Token Design as Optimization Design: Ocean and Ampleforth protocol
Clarence Thomas as Shinji's Dad

Token Design as Optimization Design: Ocean and Ampleforth protocol

By Will | Proof-of-Research | 19 Nov 2020


Where Token and Optimization Relate to each other

Back in February of 2018, Tent McConaghy gave a presentation on the many facets, or faces even, of crypto. He unpacks the token design of bitcoin as a protocol which gives incentives to users, who provide resources (energy, computational power, hashes) to the Bitcoin blockchain. It’s a very simple design and it’s very effective at achieving this goal of accumulating hashing power that emerges from users wanting the block rewards. He breaks this incentive optimisation down into an objective function, which defines certain characteristics you’d like to set, maximize or minimize. In bitcoin’s case it is to maximize the security of the network and make it harder to unravel. He likens the blockchain space to mechanism design, something many people call “reverse game theory”, because instead of being given a set of rules, and choosing the optimal move; mechanism design deals with making systems which one knows the choices that they want players to make, but don’t know what the players are thinking. He also likens these optimization designs of blockchain to evolutionary algorithms, for which he courteously provides a slide for us:de05d3663afbc3d638a67343c8cdd96b0c37f67b6cd42ec2ac61ea954e831ab8.png

 

Mechanism design, token engineering, Cryptoeconomics, tokenomics, financial cryptography, cybernetics; whatever you chose to call it, according to trent, these converge around optimization design. He, with the help of top token engineers, designed their token ecosystem to with the objective function of maximizing the relevant data; by providing an open data commons that threads together all data marketplaces, and incentivizing users to curate data by staking tokens, and rewarding those who properly predicted the data’s future popularity. He envisions access to data sets and all other data compute services being published on the Ocean Protocol. The incentive design, along with crypto-economic primitives (which drastically enhance the security and privacy of data sharing), will optimize sharing of relevant data, break the data silos, and open the data economy to its full maximum potential. 

 

Objective Functions

What I learned from this talk was the emphasis on the importance of getting the objective function of your token design right. We exist in the very early ages of token engineering, This is the objective function described for Ocean Protocol’s token design:


850d15458285f951b7fd7ecb5ee22003c0fa812115464c5fbbf2f143d8a61cb1.png

We see that just like Bitcoin, it has an expected reward and a set number of tokens at each interval. The only difference is that instead of a hash power variable, the value is described as: predicted popularity X actual popularity. This properly incentivizes users to always be curating the most important and relevant data. The rewards are designed to optimize Ocean’s goal in an objective manner, which means the users are free to behave however they want, and order will emerge as a result of the incentive scheme.

 

Ampleforth: The equilibrium machine

 

Now that we’ve sort of crash landed into token engineering, we will move to the main focus of this paper: The Ample Token design. The Ampleforth protocol is technically a “non-dilutive, smart-commodity money with near perfect elasticity”. Ampleforth has a token design which is theorized to achieve the objective of a stable value commodity money. Its objective function is to maximize a Supply/Demand equilibrium state, which results as a perfectly elastic commodity. So what does “elastic commodity money mean? In economics, We have a divergence of thought in what makes money “money”. Gold is considered a more sound money and world reserve than the dollar because its supply isn’t subject to nationalized inflation, and it dominates in terms of money hardness, meaning the cost of creating more of it is truly costly. 


76a408801ccee7c92e5238373d226cb17fea36a37cd0ccecdf3dbc396450fe6f.jpeg

The fiat monies like dollar bills are the polar opposite, in that they take the forms of “IOUs”, and more of it can be costlessly brought into the system, providing cheap liquidity in the form of monetary devaluation. Fiat currency has extremely useful properties from its elasticity (ability to expand and contract supply based on market conditions) to maintain a stable medium of exchange. It also has the ability to behave as an independent economic tool, and isn’t used for non-monetary uses, unlike gold which has correlation to the markets that use it.  While debt is very useful, that value comes out of human choice, not objective hardness, so it exists only within our imaginations. While it’s useful to have tools which help us make bets on the future, it is bad to let our imaginations run a muck with these tools, and even worse to utilize them as a base money in an economy. 

 

The Gold vs Fiat debate is incredibly deep and interesting, but it’s mostly not pertinent to proving the validity of Ample’s token design. I will go deeper into the history of currency wars some other time, because it’s very interesting and important for understanding Ampleforth’s true significance. For now; however, we will simply assume that Ample’s design objective is a desirable goal for the collective good of society, because for most that would be intuitively obvious.

 

Ampleforth’s Uniqueness as a Crypto Project

Not all tokens are based on a block-to-block emission schedule to enforce certain “work” in advance of the reward. And while some tokens even represent various degrees of governance, Ampleforth is fundamentally different from these norms. Ample’s tokenized incentives don’t come in the form of actively contributing some fungible service, like Ocean’s relevant data, but rather its incentive behaviorally affects the market within an information loop. This is to say, that instead of asking the users for a certain work prerequisite to be fulfilled, like mining or service provision, and then rewarding the users who do so accordingly, Ampleforth waits for open market price information to naturally develop on its own internal asset; the incentive rules then kick in to counteract that unpredictable behavior of “demand” on the open market. 



In Ampleforth’s design, we seek to align the users’ interests on an individual level, to get the result of a token that is bought and sold on the open market in such a way that it maintains a stable, non-dilutive value. This means Ampleforth seeks to become a stablecoin through emergent market behavior. The desired value of 1 ample is the US Dollar’s purchasing power in 2019. For simplicity we will refer to Ample’s target value as $1. Amples rules work with a target price Pt , and threshold δ.  If the exchange rate of Ample is above the Target price + δ, the protocol responds by minting tokens proportionally to users. If the exchange rate is less than Pt - δ , then the protocol reacts by contracting tokens proportionally from users.


5c4f7f1ac3c5ae2e267676046886163110a64420a1284a11960c216735600e04.jpeg

Cooperation of fast and slow traders

What is especially interesting to me about the token design is that the Ampleforth contract actually creates incentives that affect the behavior of players on external exchange markets. We now know that Ampleforth has a function in its contract known as a “rebase”. This rebase will expand or contract the supply of Ampleforth. The price of the token in the open market will apprise the contract policy, which will influence the supply. But while the market price influences the supply, it is the activity and psychology of the users who propagate the new supply information back into price


27b166d6fdfcfbd226b4948eab8614d640e8c742d2946af042545ba8540868e5.jpeg

For long term holders of Ample, it makes no difference to them whether the demand is reflected in the supply or price of goods. For example, Alice buys 1 Ample for $1.5. Later she notices that she now holds 1.5 Amples worth $1 each. See so either way, it doesn’t really make an impact on the individual’s wealth, which is still $1.5, and we as a society still benefit as a whole from the emergent property of a stable currency that finds its way back to the $1 equilibrium. Bob, as a fast trader, recognizes a third state in this occurrence. At t0, Bob has 1 Ample worth $1.5, the rebase occurs at t1, and the balance expands to 1.5 Amples worth $1.5 per coin. This is a sell opportunity. Finally at t2 the value of 1.5 amples is $1 per coin. There is a small opportunity in the market dynamics, in which people first get their rebase, and then the market goes to work, with fast traders selling off their newly minted tokens before the price has corrected itself. So you see, because the price of Ample actually relies on trader’s willingness to sell high and buy low, there is an explicit arbitrage opportunity for quick traders to make money off of the protocol’s rules to reach equilibrium. The reason this is fascinating is because we see sophisticated trading bots always trying to find arbitrage opportunities between exchanges, but here, we see it within the actual state of token supply, and this equilibrium rule is known by everyone who understands it, so there is a predictable behavior of people selling or buying depending on whether the asset is above or below equilibrium. When you give traders a distinct rule to base their orders on, they will behave based on everyone else’s quickness to respond to these price discrepancies, and trading algorithms can be made to farm for arbitrage opportunities. Because of these hard coded rules, the incentives create a race to the equilibrium. This effectively gives payout opportunities to fast traders who push the price back to equilibrium based on rational greed. (the more competitive the market becomes, the smaller the Supply/Demand discrepancies become) Most importantly, It accomplishes this critical function without at all affecting the long term holders.

 

The Ample is also meeting the requirement of being a stable store of value by being non-dilutive, meaning everybody that holds Ample will receive a direct proportion of the new supply based on their stake in the network (how many tokens that they own) This is a key component to the Ampleforth contract because it protects the interest of long term holders while scaling the demand of the currency. The dollar does not have this property, and it actually unfairly punishes savers in the ecosystem through dilution, which is, there’s really no nicer way to put it, theft.


f176b1e18f18df5de7c4d028b7e9c0cdad0a1b5a28d8c61a5f2d6810eeb2154d.png

This is not only bad design with missing features, but it is a breeding ground for massive power asymmetries between regular users of the dollar, and those who decide how new dollars are distributed and who they go to. 

The Ampleforth token ultimately creates unique trade opportunities based on Price and supply knowledge, and leverages this to create a network that self corrects price movements thanks to independent actors choosing their own moves by looking around at their local environment, and out of this, a property of Supply/Demand equilibrium is actualized on the macroeconomic level. It successfully aligns the incentives of individual users and produces a stable value primitive through simple but sophisticated token engineering. Bringing this back to the beginning where we talked about Trent comparing these designs to evolutionary algorithms, I think we do clearly see such an “equilibrium” optimization in Ampleforth, and it is really just a simple money game when you deconstruct the protocol.  I hope now that people not only understand the uniqueness of Ampleforth, but also the great potential in token engineering, and how it can design systems that benefit society as a whole, and not just the “get rich quick” ponzi governance schemes for which this space is commonly known.




Will
Will

Making blogs for the history and progress of crypto based economies. Game Theory, Applied Cryptography, crypto economics and tokenization


Proof-of-Research
Proof-of-Research

A research blog for digital currencies

Send a $0.01 microtip in crypto to the author, and earn yourself as you read!

20% to author / 80% to me.
We pay the tips from our rewards pool.