Intelligent Assets and AI Blockchains2nd June 2019
Imagining A Not so Distant Future
1. The Problem With Blockchain Markets
There have been two major events in Blockchain innovation history. The first was the creation of Bitcoin by Satoshi Nakomoto in 2008. This gave rise to a software that resolved the double-spending problem traditionally associated with digital currencies by means of harnessing a probabilistic mathematical equation in the manufacture of bitcoins.
The second major event in Blockchain innovation was the 2015 creation of the Ethereum network, which allowed anyone to deploy a similar non-forgeable unit of currency without the requirement for building a Blockchain for each individual currency. The way this worked was via redistributing computer power across the network and reusing the cost to support an additional layer of the Ethereum network on which digital tokens (as opposed to Blockchain coins) were supported.
The explosion in the number of digital currencies as a result of Ethereum’s global network propagation has resulted in an innovation irony. That irony is that while Bitcoin and Ethereum are both decentralised networks, the exchanges that digital currencies trade on are for the most part not.
Therefore, while each network confirms to the principle of preventing double-spending problems and ensuring fair and transparent market behaviour, nefarious actors have with the rise of the number of cryptocurrencies been able to simply bypass the barriers the networks themselves put up against unethical behaviour by running crypto exchanges off centralised servers which are entirely hidden from public sight.
As a result, market manipulation is rampant in the digital currency universe, with many Blockchain projects listing tokens for a minimum of $30,000 upwards only to find that there is no liquidity for their currency. In the event their currency does do well, most of the time the central server they are on (i.e. the exchange operators) quickly crashes the price at the point they are financially incentivised to put profitability over their customers’ user experience.
Part of the problem with bad actors becoming an increasing influence on digital currency markets has to do with the highly-centralised mining processes that dominate the new currency protocol. In the case of proof-of-work Blockchains, coin age usually serves as a function of mining priority.
This rule was implemented to incentivise miners to hold larger shares of coins for longer so that the price of the digital currency remains stronger as demand outpaces supply. The problem is that holding periods are by themselves not a suitable weight for determination of preference. As a result of this feature of the Bitcoin Blockchain, market makers and other liquidity providers are not incentivised to create meaningful liquidity.
In this way, they are directly incentivised to purely prioritise the liquidity of one currency only — that which is easiest for them to cash out in; Bitcoin.
POS chains have yet to offer a viable alternative to the dilemma with wallet mining favouring incumbent premine coins purchasers disproportionately, sometimes negating many advantages of the decentralised cryptographic data mining process.
In the case of many POS chains, buyers simply hoard extensive numbers of Masternodes — software mining devices — ultimately churning out coins at what are initially extremely high coin production rates so the coins can be quickly sold in exchange for bitcoins.
2. AI Blockchain Solution
In the period since Bitcoin was launched, there have been large-scale improvements in the overall usability and application of Artificially Intelligent (AI) software programs, especially in the area of financial technology. Many so-called “robotraders” employ self-learning trading systems that have been proven to profitably engineer consistent trading returns based on real-time assessments of market behaviour and pricing.
An artificially-intelligent Blockchain that was simultaneously integrated into its own exchange platform and which traded against a whole variety of non-securitised and ultimately too, securitised assets, as a digital cash product would end a lot of the problems that today’s Blockchain innovation drain suffers from, with the simultaneous and ongoing continual copy-and-paste of traditional Blockchains by new teams, and subsequent garage-sale, followed by the exchange pump-and-dump of the new chain’s currency, which after that is most often simply abandoned altogether.
Specifically, an AI Blockchain would achieve the following:
i. It is market price-efficient. By constantly scanning the exchange for price updated and volume order updates, the AI Blockchain can easily synchronise its mining algorithm with whatever price increase or decrease is likely to occur in its own currency over the short- and medium-term. This will over time make it a stronger source of purchasing power and act as a more robust store of value than any digital asset today, where there is no correlation between currency inflation and market pricing, leading to sudden one-time spikes and erratic volatility versus steady, scalable value growth
ii. It rewards superior market behaviour. One of the best features of the AI Blockchain is that it could assess how wallet holders and market makers were behaving, and ascribe accounts (and by association strongly connected accounts) with individual weights which would determine how much of the newly mined coins a wallet holder received. In doing this, it would look at factors such as how much liquidity that account brought to the network, how aggressive the account was about trying to quickly profit in the event of market price increases in the currency, to which accounts the account holder was connected with etc.
We can make a Blockchain network with additional AI engineering embedded fairly easily. To do so we would create an integrated exchange-Blockchain network, such a Waves or Nxt. This would make obtaining the desired wallet data and market-making data from each wallet in particular much easier. I will refer to this integrated wallet-exchange-Blockchain as “the Platform”.
3. AI Blockchain Platform
The Platform requires every wallet holder to stipulate a rate of return they desire on their currency. A user may input as much or as little as he wants in interest from 1% up to 100%. In calculating the user’s actual return, the Platform takes into account the user’s history, his behaviour on the exchange (for instance, providing liquidity to newer currency pairs constantly will earn a the wallet holder a better weight) and it compares the interest rate multiplied by the user account’s weighting (economic benefactor weighting) against all the other multiples of user account weightings.
The Blockchain will assess what is likely to happen to its market price over the next calendar month. Depending on how well or how poorly the currency is likely to do, the chain will ascribe a positive or negative exponent multiple to the selected interest rate of the user account.
To visualise this process, assume there are 10 user accounts, each labelled with a letter from A-J. These letters we refer to as accounts. Each account stipulates a rate of interest that will in part be determined by its own weight. Clearly, an account cannot run a deficit of coins for very long, so a logical assumption is that after the relative weighting is applied, the real interest rate (mining ratio of new coins) will not result in a greater than 100% clawback or the account is simply discarded from the data pool.
Therefore, anyone running a negative real interest rate would set their interest rate to a minimum of 1% since increasing the requested rate and multiplying by a negative would merely serve to increase the decay of the account balance. Some accounts vary wildly due to the weightings multiples. Account A, in this example, requested 1% but has 30% account balance erosion.
Most accounts in good standing get roughly what they requested in newly minted coins however, with accounts in excellent standing such as Account I (+0.65 weight) receiving a further 15 percentage points of mining interest.
4. Data & Programming of AI Blockchain
Given that every aspect of the platform would be constantly graded in weights via a series of commands that the Platform used to assess mining rewards distribution, the real economic track record of the Blockchain would also be much more transparent and easy to predict — so much so, that it is hard to think that it would be in any way comparable to any digital asset in circulation today, which is purely best-guess.
Because there is a subtraction of coins from bad economic actors while there is also a distribution of new coins, over time the real inflation adjustments are made in favour of the good actors. For instance, while there is a current gross inflation in the example provided of 8%-9%, because of coin erosion in the higher-weighted accounts, there is in fact net inflation of only half this (since those accounts will continue at the present rate to have disappearing coins).
Although controversial, the disincentive to bad actors who sell voluminously on self-created price spikes and who in the economic interest of themselves purely by hiking up their own interest rates etc. is so strong, while underlying value progression is so emphatic, that the combination of these factors would resolve the major problem for digital asset propagation as a mainstream asset class outlined in 1.
Such commands are intuitive and would involve market data analysis, some form of self-reflective analysis that would allow for sufficient flexibility so that unanticipated side-effects did not derail a more productive market (e.g. if 60% of all bad actors respond aggressively and cause a >10% dip in the price of the coins when countered with >30% decay then cap decay here until the ratio lowers).
The Blockchain’s algorithm can be expressed as follows:
((Pi / 100) * (Ni)) / (m) * (Wa / Wn)
where Pi /100 is the AI-calculated price increase or decrease expressed as a % of the coins mined, Ni is the number of coins in issuance now, m is a one-minute interval moving average, Wa is an individual account’s economic beneficiary weighting ascribed by the AI and Wn is the cumulative network economic beneficiary weightings of all accounts combined ascribed by the AI.
5. Benefactor Weightings
A crucial component of the envisaged AI Blockchain is the use of economic benefactor weightings. These are qualitative and quantitative, constantly shifting weightings ascribed to each individual account that allow the AI to determine predominantly who gets what (or loses what) amount of coins on a minute-by-minute basis.
In terms of ascribing the economic benefactor weight of a given account, the AI will individually assess this, although would take into consideration which other accounts the individual account regularly send and receives currency from on the platform (if any) and their associated weightings. Primarily, the AI’s objective is to ensure:
i) Improvements in distributed exchange liquidity across all asset classes
ii) A strengthening currency price in the AI Blockchain issued currency as supply is steadily increased
iii) A reduction of inter-week pump-and-dump price spikes and collapses (meaning +10% / -10% within a one-week-long period versus other external data APIs report)
iv) An increase in net users both transacting with an individual account and on the platform generally
v) A consistent track record of forecasting and achieving currency price increases in the domestic currency around 2–5 times over the amount at which the AI Blockchain is diluting its currency
vi) An increase in liquidity of roughly 2–3x following any currency reduction exercise
vii) To consistently run a current account surplus so that the AI does not have to “trim back” accounts in order to fulfil a mining quota
and will thus weight according to these factors in addition to notable trends and events that are supportive of any / all of these principles.
6. Distributive vs. Subtractive chains
Redistributing assets may be a more desirable method of punishing bad actors than outright seizure and destruction of them. We can hypothesise two types of AI Blockchain: distributive and subtractive chains. Distributive chains seize assets from low-weighted accounts and pass the coins to higher-weighted accounts, whereas subtractive chains merely burn the seized coins.
It’s easy to grasp how a subtractive chain would work, but what about a distributive chain? To visualise this imagine that we grade the top 40% of weighted actors and allow for a 60% distribution of the holdings seized to those actors, with the top 10% getting 30% of assets, the next 10% getting 15% of assets, the subsequent 10% receiving 10% of the assets and the final 10% receiving 5% of the seized assets. Distributions will be calculated at the moment of asset seizure.
The AI Blockchain program will retain the other 40% for its own account surplus requirements.
7. Intuitive Smart Network
It is possible to have features on the Blockchain which are only open to users who attain certain weightings, and even permissions within those features subject to correct use histories. An obvious example is that of smart Blockchain network. One of the problems with the impact that Ethereum’s network has had on digital asset markets is the sheer quantity of “junk asset creation” that has been undertaken due to the relatively quick and easy method the Blockchain offers users for creating digital assets. Digital assets are first and foremost, a form of capital asset.
If you allow anyone and everyone on a network (environment) to create capital assets of their choosing and allow the situation to run to its natural extent then you are bound to have a degradation of value on the network over time as the quality average of capital is reduced substantially by the proliferation of these junk assets.
Value is a very tenuous concept, and erosion of even very high-quality value assets can happen at times in markets where there is substantial low-grade capital constantly filling up the same context.
For this reason stock exchanges, retail chains, and even high-quality serviced apartment operators tend to bias customers and products that contribute somehow to the perception of quality that their brands are trying to attain or maintain.
By utilising the weighting criteria, the AI Blockchain can assess whether a market-maker is “asset creation worthy”. In this particular case it may not look at the total weighted criteria in the balance but may put preferential consideration on the amount of liquidity the market maker provides to the platform, the consistency of the liquidity, the number of counter-parties in OTS status the Market Maker deals with etc. as well as the regularity of profitable trading that is undertaken by the market maker without either causing or participating in extreme price volatility in the process of taking profits.
Once the asset is issued, the AI examines very closely how liquid the issued asset is, the number of holders over the length of time the asset is distributed, the amount the holders trade the asset or move it between “known” separate user accounts with similarly favourable weightings and so forth.
In this way a high-quality digital capital asset market is allowed to develop on a smart Blockchain protocol in a decentralised manner consistently for the first ever time.
8. Summary of Product Benefits
It is clear that an AI Blockchain with an account trading and asset storage platform divided between regulated and unregulated digital assets could become a fin-tech standard relatively quickly and would not involve an overly-extensive R&D period before coding of the platform commenced with a competent engineering and UX team.
One of the advantages of the Platform is that it naturally incentivises market makers to position themselves on the Platform, operating either over the exchange or on an OTC basis. Data from the exchange and the wallets can be captured via a standard Block Explorer API and coins too can be mined from the Block Explorer in the way that they are for regular POS chains. The key difference between this AI Blockchain and other Blockchains is that the running mining algorithm would be derived on a one-week minutely moving average basis from an adjoining Oracle which would not be a decentralised software.
In this sense the Blockchain is neither a public or private Blockchain but more of a decentralised automated scripting program run on top of a standard layer 1 Blockchain technology.
The economic beneficiary weighting process and constant value ascription to such processes is long overdue in the trading of digital assets and a successful prototype would likely attract major institutional investors over time, forming the central point of exchange for such assets over time for much larger volume players. Meanwhile, smaller retail players will find fairer and more cohesive pricing on the Platform that does not penalise them for their smaller investment holdings.
Note: The original paper is here.
Intelligent Assets and AI Blockchains was originally published in Hacker Noon on Medium, where people are continuing the conversation by highlighting and responding to this story.