In our current world, one can rarely monetize their knowledge in a direct manner. We almost never buy or sell (or price) a single piece of knowledge.
Seeking knowledge in an economic setting is currently mostly done by appealing to people who we assume to be involved in a field where the knowledge of interest may exist, just like there isn’t a book for every single answer, but we look up answers in books that deal with apparently relevant fields. Knowledge is a major part of our economy, and here we seek cases in which we can make the economics of knowledge more efficient and more advanced.
Formalizing knowledge in a machine-accessible format offers an advantage over books and search engines: the data isn’t a stream of bits anymore but the meaning behind the bits, and therefore we can indeed perform an automatic semantic lookup of small pieces of knowledge even in a large compilation of writings. I spoke several times about how a shared knowledge base will evolve over Tau and its logic-based discussions. In a framework where knowledge is constantly formalized and shared between users, it’s only natural to consider an economic infrastructure to facilitate all aspects of the economics of formal knowledge.
In this text, I will focus on some aspects of the economics of knowledge in Agoras while leaving some others to future writings.
First let’s consider how knowledge is generated, or “mined”. If we treat knowledge as something beyond (natural or artificial) sensory inputs, one can argue that the act of reasoning is the one that generates new knowledge from existing one. Such a process can also be done automatically. But how can we guide this automatic search process so that it’ll produce interesting results?
In the Whitepaper, we spoke about the concept of “interesting questions”. Just like answers may be correct or not, in the same way, questions may be interesting or not, but correctness is irrelevant and undefined for questions.
We argue that this defines the roles of humans and machines in our setting: machines will never be able to tell which question is interesting, as this is purely a human-nature-dependent thing. The machine’s role is to help us answer our interesting questions by automatically performing the tedious work of reasoning over the knowledge we already fed into it.
It might be worth adding a remark about terminology here: by “queries” we refer to questions that we ask the machine, and we expect the machine to have enough data in its knowledge base in order to answer this query, while by “question” we refer to a question that we want to mark as “interesting”, assuming that the answer is yet unknown over the system.
In short, questions and answers will play an important role in discussions and knowledge formalization over Tau, but this also helps to understand the economics of knowledge. A piece of knowledge is valuable if it’s interesting, and its value should depend both on the level of interest and on the hardness of answering the question. The hardness of answering is something slightly more accessible to machines since they can count steps, and besides, there is a rich theory about the complexities of various reasoning tasks (e.g. Descriptive Complexity Theory).
Focusing on the economics of knowledge, questions and answers may be seen as demand and supply respectively, although this doesn’t cover the full picture and I’ll add more details later on. Users over the platform will be able to mark questions as interesting in the same convenient, collaborative, and natural way we proposed all along: via the course of discussion.
Now let’s imagine how a knowledgeable individual may generate income over Agoras. In a broad sense, users interested in certain questions may offer a reward for an answer. Verification of answers may be done in several ways. In some cases, like common mathematical questions, the answerer may supply proof for the answer, and no dispute arises over whether or not the answer is correct. We might even have information about whether such a verification process is expected to be efficient, again by using considerations from e.g. Descriptive Complexity Theory, and there’s a lot to add about this point from a cryptographic point of view, but this is out of the scope of the current article.
However, sometimes asking for mathematical proof is too much. Sometimes one might trust an expert in the traditional way, simply by impression or recommendation or advertisement, etc. as common, and then automatically trust their answers. A simple example would be to trust some medical doctor which you already know well, and not require them to supply a mathematical proof for each and every medical advice that they give, as this will render the whole thing impractical (well at least until the singularity comes).
So the last example gives rise to one form of knowledge trading. Consider some reputable body, like a university or a trusted expert, which takes the hard task of formalizing some large and useful body of knowledge. They can then offer a subscription to users for automatic participation in their discussions. Tau will allow “auto-commenting” so it will be able to automatically participate in a discussion on your behalf, once you tell it your opinions (your Worldview) over time, by posting or by agreeing or disagreeing with others.
This lets subscribers enjoy automatic participation in discussions where the data comes from a trusted source (from the single user’s perspective). For a specific instance, a law firm offers their Knowledgebase to automatically participate in corporate discussions and possibly auto-comment on certain ideas to be legal or illegal.
Another form of subscription may be pay-per-query, allowing subscribers to ask questions and get answers without revealing their whole knowledgebase. Further, thanks to Tau’s collaborative knowledge formation aspects, a group may formalize knowledge and monetize it together.
At this point, it’s worth mentioning one of the original Agoras’ main innovations: the concept of Autonomous Representation. Users have their own local assets, being computer resources, knowledge, coins, and possibly other assets that they allow their formalized Worldview to take into consideration. Agoras will then be able to tailor a deal by looking at the available knowledge and contracts offered out there, or by publishing a bid for certain knowledge or contract, all from a coherent and logically proven plan of combining the assets and opportunities into a good deal.
One can even reach an extent that is unheard of in common automatic planners: the user will be able to ask Agoras for deals that don’t break laws and regulations, once the law is formalized over the system. This is just a small example of what a logical reasoner may perform in an economy.
Agoras will also contain a computer resources market and a future contracts market. The Autonomous Representation is therefore a holistic application that may involve all parts of Tau and Agoras’ capabilities.
Our team has some new thoughts and plans to add to this design. That all emphasized, clearly a human touch is crucial for many forms of knowledge transfer, and not all knowledge may or is suitable to be formalized in all circumstances. Seeking quick advice from a doctor by means of exchanging formalized knowledge is not always the preferred way to go, same for taking a private tutor, and so many more examples. We, therefore, intend to also have a freestyle form of trading knowledge, in the form of text, audio, and video, which we are calling Agoras Live.
A version of this text was first published by Ohad Asor, Founder and CTO of Tau and Agoras, on May 1, 2020
Ohad has been a Software Developer and Mathematician at top tech companies since 1995. He was the youngest university student in Israel, studying Mathematics and Computer Science at the age of 13. Over the years he has accumulated extensive knowledge and experience in programming and various areas of math, with a recent focus on logic, machine learning, complexity theory, philosophy of science, economics, social choice, and decentralized networks. He designed and implements Tau and Agoras. As the founder, he is currently leading the development of the project. Work history: Logician and Mathematician, expert in AI, prior work with GE Healthcare developing software systems.
Learn more about Tau and Agoras here.