Hacker News Reader

Comment by ColanR | Hacker News Reader

This makes me wonder how many presently existing encryption & security projects / organizations are owned or influenced by government agencies.

View on HN

… yes, and how many (open-source) software projects in general.

Though I know it's not perfect, free software projects at least give you some ability to discover and hinder government interference. I don't know why you seem more interested/worried about them.

Wasn't there a post here about how little review open source software actually gets due to the highly specific knowledge required to identify vulnerabilities?

something like: https://www.itworld.com/article/2922074/significant-virtual-...

Seems like creating a foundational open source project and introducing a few key bugs would be the best way to hide in the open.

Because the open source community is slowly coming to the uncomfortable realization that many eyes != security. Take heartbleed: SSL had a glaring security vulnerability open for years that none of those eyes ever spotted.

I've seen questions raised here on HN about Signal and Tor: where they came from, and where their funding comes from. If I had to bet, then I'd bet both of those are modern day Crypto AG variants.

Its one thing to acknowledge that open source software doesn't get the review it needs. Its another thing entirely to suggest that major platforms in use today are sponsored by state actors willing and able to introduce vulnerabilities without proof.

Turnkey black box solutions may be reviewed more regularly by a dedicated team but you have to admit that they're subject to flimsy and easy manipulation by state actors and the greed and coruptability of their owners.

> Its another thing entirely to suggest that major platforms in use today are sponsored by state actors willing and able to introduce vulnerabilities without proof.

I think the Crypto AG story is sufficient proof of itself to look with suspicion at all related open source projects. In situations where there are known bad actors and we are dependent on security, we should look with suspicion unless we know better. "Insecure until proven secure" is probably a good motto.

>Insecure until proven secure" is probably a good motto.

So just always insecure, as no amount of testing can guarantee there isn't some heartbleed like bug in there still.

If that's the reality, should we whitewash it?

>Because the open source community is slowly coming to the uncomfortable realization that many eyes != security

The community has been aware of this for a very long time. Many eyes still provides better security than few eyes, which is why singling out free software as your concern seems misplaced.

>f I had to bet, then I'd bet both of those are modern day Crypto AG variants.

Though it's possible Signal is compromised, it's basically known that the proprietary offshoot WhatsApp is compromised. Free software is still your best bet, likely even better than doing it yourself for 99.9% of the world

The evidence seems to show that open source crypto is more likely to be secure. The Dual_EC_DRBG [1] thing is a good example where the NSA was desperate enough to actually try to hack an algorithm and then was immediately caught at it. That known dodgy algorithm was then used in commercial crypto products for 7 years. The open source crypto community could just avoid it.

Snowden used an open source implementation of PGP based on his knowledge of the NSA's capabilities at the time. He did not go shopping at the crypto store.

[1] https://en.wikipedia.org/wiki/Dual_EC_DRBG

Pretty much all of them, I'd assume. Basically there is no secret that a major state player cannot crack open unless 1) it doesn't care 2) it's protected by another equally strong state player.

Or 3) you built your own system from scratch.

By from scratch, I mean building a CPU from transistors and writing (or at least carefully auditing) all the code on it.

People have done comparable projects (like a clone of the Apollo 11 guidance computer) as a labor of love.

I wonder if RISC-V would fall under this category. I saw recently that a decently powerful SOC was built and sold in small quantities. Is that architecture sufficiently open (and simple) that a satisfactory audit would be feasible?

The issue with #3 in practice is that roll your own tends to have its own undiscovered bugs and winds up being security by obscurity in practice unfortunately. There is a reason best practices for Cryptography involve throughly tested algorithims as opposed to just "the latest and greatest".

If you can audit completely enough to not leave anything then the question becomes why not audit the commodity? Lack of availability of auditable components or proportionate costs of doing so is the main answer I can think of.

Reply on news.ycombinator.com