It's hypothetically possible to come up with versions of algorithms that distribute the processing of your data such that each node does not get enough information to recover secrets, in a cryptographically secure fashion.
As a developer though in 99% of cases I'd much rather just pay for a few servers in multiple datacenters and load balancing than waste time researching the above, let alone proving its cryptographic security for my use cases.
It could, however, be useful for cases where you're trying to get over foreign government censorship or oppression against whatever service you're trying to provide. Imagine, for example, a free speech discussion service that was decentralized and essentially un-blockable. Bonus points if its cryptographic system can be designed such that an arbitrary processing node can never be proven from its data alone that it was helping run the service (among other services).
It's already possible to execute some algorithms on untrusted computers using homomorphic cryptosystems. I'm not sure what to what extend this will proceed - it might be possible to securely execute arbitrary algorithms on untrusted platforms. This is sort of a holy grail of cryptography - to be able to process secret information efficiently without gaining knowledge of it.
I agree it has it's cases in a Tor sort of way, but in terms of "replacing the internet" which I think was the original title of this HN post, I don't think it's feasible.
I'm not outright dismissing the technology or the thinking behind it but it certainly wouldn't fit a majority of use cases as we use the internet now.
Comments
It's hypothetically possible to come up with versions of algorithms that distribute the processing of your data such that each node does not get enough information to recover secrets, in a cryptographically secure fashion.
As a developer though in 99% of cases I'd much rather just pay for a few servers in multiple datacenters and load balancing than waste time researching the above, let alone proving its cryptographic security for my use cases.
It could, however, be useful for cases where you're trying to get over foreign government censorship or oppression against whatever service you're trying to provide. Imagine, for example, a free speech discussion service that was decentralized and essentially un-blockable. Bonus points if its cryptographic system can be designed such that an arbitrary processing node can never be proven from its data alone that it was helping run the service (among other services).
It's already possible to execute some algorithms on untrusted computers using homomorphic cryptosystems. I'm not sure what to what extend this will proceed - it might be possible to securely execute arbitrary algorithms on untrusted platforms. This is sort of a holy grail of cryptography - to be able to process secret information efficiently without gaining knowledge of it.
I agree it has it's cases in a Tor sort of way, but in terms of "replacing the internet" which I think was the original title of this HN post, I don't think it's feasible.
I'm not outright dismissing the technology or the thinking behind it but it certainly wouldn't fit a majority of use cases as we use the internet now.