On-chain data cooperatives reach mainnet
Introducing a new protocol for community-governed datasets
Imagine this: George, a machine learning engineer, sinks into his seat, takes a sip of coffee, and connects to his favorite coffee shop’s WiFi network (via VPN of course). Knees weak, palms sweaty, he can’t wait to explore a new idea he recently conceived. MRI images alone provide an incomplete picture of how the brain works, but could multimodal deep learning across other types of data characterizing neural activity — such as EEG, MEG, fMRI, etc — unlock interesting new insights? He didn’t know, but he sure was eager to find out.
- Instant access to a cloud computing instance — check.
- Finding open source machine learning models to adapt — easy.
- Getting said data to experiment with — stumped.
A couple of years ago, I had this exact experience, and many others I know have had similar ones as well. It got me thinking. If we can unlock the world’s data, how much more innovation, progress, and productivity can we drive by unlocking the innovators themselves? While not the only reason, this question heavily influenced the core thesis behind why we started Computable in the first place.
Data has a governance problem
The Internet simplified access to high performance and affordable computing resources. Open source software fostered a culture of sharing the latest advances in algorithms. Yet, neither really solved the fact that data remains in silos. Which makes sense because data introduces a wholly new set of unique challenges that open source culture and the Internet can’t solve alone. For one, not all data are created equal; namely, the usefulness of data hinges on standardization, quality control, and labeling. Furthermore, those who possess valuable data are often those who feel most incentivized to hoard it, even if that value can be better derived by sharing with others. (For example, the scientific community’s growing willingness to share data has already started generating important novel findings that might otherwise have been discovered years later.) Finally, technology-driven capitalism has raised serious societal questions about how to best respect user consent and privacy. In a nutshell, data has a serious governance problem. We need a more cooperative structure that incentivizes data sharing while also fairly distributing value and control.
Today the Computable team is elated to share a new milestone towards that vision — Ethereum mainnet alpha deployment of the first decentralized and on-chain data cooperative.
Introducing on-chain data cooperatives
A data cooperative is a collaborative dataset bound to a common use case and supported by a community of participants. An on-chain data cooperative is one that is governed via smart contracts — in this case, the Computable protocol.
Smart contracts combine the best attributes of the Internet and open source software to help us reimagine governance. They foster transparency and security by implementing permissionless governance functions in auditable open source code. We’ve seen this already with how projects like MakerDAO and Compound Finance are pioneering a new financial technology stack centered on open participation. Anyone can get fair access to the same lending products and vote on the parameters governing them, e.g. stability fees, savings rates.
The Computable protocol brings similar attributes to the world of data. In particular, any member of a data cooperative can unilaterally opt in or out of sharing, receive fair compensation when their data is used, collaboratively curate the quality of a dataset to boost its collective value, and vote on how that value is distributed via tokens. The result is an egalitarian yet market-driven approach to incentivize data sharing by both producing financial reward and tackling trust issues around data.
What’s next
Our mainnet alpha is the first deployment of our contracts, but not the last. By construction, a new set of contracts (or a new data cooperative) can and should be launched for each application with a unique focus. After all, governance would and should look differently for a mapping dataset versus a rare disease one — in terms of how data is standardized, curated, and rewarded. Different data cooperatives will likely also coalesce unique social communities around their unique causes. We have already begun market experiments to explore what some of these applications and communities will look like. Ongoing work is also happening to create rich user experiences for community participation — a significant challenge facing decentralized application development in general.
Stay tuned for more updates. In the meantime:
- Check out the deployed contracts.
- Learn how the protocol works here.
- Review our security audit.
- Read the whitepaper.
- Questions, comments, feedback? Get in touch via Twitter, Discord, or Discourse!