Considerations When Designing Blockchain Oracles

This post is part of Researchly’s weekly Blockchain and Crypto report covering ICOs, dapps, and news sentiment. You can read the full ICO report here, the full news report here, and the full Dapp report here.


Last week, Kleros finished their ICO. Kleros calls itself a “Dispute Resolution Layer” and has raised around $3 Mio. In essence, a blockchain oracle focusing on disputes. A user submits her dispute, selected jurors vote on the case and the majority vote decides what the truth is. One use case Kleros proposes is arbitration. If two parties that organized a freelance job over a decentralized platform cannot agree on the output of the work (e.g. did the work meet defined quality standards), Kleros could be used to solve the dispute. Coincidently, this is something I pointed out in last week’s report regarding Latium, a task platform;

The most important part of task platforms – ensuring task quality – lies outside their control. Blockchains provide several concepts to manage such intermediary-less task platforms (e.g. token-curated registries for reputation management) but the biggest issue – verifying task completion – remains. Although oracles could be a solution here…

Several other blockchain oracles already exist. By the way, one of them – Augur – went live last week and was one of the most popular dapps last week but also this week, based on DappRadar. Kleros puts an interesting spin on oracles by narrowing them down to disputes (even though the voters are not necessarily real lawyers).

And whereas Augur’s relative strong user demand indicates a market for blockchain oracles, they must be treated with caution.

Justifying the overhead

In last week’s report, I argued that “Although blockchain oracles could be a solution here I am unsure whether this would justify the overhead.” In the case of Kleros, this means that although decentralized settlements could be fairer than central judges, in many cases they won’t justify the overhead (e.g. for small arbitrations).

Avoiding groupthink

Secondly, blockchain oracles do not provide truth, they just reflect what everybody voted. The emphasis is on voted and not believes because those two might differ. As judges in blockchain oracles like Kleros are financially involved (they gain financially if their vote is in the majority vote), group thinking is encouraged and jurors are likely to vote based on what they believe others will vote.

Selecting the right areas for blockchain oracles

Finally, alongside overhead considerations, blockchain oracles must be used based on the evaluated topics. Not all topics are equally suited for decentralization. Settling provable predictions where there is an objective truth like with Numerai for machine learning predictions works very well. Settling, however, disputes with a subjective truth (e.g. work quality) or where bias is present (e.g. racism) is less likely to work.

Despite the short timeframe, Augur’s relatively high participation is a good indication that there is a need for blockchain-based oracles. Thus the question is not will they work but where will they work (e.g. topics with object truths, where they justify the overhead…) and how to design them (correctly incentivizing voters, preventing groupthink…)

Sign up for Newsletter: