Proposal: AGI monetary policy to drive long-term network value

Hi everyone !

I have been following the SingularityNet project for more than a year now and I’m very impressed with what has been done on the product and business development side. Kudos to the team !

I still think however that the network misses one key thing to achieve its tremendous potential: a clear, “game-theoretic” incentive to hold AGI Tokens on the (very) long term.

The SingularityNet Whitepaper 2.0. focuses mainly on the SingularityNet platform and on high level AI services but leaves key questions:

What will drive long term token value ?

Saying that JUST increased demand from buyers of AI services will push up the price is a flawed answer as it leads to my second question:

Why shouldn’t any AGI Holder - investor or data scientist earning tokens on the network - sell at any given moment ?

Yes, if I’m a data scientist earning a salary thanks to the SingularityNet, I’ll need to convert my AGI back to fiat to pay my bills. What incentive do I have to sell less than 100% of the AGI I make ?

To set things straight, I believe the long term price appreciation is key to the success of any project like this one. Price is what drives media coverage, enables infrastructure/platform funding and makes sure that different stakeholders (miners, investors, service providers, users …) have a good (monetary) reason to collaborate and to create a virtuous cycle.

Pure crypto-commodities, aka utility tokens, have little chance of sustaining high prices on the long run, as the fiat-denominated price of what they enable is by nature deflationary (eg: the fiat-denominated price of a unit of computing power is fundamentally getting cheaper as computer chips are getting more efficient).

The same logic applies to AGI tokens, as what they enable, AI services, should become cheaper as AI services becomes more mainstream and thus cheaper. There should indeed be growing competition within the SingularityNet network and outside of it.

AGI tokens being a crypto-commodity, they can be ‘transformed’ into AI services by a buyer the same way we today use oil to produce energy. The only difference with an actual commodity such as oil is that no AGI tokens are being ‘burnt’ in the process, as AGI tokens are simply sent to whoever owns the algorithm. Because oil is getting scarcer, its price is going up on the long run.

Here is a great article on the difference between cryptocommodities and cryptoassets:

Everyone in the crypto sphere has seen the recent rise of the Binance (BNB) and of the Maker Dao (MKR) tokens. The main reason for the success of these tokens remains product market fit. Binance is a leading crypto exchange and the Maker system has loaned over $200 million since its start in december 2017.

However, one structural aspect people seem to underestimate is the token burn mechanism used in both cases. Here is a high level description of what happens:

  • BNB coins are used to pay for transaction fees on the Binance exchange. Binance then burns BNB coins every quarter
  • MKR tokens are used to pay for interest rates on loans in the Maker Dao system. MKR tokens are then burnt as debt positions are being closed by users of the system

This means that even in the case of flat usage growth, both tokens are becoming scarcer, as a small pourcentage of the total supply is being burnt in a systematic way. While this system does not prevent speculation and volatility peaks, it acts as a stabilizer because it creates a structural buying force on the open market.

From the point of view of game theory, in the MakerDAO system, being the last token holder gives one the ability to fork the network and to get all the revenue generated by the system. That is a powerful incentive for people to hold as long as possible !

Where does that leave AGI ?

Given my understanding of what AGI is today, besides potential growing demand for AI services on the network, nobody has a tangible reason to buy AGI tokens now rather than in, say, two years, because of two things:

  • Fiat denominated prices of AI Services will go down as the market matures
  • AGI Tokens, the crypto-commodity used to ‘produce’ AI services on the network are not getting scarcer

With such a macro-economic environment, AI scientists would not want to earn a living on the network because the crypto-currency/commodity they are getting payed with should be depend solely on the broader cryptomarket sentiment - rather than on more predictable micro-economic fundamentals.

How to solve that problem ?

When 100 units of AGI go through the system, less than 100 units need to go back to the market on the other side. Here is an exemple:

  • John buys a service on the Network for 100 AGI
  • Once the task is completed, 95 AGI tokens are sent to Bob, who created the given AI Service
  • 5 AGI tokens are burnt
  • Bob sells his 95 tokens on the open market because he needs to pay his rent

-> Next time John wants to buy AGI Tokens to purchase services on the network, they will be relatively scarcer and thus more expensive
-> Bob now has an incentive to hold a few AGI tokens rather than sell everything he has as he knows John will need to use his algorithm again in the future
-> As AI services will be priced in fiat, all other things being equal, the amount of AGI burnt should go down as the price of AGI goes up.

As a conclusion, giving people a strong incentive to hold the token on the long term is as important as having a great infrastructure on which the token can be used. This incentive is yet to be implemented for AGI token holders such as myself.

What do you think about implementing a token burn mechanism for the AGI Token ?