The Oracle Problem as Applied to Social Impact and Impact Tokens
“Oracles retain an enormous amount of power over smart contracts in how they are executed because the data they provide determines how the smart contracts execute. Therefore, data feeds from third-party sources give that data substantial influence over the execution of a smart contract, removing its trustless nature as part of a decentralized network.”-Medium
Overview
This post is not a full explanation of Oracle Problems and their effects across different token/smart contract applications, that can be done on other venues like this overview and this analysis. In short, the Oracle Problem introduces a single point of failure into a system whose entire purpose is to ensure there are no single points of failure. This single point of failure has different implications depending on the application, meaning the risks introduced by the Oracle Problem in Defi are different than those introduced in Social Impact.
As more and more social problems are addressed through the use of Impact Tokens, the value of these tokens will be determined by the materialization of benefits to specific actors. The quantified definition of these benefits serves as the Key Performance Indicators (KPI’s) for the Outcome State and the automated minting of the Impact Token via the Smart Contract. If we consider the KPI’s to be the Oracle output, the complexity of the relationships measured by the KPI’s require transparency on the level of certainty in their measurement. In the field of international development, this is partially addressed by the use of confidence levels (for statistical calculations) and stating methodological limitations (with implications for interpretation of results) for relevant measures.
Similar practices could offer lessons in how the Oracle Problem is addressed in the tokenization of social impact through the use of Impact Tokens. But it also highlights additional methodological and decision-making gaps that need to be addressed in order to ensure this single point of failure does not alleviate the value-add of the use of an Impact Token.
Assessing Oracle Performance
For the smart contracts that mint Impact Tokens, the single point of failure introduced by the Oracle should be addressed by rigorous protocols accumulated evidence from relevant technical fields. For example, if an Impact Token is meant to alleviate water borne illness, then WASH experts should inform the design, measurement and improvement of the WASH Token. WASH Token design should have an accompanying Learning Agenda that is informed by the application of WASH Tokens to address various WASH related problems, with evidence driven quality controls for assessing Oracle performance based on their ability to accurately measure and communicate the accumulation of benefits.
The Inherent Un-Certainty of Social Problem Solving
In general, we know how to build a bridge (meaning more often than not bridges fulfill their function as planned). The physical laws that underlie the engineering principles and methods to construct and measure the “health” the bridge are highly predictable. This is why you don’t hold your breath when you drive across a bridge as you feel certain you will reach the other side.
We have a very limited understanding of how to solve social problems like domestic violence, poverty, food security, etc. Social problems stem from social systems and the human actors that comprise them and thus far, human actors are the most complex thing in the universe. Human behavior is highly un-predictable and difficult to account for in designing and testing solutions to these problems. While there is a significant lack of will to address these social problems, there is also a significant lack of evidence in how to address them.
This complexity, and the un-certainty it causes, has implications for Oracles providing inputs for Impact Token smart contracts. The primary risk of the Oracle Problem for Impact Tokens is that the level of probability that the Outcome State has been achieved is not correctly measured or communicated, creating a lack of trust that the value of the Impact Token correctly reflects the materialization of benefits that defines the Outcome State.
The Oracle Problem as Applied to Social Impact
Social Impact is fraught with measurement problems that will be inherited by Oracle providers for Impact Tokens. Social behavior is highly complex and un-predictable and the benefits that accrue from simpler social interventions like vaccination or de-worming campaigns, can yield differing measures of resulting benefits. For example, the Worm Wars demonstrate the difficulty in duplicating the same benefit measures from the same type of intervention using the same methods of measurement on a highly “predictable” intervention (de-worming). Hence even with simpler interventions with conceptually higher levels of predictably, there is wide variance in measures.
If Impact Tokens are to be used for more complex social problem solving, then the un-certainty that is inherent to those problems must be accounted for. Consider measuring the costs and benefits that materialize from two assets, one a bridgeand the other a tree in a forest.
Cost-Benefit Analysis is a useful model to assess possible solutions to problems by measuring the anticipated cost of the solution over its lifespan in relation to the anticipated benefits that accrue from the solution over the same lifespan. For a hard infrastructure asset like a bridge, the primary costs are initial construction and maintenance/upkeep over a lifespan where the costs are high in the beginning, plateau during the middle, then increase at the end as the bridge needs more and more repairs to remain functional as informed by direct measures of the integrity of the bridge.
The benefit streams are quantified in time and costs savings from travel, increase in economic opportunity, improved access to services, etc. The benefit streams are largely the result of simpler human behavior, people taking the bridge to transport goods, to go to the new job in the city, etc. While mistakes happen, bridges tend to be fairly predictable solutions to various barriers like cost prohibitive access to goods and services. This is partly due to the relatively simple direct measurement methods for determining the state of the bridge and its maintenance needs. Measuring benefits can be done directly through standard traffic studies, time/cost saving calculations, etc.
Measuring the cost/benefit of a forest is more difficult. The level of carbon offset of a tree is dependent on the health of the tree and its level of maturity. Both of which are dependent on the health of the surrounding ecosystem, etc. as trees offset varying levels of carbon depending on their growth lifecycles.
The measures primarily include less direct proxy measures (soil samples, atmosphere readings, growth coverage, etc.) that are then triangulated to produce quantified measures of the benefits accruing from individual trees. But these measures are caveated with limitations for their use in decision-making and levels of confidence given so that the end user knows the probability that the benefits have actually materialized.
If each asset were tokenized, the bridge would have a greater reliability in the measurement of costs and benefits while the Oracle provider for the forest would need to caveat their measures with confidence levels to communicate the probability that the measures are reliable.
It is not that an Impact Token for trees is impossible, it just means that it has to account for the level of certainty in the measurement of the benefits.
First Principles for Impact Token Oracles
There is a need for an evidence driven protocol for selecting and assessing the performance of Oracle providers for Impact Tokens. This protocol must be built on initial principles that are then integrated into criteria and processes which are co-created by Social Impact stakeholders (sector experts, decision-makers, measurement providers, etc.), token engineers and perhaps most importantly, those directly involved with the problems at hand (the communities and actors that contribute to or are harmed by the problem).
Principle #1: The Key Performance Indicator for Oracle Performance is the level of confidence that the benefits of interest are correctly measured and Communicated. This Indicator can only be calculated by assessing Oracle data sources, methods of collection, analysis and reporting using the standards of the relevant technical fields adapted to the use of tokens.
Principle #2: The Value of Impact Tokens is directly correlated to the value of the benefits that result from solving the Problem of Interest. The transparent communication of confidence levels (and the method of their calculation) in the Oracle outputs is necessary to ensure that the value of the token as reliable as possible.
Principle #3: Solving the problem of interest is the only function of the Impact Token. Social problems are solved through the creating new benefits for specific actors in specific places at specific times. This formula sets up how problems of interest should be defined and their alleviation measured by the Oracle.
I would be grateful to any who add to this list and want to contribute to the conversation.
