text stringlengths 0 1.36k |
|---|
**Rachel White:** I think it's dumb, but I like making Mikeal say it. \[laughter\] |
**Alex Sexton:** That's fair... That's very fair. |
**Mikeal Rogers:** That's really fair. Alright, I'm Mikeal Rogers... |
**Alex Sexton:** I'm Alex Sexton... |
**Rachel White:** And I'm Rachel White. |
**Mikeal Rogers:** Alright, everybody! Let's get this party started, let's just dive right into the first topic. So Google broke the internet. I don't know why they keep pointing out flaws in the internet's security, but \[whispering\] they broke the internet again! |
**Alex Sexton:** I thought they helped to disclose it, but wasn't it like some German W, some acronym...? I think it was the Germans, it's all I'm saying. |
**Mikeal Rogers:** It was the Germans... Likely story. Anyway, so SHA1 hashing algorithm has been cracked. I guess in 2005 there was a paper written that said theoretically it could be cracked, but nobody had done it yet. Apparently, as of like 2010, the federal government said no government encryption can use any SHA1... |
**Alex Sexton:** Yeah... The only person I've seen strongly support SHA1 for the last six or seven years is Linus Torvalds in Git. |
**Mikeal Rogers:** It's so annoying... |
**Alex Sexton:** And not just kind of... He really was like, "You guys are all super dumb for caring about this." |
**Mikeal Rogers:** I know, it's really crazy. He's still downplaying it, actually. So backing up a little bit - let's just get into what is SHA1 and what does it do. Does anybody else wanna take a crack at this, or do you want me to explain it? |
**Rachel White:** I only know it from Git related stuff, so that's all... |
**Mikeal Rogers:** Right, that's actually a really good way to explain it though. The way that Git uses SHA1 is kind of indicative of how everybody uses it, which is that you take a bunch of data and you say, "I want a unique identifier for this data", so you hash it. That's what Git does to every change that comes int... |
If you go to GitHub and you go to a project and then you click on Commit To and then you click on one of those Commit links, in the URL bar you'll see this randomly-generated identifier, and that is a unique identifier for that hash. The problem is that if you could forge these - that's a very small amount of data, rep... |
**Alex Sexton:** It still costs like a hundred thousand dollars. It will be cheaper, but right now, with the current algorithm... It's insane how much faster they can do it, but still, with AWS spot instances it costs around a hundred thousand dollars to break a random thing. |
**Mikeal Rogers:** But how much do I have to pay Russian hackers that have a botnet? Like one Bitcoin, which is roughly twenty thousand dollars? |
**Alex Sexton:** I wouldn't necessarily be worried about this... Sorry, I didn't hear that. |
**Mikeal Rogers:** I was saying I'd probably need to pay one Bitcoin, which is roughly twenty thousand dollars, to get Russian hackers to break it. |
**Alex Sexton:** \[04:04\] Oh yeah, for sure. The cost is still prohibitive to the point where no one's gonna troll you with this. Someone really needs to want -- there has to be a reason someone's doing this at this point. But that will only be true for like two months, or something. People will make this better insta... |
**Mikeal Rogers:** Right, and it's pretty much a given now that governments can do this at will. What that means is that if your integrity checks involve you hashing with this algorithm, now if you're just using those checks, people can just inject malware whenever they want. |
**Rachel White:** I have a question. If this has been relatively not super secure for a while, what was the catalyst for people to be like, "Okay, it's finally time to stop using this thing"? Was it something that Google did that you said...? |
**Alex Sexton:** Oh yeah, yesterday. |
**Mikeal Rogers:** Yeah. Well, honestly, I think most people in the security community have felt since 2005 that you should stop using this. There are other algorithms that are just as good that don't have this problem. And in 2010, most reasonable companies said, "Hey, we should stop using this." |
**Alex Sexton:** Browsers already don't allow -- you'll get a very big red X instead of a green lock if SHA1 is used for web security stuff. It's been well known to be very crackable by someone with a ton of money for a long time. |
**Mikeal Rogers:** Yeah, but like Alex said, Linus Torvalds has just remained unimpressed by evidence, so it is still in heavy use in Git, in GitHub, and a bunch of other Linux-related stuff. |
**Rachel White:** That's fine, because I'm wholly unimpressed by him, so it's okay. \[laughter\] |
**Alex Sexton:** You're only gonna make him stronger. |
**Mikeal Rogers:** To answer your question though, the thing that happened yesterday was that some people from Google and 'zee Germans' came out and just said, "Hey, look, we cracked it. Here's exactly how we cracked it." So it went from theoretical to "Here is an open version of this." |
**Alex Sexton:** To be totally clear though, they have to try a ton of things... They were able to reduce the subset that you had to brute force to a small enough amount to be significant. But it still takes a hundred and ten years of computing time, or something like that. You had to put a lot of machines into it. But... |
**Mikeal Rogers:** Yeah. So if you're future-proofing, don't use SHA1. |
**Alex Sexton:** Or past-proofing. If you're just proofing at all, don't use SHA1. There's SHA256, which is essentially exactly the same, with much higher entropy, so just use that instead. |
**Mikeal Rogers:** I've actually become a big fan of multihash. Have you ever heard about this? |
**Alex Sexton:** No. |
**Mikeal Rogers:** Okay, nobody has. Juan Benet has been pushing this really hard for quite a while. He's one of the people behind IPFS, so lots of kind of distributed, peer-to-peer crypto stuff. He has really wanted to future-proof everything that he's been working on, so he started this little open source project cal... |
**Alex Sexton:** \[08:07\] It's very similar to .mkv or .mov - all the container things for video codecs, I suppose. |
**Mikeal Rogers:** Right. Although containers, oddly, do implement a bunch of features. |
**Alex Sexton:** True. |
**Mikeal Rogers:** This gets really ugly, actually, in codecs and containers. For multihash, for instance, there's libraries in pretty much every language ever, including a very well maintained JavaScript implementation that works in the browser and in Node. So that's what I've used in a couple of projects recently. Bu... |
**Alex Sexton:** I mean, it does make it a lot harder, for what it's worth. |
**Mikeal Rogers:** It does. It does make the attack a lot harder, but I do feel that rather than future-proofing or moving to just a better algorithm... He's just kind of dangling out this, like "Oh, prove me wrong, computer scientists!" \[laughs\] Which didn't work out that well for his last round of this... |
**Alex Sexton:** Right... It seems silly to be like, "Well, you only half-broke it, so I'm gonna continue..." If you got through a half of my lock, I'm gonna go ahead and just change out the whole thing. |
**Mikeal Rogers:** Yeah, rather than just like, "No, I'm gonna continue to put these half-locks on things." |
**Alex Sexton:** Right. |
**Mikeal Rogers:** This kind of reminds me that the way we think about security on the web tends to be like, "Oh, I put CloudFlare in front of it, so I'm secure now." "I added SSL, or I added TLS, so I'm secure now." But really, security is this really multi-layered thing where when you break off one layer of the onion... |
**Alex Sexton:** Yeah. I think you almost can't even break off any of the layers of the onion. Security is really, really hard. It needs to be there at every layer, otherwise the other ones have no effect. I think an onion is a poor metaphor. A chain is much better - if you have a single weak link, then it doesn't matt... |
**Mikeal Rogers:** Yeah. I mean, if you look at some of the stuff people have been doing with OAuth for a while... OAuth jumps through all these hoops to basically do an extra layer of encryption. Initially, they kind of did that so that you could do OAuth over HTTP without TLS. But even when you added TLS to it, it's ... |
**Alex Sexton:** Sure. If you operate under the assumption that TLS is broken though, then the entire internet is broken already... The OAuth channel - if you had that extra encryption - would be broken, but then as soon as you got to that website and used it, you'd be screwed anyways if TLS is broken. So I don't think... |
Maybe someone doesn't get your authentication credentials, but hopefully you don't reuse those anywhere else, so... |
**Mikeal Rogers:** Alright. Do we have anything else to say about hashing algorithms? This is a pretty deep topic to start a JavaScript show with... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.