Arguably the main feature of token ecosystems, also known as public blockchains, is getting people to do things.
The incentives are strong. But like AI design and optimizers, such incentives are hard to get right.
The Block chains can even be framed in life. In this context, what would happen if a rogue life form appeared that sucked life energy from the planet? More specifically: has become Bitcoin Rebellious way of life? This article explores these questions, in the first installment of a larger series aimed at improving the token design process.
We will start from the perspective of optimization and AI and go back to blockchain and incentives.
artificial intelligence game
For years of experience people who have worked in creative artificial intelligence encountered problems like Wires hang randomly.
These issues were resolved by adding constraints coded in computer terminology, such aseach node must connect to > one edge, Trying to fix the dangling wire problem.
But these solutions may have created other problems, such as the current in the wire being 100 times higher than normal, which could cause the circuit to explode. The problem was resolved and the process was repeated. All of this can be fine within a certain number of restrictions. After that, constraints became a really heavy and cumbersome task.
One of the lessons to be learned from this type of operation is that, just like debugging software, challenges are placed between the operator’s intent and what the machine can understand the operator wants. Another lesson is that defining objective function and combining it with constraints is difficult.
The Clip Maximizer
Communicating intent is difficult. This idea is the basis Nick Bostrom Clip Maximizer:
Suppose we have an AI whose sole purpose is to make as many syllables as possible. The AI will quickly realize that it would be much better if there were no humans because they might decide to turn it off. If humans did that, there would be fewer syllables. Also, human bodies contain many atoms that can be turned into paper clips. The future that AI will try to head towards is going to be a future where there are a lot of clips, but no humans.
In this scenario, humans pass on the primary goal – maximizing syllables – but ignoring a key limitation, which is not to destroy humanity. But how do you define this last limitation? Yes it is difficult.
It doesn’t even matter if the AI is really dumb. As long as he has access to resources to keep growing, our world may end up with paper clips. Optimizers and AI don’t care what you mean, they happily multiply it. Some people might think –Well, just turn off the AI-. This works with centralized AI, but not with decentralized AI (AI DAO).
Blockchain as trusted machines
We have provided context for optimizers. Now let’s give context to an extension block chains. Next, we combine the contexts.
Blockchains have many great features that go beyond traditional distributed systems: they are decentralized (not owned or controlled by any single entity), immutable (once written to the blockchain it is as if they are set in stone) and they facilitate the issuance and transfer of assets.
This makes them trustworthy machines. These features unlock high-level capabilities such as smart contracts.
Blockchain as motivational machines
“Show me the trigger and I’ll show you the result.” –Charlie Manger
The blockchain community understands that blockchain can help align incentives among a group of token holders.
Each token holder has a part in the game. But the benefit is actually more general than just incentive alignment: you can design whatever incentives you want, and give them big rewards. In other words, you can get people to do things by rewarding them with tokens. Blockchains are motivational machines.
This can be considered as a superpower. The block rewards feature defines what you want network participants to do. So the question is: What do you want the people in your network to do?
This question has a crucial corollary:How far can you communicate this intent to machines? This is a very hot detail. Do we really know how to design incentives?
Blockchain like life?
Erwin Schrödinger frames life simply as physical processes in his treatise What is Life? Recently, physicist Jeremy England gave it a thermodynamic framework: it’s about entropy. Carbon is not a god.
The Artificial Life (A-Life) community acknowledges that the definition of “life” is controversial. There are clearly things that are not life, like a hammer, and others that are clearly life, like a puppy. But there are shades of gray in between. We can think of it as a checklist of, say, 20 items. Independent mobility? Yes, self-cloning? verification. Decision making? verification. and so on. Carry out adequate checks and this is life.
Ralph Merkel wrote of Bitcoin as life:
Bitcoin is the first example of a new lifestyle. Live and breathe online. He lives because he can pay people to keep him alive. He lives because he does a useful service that people pay him for. … on fire. It can’t even be interrupted. If a nuclear war destroyed half of our planet, it would continue to live without damage.
Has Bitcoin become a criminal?
To summarize the last few sections:
- Designing the goals and constraints of an optimizer/AI is challenging.
- An AI that has access to massive resources, but weak goals and constraints can end badly for humanity (maximum in the clip).
- Bitcoin can be considered a lifestyle, or a very stupid artificial intelligence. It is almost impossible to stop.
Let’s put it all together. Let’s recall the Bitcoin block reward function (also known as the objective function): maximizing security, maximizing hash rate, and maximizing electricity usage.
And it is improving against this objective remarkably well. So good that it was on track to overtake the US in July 2019. Energy is perhaps the most important resource on Earth. This is why we humans start wars. (Oil, remember?)
Briefly:
We have a lifestyle that we basically can’t stop, and that is manually optimizing this precious resource: energy. This way of life is called Bitcoin.
What do you think of the strength of incentives? We must get the incentives right when we build these token ecosystems.
conclusion
Satoshi almost certainly did not want to suck the life force off the planet. Objective job design, also known as incentive design, is challenging. But you have to try. To do well, we need strong engineering theory, practices, and tools. That is, token architecture. The next article in this series delves into it.
Original article by Trent McConaughey at https://blog.oceanprotocol.com/