Banks and other operators of sensitive data may like the efficiency of blockchains, but not the lack of privacy that goes with them. And a bunch of banks operating a closed private shared ledger system still require that their data is to a large degree kept private from one another – a point which is often overlooked.
Oz Nathan and Guy Zyskind, the creators of MIT's Enigma, have developed a guaranteed privacy system, where data is stored, shared and can be analysed without ever being revealed to any party.
Nathan told IBTimes UK: "Together we formed this decentralised super-computer and every computation is done between different participants in this network and in that way the actual data is never revealed to anyone."
This is has been accomplished through a distributed network which is inspired by Bitcoin, said Nathan. "It's different computers that are talking to each other, but they don't do mining, they just provide resources to the network, bandwidth, some of their hard drives, some of their CPU power".
To break the system would mean taking control of all the servers on the network. It can work alongside any blockchain - Bitcoin, Ethereum, or private chains that banks want to form.
Nathan added: "One of the banks I have spoken to told me even a private blockchain doesn't help them because even if they do a blockchain between ten banks, they still can't have their competitors know certain transfers and certain clients that they have."
Commercial applications of blockchain technology on the horizon are driving the need for privacy among those entities that will share ledgers. Blockstream recently announced some privacy-enhanced sidechain elements, to work between exchanges on the Bitcoin network.
Generally speaking, concerns over security are on the increase, while data is becoming more and more powerful, meaningful and valuable. If people have guarantees about their privacy and their security, they will likely share more data with companies; companies will share data with other companies, which will create more business value.
Zyskind is currently involved in a blockchain workshop about privacy and identity with some banks and some people from very large medical companies. He said: "Medical companies are saying, well we can't put everyone's medical information on the blockchain, and the banks, they can't put everyone's finances on the blockchain and so forth.
The Enigma system breaks down data into pieces and also masks it using some clever mathematics. "Encryption is not the right word: it's called secret sharing, and it guarantees mathematically that each of these pieces are completely masked, completely random and completely secure," said Zyskind.
"There is no way to infer anything from looking at each of these pieces alone; you can't get anything about the original data back. So it's actually even better than encryption because it gives you what we call in security a 'perfect secrecy', which is the best type of security that you can imagine."
The computers interact without compromise to the data, and the system allows them to execute any code, do any processing on the encrypted data.
Nathan said: "Collectively we let a large network do the computation instead of just one computer, which is also better for integrity, and for resiliency. If you want to break encryption, you would need to get the key; if want to break Enigma, or blockchain by the way, you would need to get control over all the servers. So instead of just getting the key, you would need to get control over all the servers, and then collectively you can reconstruct the original data."
Nathan pointed out the Enigma's system differed from homomorphic encryption, whereby computations are carried out on ciphertext, thus generating an encrypted result.
He said: "Right now the best [homomorphic encryption] is a million times more expensive. What we are doing is one order of magnitude, maybe about ten times more computationally expensive. Also, using our system you can actually compute on data that has different owners. In homomorphic encryption you need to encrypt it with the same key, and here you can use multiple pieces, because they don't have a key basically. So you can aggregate data from different users, different companies whatever the use case."
As far as use cases are concerned, Nathan mentioned some specifics such as privacy in oversight: "So you don't have to look at each person's balance but you know if a transaction is above a certain amount, or it comes from certain sources, then you put up a flag and then you know something about it."
Other uses for banks could include things like proofs of solvency, or proving things to the regulators. In such cases, the banks would not have to reveal the actual internals, just the proof that they are abiding by the regulation. Another financial use case could be dark pools that are really obscured, so nobody sees the bids.
Nathan said the system also applies to any type of machine learning that has to do with aggregated data, or just something that's very private so companies can reassure their customers that their data is safe and private and they don't have access to it.
Zyskind added that medical companies often store very sensitive data, which must be kept private from their in-house researchers or other companies or universities are collaborating with.
"They can't just give them the data to do research because it's so sensitive. So if they can run their machine learning models, or predictive models on data while it's secured then that's a game changer."