News for hackers. Stuff that matters.

davidcerezo

Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @06:14PM (#13546404)
No, you can't reverse anything. There are more details at: https://www.calctopia.com/2017/02/01/on-the-security-model/
Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @06:07PM (#13546351)
OK, I'll make much easier examples: other people which I've asked thought them OK.
Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @06:03PM (#13546308)
You're right about that: just trying to explain it to business people.

I've put more detailed information for math/crypto people here: https://www.calctopia.com/2017/02/01/on-the-security-model/

More to come soon!

Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @05:59PM (#13546265)
>Honest question: Can't I do that already by using SFTP or ssh and uploading a spreadsheet? I just feel like I'm missing something. Do you have some kind of real-time aspect going on?

No, you aren't encrypting the computation, just encrypting some data. Using secure computation (garbled circuits, secret sharing, homomorphic encryption, ...), you are encrypting every mathematical operation.

> Second question: How do I know that your closed-source app isn't exfiltrating the data clandestinely? >Assuming no flaws are ever going to be discovered in your crypto scheme in the future is pretty bold.

What I've written is: if you are carrying the secure computation on your private LAN and nobody is sniffing the traffic (potentially gigabytes), in the future they will have no way to cryptanalyze what the inputs were. And you can check that no information is being sent to anywhere during the secure computations: all the encrypted data is sent between the two computers carrying out the secure computation.

> Given that anything orbiting the earth can be observed from the ground with sufficient resolution, there's no such thing as a secret orbit. This sounds more and more like snake oil.

You can learn more here: https://eprint.iacr.org/2013/850 https://eprint.iacr.org/2016/319

Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @05:44PM (#13546161)
You can read more details about the algorithms/protocols here: https://www.calctopia.com/2017/02/01/on-the-security-model/

I don't agree with the statement "because is open-source is more secure"; for example, OpenSSL has been open-source for almost 20 years and it's difficult to consider it really secure. Only code reviews by experts make software more secure.

BTW: quoting Harry Potter doesn't make you look serious.

Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @05:35PM (#13546073)
Absolutely not: outsourcing the secure computation to third-party servers is not on the first release.

The secure multi-party computation runs between the computers of both participants (over LAN/WAN).

Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @05:33PM (#13546061)
Actually, ABY is faster and more capable than Tasty:

https://github.com/encryptogroup/ABY

Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @05:31PM (#13546047)
Sure, I've just published more details taken from the documentation:

https://www.calctopia.com/2017/02/01/on-the-security-model

Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @02:18PM (#13543962)
It's not homomorphic encryption: it's garbled circuits and oblivious transfer (order of magnitudes faster than homomorphic encryption, a cryptographic technique with lots of publicity but it'll take some years before it's practical) supporting floating-point operations (there are no opensourced libraries with floating-point support).
Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @02:00PM (#13543767)
It's running over HTTPS (protecting the network traffic between the user and their servers), but it doesn't offer encrypted computation.
Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @01:53PM (#13543695)
It's not using homomorphic encryption: it's using garbled circuits and oblivious transfer (order of magnitudes faster).
Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @01:51PM (#13543675)
Both parties are computing some spreadsheet formulas contributing their own secret inputs, in such a way that both learn the result but not the input of the other party: it's cryptographically impossible to do so. Regular collaborative spreadsheet software doesn't do that: actually, this is the first program for normal computer users that is able to do so (and not cryptographers specialized in secure multi-party computation).

It depends on your business: if you are bank, a financial firm or any other that handles very valuable and confidential data, you need secure computation on any of its forms (garbled circuits, secret sharing, homomorphic encryption).

Encrypting and emailing is encryption at rest, you're not encrypting the computation.

Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @01:31PM (#13543443)
The point is letting others use your data (vg. your valuations, prices, ...) without others learning/stealing your information, et vice versa.

It's usually explained using the Millionaires' Problem (https://en.wikipedia.org/wiki/Yao's_Millionaires'_Problem): "two millionaires, Alice and Bob, who are interested in knowing which of them is richer without revealing their actual wealth."

Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @01:20PM (#13543348)
It's not about encrypting files (encryption at rest): you can do the secure multi-party computations over your private LAN, and the encrypted network traffic will never leak to anybody (ie. no way to cryptanalyze anything in the future).

It's about encrypting the computation between the parties: this kind of technology is being used to prevent satellite collisions on secret orbits, for example.

Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @01:09PM (#13543243)
There's a succint description of the steps at the webpage:

1- Both parties have to agree to the same formulas that will be computed 2- The spreadsheets are different because the inputs of the two parties are different, not the formulas 3- When both parties agree to compute the same cell, the secure multi-party computation between both parties starts: when it's finished, the result is displayed on a destination cell or as a comment

The point is that computer users, not knowing advanced cryptography, will be able to securely compute in a transparent way (ie. "cryptography dissappears").

Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @01:02PM (#13543155)
Yes, it's state-of-the-art secure multi-party computation. The code is closed for now, but it'll be gradually open-sourced in the future.
Re:Show HN: SECCOMP – A Secure Spreadsheet on Wednesday February 01, 2017 @01:00PM (#13543141)
It's secure multiparty computation, specifically tailored for spreadsheets: your secret inputs never get out of your computer in plaintext form, only in encrypted form between the participants of the computation. For the moment, it's closed-source: but it'll be gradually open-sourced in the future, just give it some time.