Data is the oil of the 21st century. It flows through the veins of the internet like black crude. But, anonymous internet data, is impossible to trust. Information, without a reputation to judge it on, cannot be believed. Also, information can only be judged on its quality, after it has been provided. The Erasure protocol can trustlessly, tap into the messy crowdsourced data flowing through the web.
Richard Craib and team have built two projects using the Erasure protocol, Numerai and Erasure Bay. Numer.ai is a crowdsourced AI, designed to control a hedgefund. It’s predictions are based on machine learning models, compiled by data scientists across the world. Erasure Bay is a data marketplace. Its users can buy/sell ANY type of data! …NOT just market data.
The Erasure protocol consists of financial incentives, penalties, and staking. It can refine quality data, from untrusted anonymous sources. Good data is rewarded. Bad data is penalized. Data is obfuscated, so users don’t even know what they are predicting. Users must stake funds, for the chance to sell data. The financial risk of staking, improves the data quality massively. This system of rewards, punishments, and staking, creates a trustless marketplace for users to buy and sell high quality anonymous information.
In the article below, i’ll discuss Numerai, Erasure Bay, and how they can enable trustless, data brokering on the blockchain.
What is Numer.ai? How Does it Work?
Numerai is a crowdsourced AI, designed to control financial capital. The AI is constructed from crowdsourced machine learning models, uploaded by data scientists from across the world. The AI predicts financial markets, then allocates capitol based on these predictions. Numerai provides its users with obfuscated stock market data, which is used to build machine learning models. The best models are then aggregated, into a superior meta-model. The “meta-model” has a higher statistical rate of accuracy, with a sharpe ratio is as high as 2.09. (image below)
How Does Numerai Turn Anonymous Internet Data, into Trusted Data?
Numerai is able to build a statistically superior stock market meta-model, with crowdsourced data. This meta-model is comprised of 100’s of user uploaded data models. Users compete with their models in Numerai’s weekly data science tournament. The top models are awarded in accuracy and originality. In the tournament, contestants are given blind stock market data, to build their own machine learning models. The data is obfuscated, so users don’t know what they are predicting. They just predict what will happen next. The best models are aggregated into a superior,”meta-model”, which is used to control the hedgefund.
How Does Numerai Reward its Users for Creating Data Models?
Users with the lowest logarithmic loss error on their predictions, get the most money, paid out in Bitcoin. Users provide predictions on live market data, as well as test data. The test data is used to validate the accuracy of their models. Users are paid for accurate predictions on live data. Models are rewarded on consistency, originality, and concordance. As the money being managed by the platform increases, the users will earn larger percentage of the profits.
How Does the Numerai Platform use Staking?
The team found that staking produces higher quality data! When money is at risk, better data is generated. On the Numerai platform, staking can even be automated, with payouts compounding over time. As more tokens are staked, the potential reward increases. Positive and negative economic incentives, weed out low-quality data providers. Auto-staking lets users just focus on just the data science. Profit is generated as a result.
The supply of Numeriere (NMR) tokens is capped at 11,000,000. NMR tokens have value and utility due to the fact that it is needed for staking on the Numerai and Erasure Bay platforms. Higher levels of staked NMR tokens, will generate a higher return for the data scientists. This creates demand for the token. As the platform grows, this should cause the NMR tokens to gain value.
Erasure Bay – the non-financial data marketplace
Erasure Bay is a trustless data marketplace, built on the Erasure protocol, similarly to Numerai. This marketplace grew out of the knowledge that staking increases the quality of crowdsourced data. Erasure Bay users can buy and sell ANY type of data, NOT just market data! This includes: current news, sports, and science. Data sellers must stake NMR tokens. The receiver of the information can destroy the seller’s NMR stake, if the quality is poor. This gives sellers the incentive to provide good data. The data must be accurate, because data providers have their own money at risk! Erasure Bay, if successful, can provide the world with a source of verifiable, true information.
In a world where lies and propaganda rule our tv screens, true information is more valuable than gold. The implication of a trustless data marketplace, extends far beyond just predicting stocks. Richard Craib and team have found a way to extract value, from the raw brainpower of untrusted internet users. The protocol of incentives, penalties, and staking creates a system, that FORCES a user to play by the rules. If they provide bad data, they lose money!
Numerai has an edge in the stock market. Its crowdsourced predictions have a superior rate of accuracy, compared to machine learning models. As the platform grows, and more data models are generated, this edge will increase. Numerai is mum on the exact profitability of its system, but it makes logical sense that more accurate predictions will produce profits.
Blockchain enables many types of anonymous trusted marketplaces. Stock market data is just the start! I’m most interested in the data it can obtain concerning politics, and current events. Maybe we could use this platform to find the truth about the world around us. Numerai and Erasure Bay are interesting projects, with great potential. Tapping the brainpower of the internet, trustlessly, allows us to access true information, and insights we’ve never been able to previously. I think this is a project to watch.
If you liked this article, follow me @defipicks