Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Craig Gentry named MacArthur Fellow for work on fully homomorphic encryption (macfound.org)
152 points by sweis on Sept 17, 2014 | hide | past | favorite | 21 comments


Yes. Yes yes yes.

Craig Gentry did more than just discover the first theoretical scheme, he reopened an interest in FHE. Yeah all the schemes are inefficient and will remain so for decades. But already the progress on FHE is incredible.

2007 for FHE was the 1982 for Secure Multi Party Computation. It will take 30 years, but I expect someday we'll see it privately allocating contracts in sugarbeet auctions.


> contracts in sugarbeet auctions

I highly recommend that everyone reads http://csrc.nist.gov/groups/ST/PEC2011/presentations2011/tof...

Crazy stuff.


I can think of few people who deserve it more. His recent work has been truly groundbreaking. Not just on FHE, either - multilinear maps, IO, lattices... I could go on. Just a brilliant dude.


Interesting stuff. Who are the people you think might deserve it more?


"I can think of few people who deserve it more" basically means the exact opposite of... "I can think of a few people who deserve it more".

I read it wrong the first time too.


Nobody deserves it 'more' but another person who has been doing similarly groundbreaking stuff is Sanjam Garg.


This is awesome. Reading about his work and the subsequent work on FHE in general was one of my favorite "holy shit this is the future" moments. (And I don't have those moments hardly at all ... kind of silly. But I was giddy, and it was amazing, and so I was silly.)

It was also one of my foremost "I will never ever in my life be as smart as these guys" moments. Alas.


Google should be funding the heck out of this guy. Privacy concerns are only going to get bigger in the future, and the reason for that is simple: companies like Google are inevitably going to want more and more data. So at least they should be trying to do that in a privacy-friendly way.


I'm certain ibm is already paying him quite well to sit in a room and stare into space, and will continue to do so for as long as he desires it. Google is not the be-all-end-all of cs research funding.


Why would they willingly deprive themselves of large amounts of user data?


Adapting Ur-Beowulf to Ar-Curunir and Grendle to Google:

"So the Shieldings' hero, hard-pressed and enraged, took a firm hold of the hilt and swung the blade in an arc, a resolute blow that bit deep into her neck-bone and severed it entirely, toppling the doomed house of her flesh; she fell to the floor. The sword dripped blood, the swordsman was elated. (1563-1569)"

Sometimes the better product makes less profit. It's hard for someone with Capitalistic incentives to respond with a better product if delaying/discouraging/regulating-out-of-existence is cheaper.


Oh I would love to create a search engine that would preserve the privacy of its users, but I just don't foresee Google doing that.

I'm aiming at studying Theoretical Cryptography, with an interest in MPC specifically, so this would be right up my alley, but my pessimistic world view tells me that any efforts to create more secure and privacy-preserving services will fail because of institutional opposition and low market demand.


We're on the same page. There's been tons of great work in the past couple years (Smart's SPDZ and others).

The financial model has to use something like a pay-per-search instead of the usual ad model (although technically even better ads could be served over SMPC!).

If you're interested in studying this you're also probably interested in 1-k Oblivious Transfers (get me a static webpage I request out of a choice of k without revealing which I've requested), in Functional Encryption and also in Securely Obfuscated Programs.

Also, there was a really awesome paper at CRYPTO this year: "How to Use Bitcoin to Design Fair Protocols" (http://eprint.iacr.org/2014/129.pdf)

:)


Ex patent lawyer. A very select crew.


Patent lawyer or no, he's an incredibly gifted and accomplished cryptographer. You don't have to agree with everything in someone's past to appreciate their contribution to the field.


I think that was just a joke.

http://xkcd.com/1067/


Ah - I thought it was sarcasm. Didn't get the Einstein reference.


If anyone ever had any doubt about the coming reality of ubiquitous cryptography, here is the next move.


Oblivious Transfer, ZKPs, SMPC, Functional Encryption, Attribute-Based Encryption, (better) cryptocurrency are all things I expect to see before FHE.


Homomorphic encapsulates some of this, no? Certainly functional crypto, c.f. http://link.springer.com/chapter/10.1007%2F978-3-642-45239-0...

And there will certainly be many more advances, but if/when functional crypto gets totally sorted out it will change everything.


Yes, it's more general. The schemes that just do functional encryption are more efficient because they don't need to do everything else FHE needs to.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: