asciilifeform: btw the whole homomorphic business reminds me of an old sf story, where some malefactor stumbles upon a pair of large integers whose multiplication doesn't commute.
asciilifeform: it's a perpetuum mobile, but that never stopped the desperate idiots with infinitely deep pockets, has it.
asciilifeform: the inevitable boojum of conventional computing is that a boobytrap, no matter how clever, could eventually be stumbled upon by some clever soviet kid with a copy of 'ida'
asciilifeform: they would like to create (and then mandate) computing machinery that cannot be audited for security even in principle.☟︎☟︎
asciilifeform: in principle, 'all blades have two edges,' etc. but in practice, the one and only objective of the folks funding homomorphic is a kind of satanic opposite of my purpose outlined in 'don't blame the mice'
asciilifeform: hence, if someone asks you to run homomorphically cloaked code...
asciilifeform: it is thus useful to know how to tell which end of the barrel you are being placed on.
asciilifeform: correct. ideally one should like to be on the 'correct end of the barrel' for this one.
asciilifeform: and this fact not being deducible from the code.
asciilifeform: a 'soft pgp' that appears to work fine, but when you encrypt a message to a certain pubkey, the message becomes 'eat shit and die'
asciilifeform: but nobody seems willing to say it publicly
asciilifeform: i venture to say that the actual goal of general-purpose (or whatever approximation is possible) homomorphic crypto is quite different.☟︎☟︎☟︎
asciilifeform: if it 'keeps key safe' via the whole proggy becoming the key, you haven't accomplished much
asciilifeform: but for the subject of the article, one is left to wonder what problem is actually being solved
asciilifeform: the building blocks of homomorphic crypto (say, the 'millionaires protocol') are certainly interesting, not only in the mathematical but practical sense.
asciilifeform: anyone who wants 'software cardano' can grab 'gpg' today.
asciilifeform: i live right outside of a big uni, whose comp sci dept. pisses out a never-ending supply of this crud.
asciilifeform: the whole field appears to be funded by the desperate 'media' conglomerates, who dream in their opium haze of unbreakable copy protection.
asciilifeform: so you're stuck with some special-case exotica, that idiots will inevitably embed in conventional code.
asciilifeform: there is already a hard impossibility proof of turing-complete computation in a homomorphic turd.☟︎
asciilifeform: so fine. i extract the turd and feed it whatever i like
asciilifeform: and on top of that, is only willing to sign giblets beginning with 1.
asciilifeform: so let's say you have a homomorphic turd that signs arbitrary giblets with your key (which, for the sake of argument, assume cannot be extracted, as claimed)
asciilifeform: naturally, we have the boojum: '...the attacker would be able to *use* the secrets only in the way that the software allows...'
asciilifeform: 6) All of this is based on mathematical computational assumptions, which may be proven false by future algorithmic advances.
asciilifeform: 5) Obfuscation is really a terrible name, but unfortunately it is the mathematical term that has become widely used in the academic (non-hacker) community.
asciilifeform: 4) In no direct way does secure obfuscation have anything to with DRM, software copy protection, etc.
asciilifeform: secrets only in the way that the software allows, but not recover these secrets in any way beyond that.
asciilifeform: 3) The right way to think of what secure obfuscation allows is to create, *under many technical conditions*, software that has secrets built into it. These secrets are used by the software to compute output, and yet the secrets remain hidden even if an attacker obtains the entire machine-level code of the software, which of course the attacker could run and analyze. Thus the attacker would be able to *use* the
asciilifeform: 2) Secure obfuscation is a mathematical term of art. It is unfortunate that the word "obfuscation" has an ordinary meaning that is very different than what we mean. An analogy might be the word "countable" which for mathematicians usually refers to infinite sets, while lay readers would likely not think this way.
asciilifeform: 1) The title is misleading: secure obfuscation does not necessarily create "unhackable" software, whatever that means.
asciilifeform: I am a co-author of the research quoted in the article. I realize it is likely pointless to write this comment in an Internet forum, but let me attempt to clarify the situation, in the hope that some thoughtful participants will find it useful. Please note: I will not be monitoring this forum in the future. If you are a researcher, please feel free to contact me.
