitive, and you’ll have objections. But let’s get there in a few steps.
- Much value creation is already digital. If you’re reading this, you’re probably an information worker. You may not have thought about it this way, but the majority of your waking hours are probably spent in front of one screen or another — a laptop for work, a phone on the go, a tablet for reading, and so on. So, most of your life is already spent in the Matrix, in a sense, even before the advent of widespread AR/VR. Short of a pullback to an Amish or Andaman Islander existence, most of your life is and will be digitally influenced in some form. Moreover, much of the value in the physical world comes from blueprints created on a computer in some form; eg, the iPhones manufactured in Shenzhen gain much of their value from the designers in California. So, a good fraction of value creation is largely digital.
- More value creation is becoming digital. Read Packy McCormick’s article on “The Great Online Game,” and think about every information worker essentially pressing buttons to earn cryptocurrency in a giant globalized internet economy. That’s what 2030 or 2035 is on track to look like.
- Much spending is already digital. Think about what fraction of your spending already goes to digital goods like books, music, software-as-a-service, and the like. Now think about what fraction of the remainder goes through a digital interface of some kind, whether through an ecommerce website like Amazon or a point-of-sale terminal via Apple Pay. So, it’s already fairly uncommon for people in industrialized societies to do a fully offline purely physical transaction, which might be conceptualized as “hand a five dollar bill over at a farmer’s market for several tomatos.”
- Many actions can be analogized to printing. Now take this one step further and think about the remaining offline components as “printing” something out, though you can use the word “materializing” if it suits. You hit a button on Amazon and a complicated multijurisdictional delivery process ensues, resulting in a box landing at your front door. You hit a button on Uber and a car arrives. You hit a button on Doordash and food arrives. You hit a button to rent an Airbnb, and then another to open the smart lock, and the door to housing opens. You can do the same for the door to your coworking space office, or the door to your electric car. So, more and more of the goods people prize in the physical world are in a sense “printed” out.
- Many printing actions can be fully automated. Today, there’s a human in the loop for things like food delivery. But as robotics improves, this could in theory become a completely electromechanical process, just like printing. Every individual step from the farm to table could be automated. As this visual shows, there are already robots for each step: robots in the fertilizer factories, for harvesting, and for last-mile delivery. As an exercise, it’d be useful to a full stack example where someone “prints” out an apple and it’s fully robotically grown and delivered, even if in practice you’d have stockpiles of apples rather than (slowly!) growing them on demand.
So, if you put all that together, all value is digital. Everything starts on the computer, generates cryptocurrency, and can be used either to buy digital goods or to pay robots to materialize things in the physical world.
Humans will still exist, of course, but the economy will become the cryptoeconomy. All value goes digital.
What is the productivity mystery? Well, we really should be in the middle of a golden age of productivity. Within living memory, computers did not exist. Photocopiers did not exist. Even backspace did not exist. You had to type it all by hand.
It wasn’t that long ago that you couldn’t search all your documents, sort them, back them up, look things up, copy/paste things, email things, change fonts of things, or undo things. Instead, you had to type it all on a typewriter!
If you’re doing information work, relative to your ancestors who worked with papyrus, paper, or typewriter, you are a golden god surfing on a sea of electrons. You can make things happen in seconds that would have taken them weeks, if they could do them at all.
We should also be far more productive in the physical world. After all, our predecessors built railroads, skyscrapers, airplanes, and automobiles without computers or the internet. And built them fast. Using just typewriters, slide rules, & safety margins.
This is a corollary to the Thiel/Cowen/Hall concept of the Great Stagnation. Where has all that extra productivity gone? It doesn’t appear manifest in the physical world, for sure, though you can argue it is there in the internet world. There are a few possible theses.
The Great Distraction. All the productivity we gained has been frittered away on equal-and-opposite distractions like social media and games.
The Great Dissipation. The productivity has been dissipated on things like forms, compliance, and process.
The Great Divergence. The productivity is here, it’s just only harnessed by the indistractable few. The founders of tech unicorns, for example, may have more ability to focus online than most.
The Great Dilemma. The productivity has been burned in bizarre ways that require line-by-line “profiling” of everything, like this tunnel study.
The Great Dumbness. The productivity is here, but we’ve just made dumb decisions in the West while others have harnessed it. See for example China building a train station in nine hours vs taking 100-1000X140 that long to upgrade a Caltrain stop. Now, yes, I’m sure not every train station in China is built in nine hours, and wouldn’t be surprised if some regions in the US (or the West more broadly) do better than SFBA. But feels likely that a systematic study would find a qualitative speed gap, 10-100X or more.
The Great Delay. The productivity will be here, but is delayed till the arrival of robotics. That is, for things we can do completely on the computer, productivity has measurably accelerated. It is 100X faster to email something than to mail it. But a slow human still needs to act on it. So, in this hypothesis, humans are now the limiting factor.
Essentially, representing a complex project on disk in something like Google Docs may not be the productivity win we think it is. Humans still need to comprehend all those electronic documents to build the thing in real life.
So the problem may be in the analog/digital interface. Do we need to actuate as fast as we compute? That would mean zero-delay robotic task completion will be the true productivity unlock. And that we haven’t gone full digital yet. So long as humans are still in the loop, we won’t get the full benefits of digital productivity.
I don’t know the answer, but I think the line-by-line profiling approach used on the tunnels is a good but slow way to find out exactly what went wrong, while the approach of looking at other countries and time periods – namely, studying history – could actually be the fast way of figuring out what might be right.
If the organic borders of the physical world are rivers and mountain ranges, the organic borders of the internet are software incompatibilities and language barriers.
The first of these is obvious: Facebook’s ecosystem is distinct from Google’s is distinct from Ethereum’s, because the backends don’t fully overlap, because they’re incompatible at the software level.
The second is a bit less obvious. You can imagine the internet being cut up into continental-scale pieces, with the English-language internet being the largest with billions of people, the Chinese-language internet being the second largest with 1.3 billion, and so on for the Spanish-language, Japanese-language, Korean-language, Russian-language internets.
One huge difference between the English internet and Chinese internet is that the former is global and arguably decentralized while the latter is heavily concentrated in China with the CCP maintaining root control over most key nodes.
Another important consequence is that the English internet is about to admit a billion new users in the form of all the Indians who are newly coming online. And because the Indian internet becomes a much bigger part of the English-language discourse, it will be difficult for the US establishment to censor the English-language internet as much as they want to, because hosting can be based in the sovereign country of India.
A network defect is when increasing the size beyond a certain point decreases the value of the network. Metcalfe’s law doesn’t include this dynamic as utility is projected to just increase to infinity as network size grows, but there are a few different mathematical models that predict this outcome, such as congestion-based models or this post by Vitalik.
Repulsion within a network is a key dynamic that can lead to network defect. The idea is that two or more subgroups within a network have such conflict that it reduces the global value of the network for both, until one of them defects to another network. So it’s a network “defect” in both senses of the term: a failure and a political defection.