In 1965, while working on transistors in dense integrated circuits for Intel, Gordon Moore proposed a theory based on his observations. The idea that technology would get twice as fast and half the price has held true much longer than the decade that Moore himself predicted it would and has been observed in all forms of technology, so much so, we turned his theory into a law.
However, due to the sheer physics of the microchip, many fear we are fast reaching the limitations of Moore's law. And since our computing requirements and need to process and store data grows every year, many have begun to start thinking of solutions to the potential impasse being approached.
One of those people is Ivan Ravlich, CEO and co-founder of Hypernet, a parallel programming model. We spoke to Ivan about how his model is superior to the cloud and how they may have found an answer to our core data problem before it became a crisis.
Visit Innovation Enterprise's DATAx New York on December 12–13, 2018
Innovation Enterprise: Is Moore's law still relevant today?
Ivan Ravlich: Moore's law states that the number of transistors in a densely integrated circuit doubles about every two years. With transistors now reaching their quantum limits in terms of size, Moore's law could be quickly approaching obsolescence. However, this does not mean that computers will stop getting faster. Since there are minimal gains left to be made in terms of transistor density on a chip, chip producers are now focusing on increasing the number of cores. By increasing the number of cores, they increase the number of parallel tasks that can be conducted on a device. This means that during real-world use, devices will seem faster, even if their transistor density remains unchanged.
IE: How will Hypernet get around the limitations of Moore's law?
IR: Hypernet's innovation is that we have created a privacy-preserving protocol which collects all of the cores mentioned above and connects them in a way which allows them all to be utilized in complete harmony. This is revolutionary because until now, no one has demonstrated an ability to do this over a dynamic and heterogeneous network of devices.
IE: How do you ensure the security of the latent devices Hypernet utilizes?
IR: The distributed nature of Hypernet's algorithm means that privacy is built into the computation layer itself. Instead of data being centrally located and a hacker only needing to hack one machine to access all the data, with Hypernet, that data could be computed across thousands of devices and is computed on the device. This means that in order to gain access to all that data, a hacker would need to breach each of the thousands of devices, individually. This would be so inefficient, that it wouldn't even be worth the effort.
IE: How much knowledge of blockchain will users need to get utilize Hypernet?
IR: Hypernet's technology may sound complex, and indeed it is, but our goal takes all of this and place it behind a simple and elegant user interface. We want Hypernet to be so simple that a child could teach their grandparents how to use it. This means the users will need no background knowledge of blockchain or cryptocurrencies work. All the complexity will be handled on the backend.
IE: How will you recruit users to participate and allow you to utilize their devices? Will there be any benefits?
IR: The simplicity of our product/protocol will naturally be helpful in attracting new users to join the network, but there will be other reasons to join, too. Of course, users will get paid to put their devices on the network, which will be a main attraction. However, we're also excited about implementing our ideas to create a social computing environment in which people want to engage with simply because it is fun to do so.
IE: What are the benefits of this system versus the alternative computing systems like IoT?
IR: The ability to compute data in a private manner is the main benefit. The other advantage includes massive savings over cloud computing infrastructure. Since Hypernet utilizes people's individual devices (which they own and take care of) it does not have to pay for the infrastructure costs associated with cloud computing. Since infrastructure (e.g., land, buildings, taxes, computers, air conditioning, maintenance, employees) account for 80% of the cost of cloud data centers, Hypernet represents an extreme increase in value. There are several more tertiary advantages, but the main ones are privacy and reduced cost.
IE: How do you envision Hypernet and similar technologies will transform the world of computing?
IR: Overall, the grand vision for Hypernet is to put individuals back in control of their own data. Being able to run large computations on data that never has to leave its location has massive implications for data collection, privacy and ownership. For example, imagine if Facebook never had to store your personal information, but it could still function and generate profit in the exact same way it currently does. It could provide the exact same services it does right now, but it would not need a single data center to store any of your private data, because that data would be housed on your own secure device. Or, what if hospitals had a way to privately and securely analyze patient healthcare data in a HIPPAA compliant manner? We could greatly enhance care and establish very robust markers for every type of ailment.
To learn more about how to best separate the signal from the noise, visit Innovation Enterprise's Machine Learning Innovation Summit in Dublin on November 29, 2018