Mencius Moldbug is dismissive of the idea of replacing rule by a King with rule by a smart contract, he calls it a "Quantum Exponential leap".
Here's why I think he's wrong:
Here's why I think he's wrong:
![Thread 🧵](https://abs.twimg.com/emoji/v2/72x72/1f9f5.png)
First of all, you can see the whole section from @kaschuta's show here (skip to 1:11:18):
The criticism is two-pronged:
(1) We don't have the technology to use smart contracts to make real-world decisions and blockchain tech is slow and expensive, and you would also need AI tech that we don't have because it's AI-complete.
(1) We don't have the technology to use smart contracts to make real-world decisions and blockchain tech is slow and expensive, and you would also need AI tech that we don't have because it's AI-complete.
(2) You can't rule the world with set of rules or laws because there are always exceptions, or because humans can't have emotional connections with computers, or because it's necessary for humans to follow orders from other humans.
(3) Actually there's a third prong: a smart contract of some kind can't enforce its will unless you give it cryptographic control over weapons (which would be bad, because it could lead to Skynet)
So, what's the response?
Prong (1) is wrong because Moldbug is a bit behind the times on technology. People in modern AI don't talk about "AI-complete" any more, it's an artifact from the 90s and modern state-of-the-art systems don't exhibit this "completeness" property...
Prong (1) is wrong because Moldbug is a bit behind the times on technology. People in modern AI don't talk about "AI-complete" any more, it's an artifact from the 90s and modern state-of-the-art systems don't exhibit this "completeness" property...
GPT-3 can write poetry without being able to fold laundry, Alphastar can play pro-level Starcraft without being able to write poems, etc. AI is moving forward quite rapidly with strong narrow capabilities but for the foreseeable future there probably won't be a "general" ...
... AI system that can solve arbitrary problems. Instead, we have capable systems in various domains with gradually broadening abilities. AI is going "Depth first" rather than "breadth first" through problem-space.
I think a lot of people on the bird site don't understand just how fast AI is moving, you can play around with the openAI blog to see what it's capable of.
https://openai.com/blog/dall-e/
For example, DALL-E:
https://openai.com/blog/dall-e/
For example, DALL-E:
You can literally tell the AI to draw a picture of an avocado wearing a helmet walking a dog, and it will produce that picture for you. This AI has mastered the mapping between words and pictures.
On the blockchain side, there are a lot of people working hard to make blockchain tech better along multiple axes. I can't explain all that tech in a twitter thread, but what we have now is not what we will have in 10 or 20 years.
So, prong (2) "You can't rule the world with set of rules or laws because there are always exceptions"
We already rule the online world with rules and have ways to escalate exceptions. Try posting hardcore porn on twitter or facebook, an AI is almost certainly scanning for it.
We already rule the online world with rules and have ways to escalate exceptions. Try posting hardcore porn on twitter or facebook, an AI is almost certainly scanning for it.
"You can't do that!"
But it's already happening. Copyright on Youtube, porn on social networks, etc.
OK, you can have some rough-and-ready rules, but how do the exceptional cases get decided? One approach to this is economic: people can vote with their wallets.
But it's already happening. Copyright on Youtube, porn on social networks, etc.
OK, you can have some rough-and-ready rules, but how do the exceptional cases get decided? One approach to this is economic: people can vote with their wallets.
Augur's Oracle works like this: a question is submitted to holders of the REP token, if they disagree they can keep escalating, if there is disagreement at the highest level, then REP forks into two chains. The idea is that the fork that chose the more accurate answer ...
... will hold more value long-term, because an inaccurate court or oracle is worth less than an accurate one.
This kind of economic solution wouldn't work in a global totalitarian dictatorship, but in a multipolar world where no single lie has a crushing advantage, it works.
This kind of economic solution wouldn't work in a global totalitarian dictatorship, but in a multipolar world where no single lie has a crushing advantage, it works.
The word in 100 years is going to be mostly an online world. People will willingly "plug into the matrix", but what that matrix is like matters a lot. And everything that matters will be decided by some kind of moderation system.
Social networks are already governed by content moderation systems, and as we move forward these social networks are gradually going to become everything that matters. The Pandemic accelerated this, but it was going to happen anyway.
Moderation and promotion aren't only downstream of power. They are also upstream of it. The entity that controls the censorship and promotion functions can isolate all its opponents from each other and unite its supporters. Youtube will start hiding dislikes, for example.
We also know that YouTube deliberately suppresses certain videos - limits or throttles the number of shares, views or recommendations.
Given that government by algorithm is already happening, I don't think it's something you can ignore. shape that future, or not.
The difference I am looking to shape is opaque backroom dealings versus transparent crypto contracts.
The difference I am looking to shape is opaque backroom dealings versus transparent crypto contracts.
The refutation of the third prong follows from this. As humans put more and more of what matters online, you don't need a terminator or a killbot to enforce the rules. Yes, we will eventually have police killbots (or stunbots) that use violence to make people physically comply.
But most human activity will end up online anyway. People will plug in, people will upload themselves, etc, and even if they do stay in meatspace, almost everything will get decided by algorithms.
Onto some objections: "humans have interests and those interests will creep into whatever system you make"
https://twitter.com/kwamurai/status/1388228873964371983
https://twitter.com/kwamurai/status/1388228873964371983
Yes, people have interests and power will try to creep into whatever system you make. But blockchain is building systems that are designed to be censorship-resistant and transparent.
"Science", "Experts", "Independent Press", "Rule of Law" are all aging institutions.
"Science", "Experts", "Independent Press", "Rule of Law" are all aging institutions.
Old-fashioned institutions where things get decided through informal channels and laundered in committees lead to oppression. The Woke problem in modern institutions is a manifestation of this, but it's not the only one. It's just a special case.
Of course there is a reflexive midwit anti-crypto/anti-blockchain sentiment about. It will age as well as the "world market for maybe five computers" sentiment.
Crypto/blockchain is a new technology and it's full of scams, most people can't separate the wheat from the chaff so it's understandable that there will be a lot of fakery and skepticism now.
But my job is to see clearly through that stuff. The future of power is digital.
But my job is to see clearly through that stuff. The future of power is digital.