Data Collection Consensus

Shared accurate knowledge of emissions, resource use, or resource production data is necessary for a clear climate consensus to emerge. However, voluntary disclosure protocols have largely failed to incentivise actors to implement strong data reporting practices. Without enough commitment from actors to making accurate, up-to-date data reports, accurate shared knowledge of the Earth system state is impossible.
The Carbon Disclosure Protocol (CDP) is a repository collecting emissions data from companies and cities (“actors”). While large, the database remains fraught with several limitations. Data may be uploaded by actors themselves in nonstandard formats, may not be updated year-on-year, may be kept private or may simply be falsified or inaccurate. CDP also collects data directly from actors but does not make their data publicly available, instead charging for access to this “proprietary data”. For data providers and aggregators to clean and make the data usable, there is a high overhead that they are often reluctant to provide their data for free.
Beyond our understanding of the current Earth system state, accurate data is also necessary to track how well the actors perform with respect to their emissions reduction commitments. CDP tracks such commitments, obtained from Corporate Social Responsibility (CSR) reports, public commitments or voluntary disclosure to the platform itself. However, the tracking can be challenging or misleading if 1) reported data is inaccurate or insufficient or 2) reported data is required to be kept private and cannot be checked against commitments.
Three related questions emerge from the preceding discussion: 1. How can we incentivise actors to accurately disclose their data? 2. How can we incentivise actors to make significant and ambitious commitments? 3. How can we incentivise actors to fulfill the commitments they make?
Currently in our proposal:

PreciDatos: Staking on data report (1)

There is no obvious reward for actors to implement robust data reporting practices. At the same time, there is no penalty either for reporting poorly, while actors’ benefit in reputation from the act of reporting itself, regardless of the quality. We propose that actors stake some amount in order to commit to the quality of their data. The stake can come in the form of either cryptocurrency, or with fiat money with an on-chain record of that commitment.
Types of actors In this system, we consider the following three actor types:
  • Honest and committed actors: Data is disclosed accurately and properly.
  • Honest but uncommitted actors: Data is accurate but is not reported properly.
  • Dishonest actors: Actors disclose inaccurate data.
System outcomes
  • For honest and committed actors: The staked amount is returned in full with a reward.
  • For honest but uncommitted actors: The staked amount covers additional data processing steps should they be required, paying for bounties or open-sourcing the work.
  • For dishonest actors: The staked amount is used to produce an investigation into the accuracy of the data when whistleblowing takes place.

Whistleblower mechanism (1)

Data reporting may be falsified by the actors, while whistleblowing may be individually damaging to the whistleblower. The mechanism offers a private, verifiable procedure for an informed whistleblower to act anonymously.

NLP of actor disclosures (1)

Actors may upload ill-formed data, e.g., PDFs or spreadsheets. NLP algorithms preprocess this data to understand its contents and open appropriate bounties for cleaning up. Bounties are paid from the initial stake.

NLP of actor commitments (2)

Holding actors accountable for the commitments they pronounce provides a strong incentive for them to not shirk. The NLP data processing analyses commitments to quantify how concrete the propositions are and how likely they are to be fulfilled.

Homomorphic computation + Verifying against commitments (1,3)

Assuming that the reported data is accurate and well-formed (i.e., what PreciDatos purports to do), we propose a mechanism to obtain aggregate quantitative measures of private emissions data that can be checked against prior commitments.

Out of proposal ideas

Staking on commitments (2, 3)

Companies are incentivised to commit to and fulfill larger emissions reduction. The mechanism should guard against penalising too harshly companies who do not fulfill their commitments, as this would incentivise lower commitments altogether.
Here too we can define a few actor types:
  • Committing and fulfilling: Actors have made public commitments and implemented capacity to fulfill them.
  • Committing but unfulfilling: Actors have made public commitments but did not act on them.
  • Uncommitting: Actors have not made commitments.
Other solutions exist: a token economy supports systems such as Zei, where consumers receive tokens for purchasing ecologically responsible products. Tokens can be used as rebates on future purchases. Zei also tracks companies making voluntary commitments uploaded on the platform (although these rankings do not seem to bear on the token economy).

Emissions reporting use case: Google emissions data

  1. 1.
    Google uploads big PDF into PreciDatos, with a stake.
    • PDF is processed using NLP techniques, extracting as much data as possible.
    • If data is private, homomorphic computation may raise issues.
    • Bounties are opened for final processing steps, if necessary.
    • Meanwhile, any one of the Google verified company members can activate the whistleblowing mechanism to signal data fraud.
  2. 2.
    Stake + rewards are paid back, minus costs associated with bounties/whistleblowing/issue investigation.