Power stems from things beyond the promise of money and the threat of guns these days.
Intelligence -- the ability to take large amounts of data and turn it into meaningful insight -- is becoming necessary to compete on the global stage. Facebook, Google, Cambridge Analytica, the CIA: all of these groups possess not just a huge amount of data, but moreover, offices filled with analysts selling their intellect to corporate power.
Blockchain technology is not just here to decentralize finance. Ambitious projects are hoping to democratize our access to artificial intelligence. This means not just accessing data sets like those that Tesla's self driving cars draw upon; but also, to be able use that data in powerful algorithms, affordably and on demand.
The Ocean Protocol and SingularityNET are just two of the projects that are taking on this monumental task. Both have already had successful initial coin offerings. There is a lot of overlap among their advisers, indicating a cooperative rather than competitive strategy. SingularityNET was previous covered here; this article will focus on the Ocean Protocol.
In the words of Trent McConaghy, one of the founders of the Ocean protocol:
The lifeblood of modern society is data, yet data is held in centralized silos like Facebook. This means they wield more power than most governments; yet we can't elect them, and don't have any real control over the decisions they make [...] AI loves data and so companies are incentivized to get lots of data and hoard it [...] to maximize profit while holding competition at bay.Ocean's approach: change the incentives at the heart, away from hoarding/siloing data and instead to sharing data / making it available.We're doing this using blockchains as incentive machines.
The Heart of Ocean
The Ocean Protocol is registry of data sets and algorithms, based around a proofed curation market. This token system evaluates the system's assets and incentives high quality data sets and intelligent algorithms.
The "proofed" part refers to cryptographic proofs about the popularity of an algorithm or dataset: how many times has it been used?
"Curation" refers to an increasingly popular way of using the wisdom of the crowd to determine the value of something. Similar to WINGS or Gnosis, users can stake their tokens on the data sets and algorithms they believe will be the most useful. If their predictions are right, they receive a share of the proceeds, and this helps elevate the best content, or at least the most desired.
The combination of these two mechanisms is not just a way to provide these services without relying upon a centralized organization. It also allows people to upload data sets and algorithms for free and still receive a profit from it (by betting on its popularity). Of course, for profit algorithms and data sets can also be entered on the list, and users of the network can pay in Drops for access. Because the data sets are not stored centrally, sensitive information can remained privately siloed.
The paper points to a handful of use cases to demonstrate the utility of their system. Self driving cars are the first example, and the BigChainDB team (of which there is significant carryover to Ocean) has already been built a prototype to this end. The better chunk of a trillion miles is estimated to be necessary for the safe operation of self-driving vehicles. Having each motor company run the same trillion miles is one of those wasteful exercises of capitalism that can probably be avoided with a little cooperation. But if any company gave that data set to a competitor, they might copy it and sell it to even more corporations. The Ocean Protocol is a way of preserving ownership since the data does not actually leave its home.
This secure way of accessing big data is also anticipated as being a break through medical prediction. IBM's AI called Watson, for instance, is able to take in the same diagnostic data as a physician, compare it to previous cases, and then make a call on whether or not a patient should pursue further diagnostics. In other words: replacing most of what oncologists do. A huge amount of medical data is needed in order to give this system intelligence. But that data is nearly worthless without complex machine learning algorithms to drives those recommendations. The Ocean Protocol means that smaller companies will be able to provide similar services on demand, without having to negotiate costly licenses with IBM and the medical record providers. Eventually, this might mean that our personal A.I. assistant are able to make informed recommendations about our health, allowing human doctors to focus on those who need help the most. Eventually it seems that a hybrid/human system becomes the most effective solution.
The Ocean Protocol is more a marketplace than a provider. Just as the data sets will be stored in many formats in many places, the algorithms will be processed on other systems such as Amazon Web Services (centralized) or Golem or i.Exec (decentralized). Specialty systems like SingularityNET will provide AI services, and privacy-centric processes can be performed on Enigma.
In order to post content to the network, an actor must have some tokens staked. Furthermore, each asset that is added is verified to insure that it was not lifted from another source. Until the data set or algorithm has been verified, any funds earned from its use are held in escrow.
Ocean is largely based on existing technology. It is an evolution from Ascribe, a platform for maintaining declaring intellectual property, and BigChainDB, a system for storing large sets of data with some of the benefits of blockchains.
If you want to learn more, check out Ocean Protocol’s whitepaper here.
Concerns & Conclusion
I asked Trent McConaghy, one of the project's core founders:
Can you envision any societal dangers that might stem from Oceans?
[...] The answer is maybe, but we've gone out of our way to mitigate the danger.Ocean is designed to optimize an objective function: maximize the supply of relevant AI data & services. It does this by rewarding people who curate with stake and make data/services available when asked. We went out of our way to design this objective function thoughtfully.But there's always a chance that the Ocean objective function won't act well, since they're very hard to get right in both the AI domain and the blockchain domain. Case in point: Bitcoin is on track to use more energy than all of USA by mid 2019! That was almost certainly not the intent of Bitcoin's creator(s).What's our fallback? Good enough governance such that we can change the objective function as need be, in the early days.