Bruce Schneier: Chrome OS's Security Claims "Idiotic" • Hacker News Board

Re: Bruce Schneier: Chrome OS's Security Claims "Idiotic"

#31

Post by imgabe » Thu, Jul 09, 2009, 6:09 PM UTC

moe wrote: post #20

imgabe wrote: post #15

I think making a claim like users don't have to deal with viruses, malware, and security updates is potentially more dangerous than having an OS with a less robust security model. Even linux and BSD systems are vulnerable if malicious programs are given the necessary permissions to run. If a casual user hears something like, "This OS is immune to viruses", they're likely to be a lot less cautious about running progra…

By now most Windows users know better than to click OK when a website wants to install something on your PC. Have any data to back that up? Because my impression is the opposite.

Perhaps "most" is an overstatement, but certainly a greater proportion know better today than did 10-12 years ago. As time goes on that segment of the Windows user population is only going to increase.

Re: Bruce Schneier: Chrome OS's Security Claims "Idiotic"

#32

Post by tptacek » Thu, Jul 09, 2009, 6:45 PM UTC

alexgartrell wrote: post #30

tptacek wrote: post #23

Earlier quoted context omitted.

You'll note that I didn't comment on his post about Gentry's homomorphic encryption scheme, for two reasons: (1) homomorphic encryption is a very boring topic, and (2) I don't feel like I have an authoritative argument for Schneier not being qualified to talk about it. Having dispensed with the straw man in your second graf, I'll take the 1st and 3rd in order. I have been interviewed by reporters. And I have, in fact…

Don't you kind of think homomorphic encryption is a big deal for what it allows? I mean, at it's core, running arbitrary computation on encrypted data for the later consumption of the decrypter is a very big deal, and can be a Cloud game-changer. Schneier may not be the world's greatest guru, but he knows a lot and he writes well, which makes his opinion more relevant on average than almost anyone else's. If you want…

I start caring about crypto (and security) when it gets deployed in the real world, so I can break it. We ship a product, but for the most part, I am myself a professional abuser of software. So there you go, re: homomorphic encryption.

As regards "street cred", look, you can assign whatever credibility you want to the guy. I'm telling you, from the trenches, you are often going to be worse off for basing decisions based on what he says. Sure, you'll say, you don't base decisions off what some random pundit on the Internet says, and I say to you, "good on ya". But lots of people do, and so taking the piss out of him is a noble enterprise in my view.

And I am all about the nobility.

Re: Bruce Schneier: Chrome OS's Security Claims "Idiotic"

#33

Post by Retric » Thu, Jul 09, 2009, 7:00 PM UTC

tptacek wrote: post #29

Retric wrote: post #27

Earlier quoted context omitted.

Sorry, that's a better explanation of what you meant but it's still not a problem. You don't need to figure out what a program can do ahead of time if you limit what it can do at run time. You don't even need to let the user do anything at run time.

You are arguing with the wall. I'm just explaining what Schneier meant by "2+2=3".

Lol, ok sorry I just had a someone is wrong on the internet moment. Which had nothing to you just the idea about the limits on what an OS could do. http://xkcd.com/386/

PS: I also just noticed that that was commic #386.

Re: Bruce Schneier: Chrome OS's Security Claims "Idiotic"

#34

Post by ori_b » Fri, Jul 10, 2009, 12:47 PM UTC

Retric wrote: post #28

ori_b wrote: post #17

Earlier quoted context omitted.

Allowing friendly programs to access resources but disallowing unfriendly ones is impossible, because it is impossible to distinguish between the two without some form of enforced signing that is impossible for the user to subvert(and who wants that?). And no, defaulting to reduced permissions for some programs is insufficient. People will still click "Ok, allow higher permissions" on programs if prompted. What's the…

Not much, but nothing says the user needs to be able to type 'rm -rf foo'. Most people are used to OS that are designed to be fast and permissive, but you can code a defensive OS.

What sort of defensiveness do you think will work to effectively prevent -- to make up an example -- an application from popping up a fake login screen that asks them to enter the details, then mails the login info to the attacker and the program to all the victim's friends, without making the system unusable? Signed binaries where the signature describes the level of access might work if you don't allow it to be subverted by the user, but at the cost of essentially losing control over what you can run.i

I can't think of an especially usable uber-paranoid defensive system, although I'd definitely love to be proven wrong. Essentially, the problem isn't a totally technological one. The core of the problem is that the user can effectively be convinced to be their own attacker.