WEBVTT - Automatic transcript; recognition may contain errors. Reviewed for technical wording and speaker labels. Times match the published episode.

00:00:20.460 --> 00:00:23.140
<v Aaron van Wirdum>Live from Utrecht, this is Bitcoin.

00:00:23.980 --> 00:00:24.500
<v Sjors Provoost>Explained.

00:00:24.860 --> 00:00:25.300
<v Aaron van Wirdum>Hey, Sjors.

00:00:25.600 --> 00:00:25.980
<v Sjors Provoost>What's up?

00:00:26.380 --> 00:00:30.480
<v Aaron van Wirdum>Did you see the amazing announcement that Bitcoin Magazine made this week?

00:00:32.250 --> 00:00:34.270
<v Aaron van Wirdum>The conference is coming to Amsterdam.

00:00:34.530 --> 00:00:35.010
<v Sjors Provoost>I heard.

00:00:35.170 --> 00:00:35.930
<v Aaron van Wirdum>Around the corner.

00:00:36.350 --> 00:00:36.870
<v Sjors Provoost>Exciting.

00:00:37.090 --> 00:00:38.790
<v Aaron van Wirdum>It's about half an hour from Utrecht.

00:00:39.390 --> 00:00:41.510
<v Sjors Provoost>Yeah, no, it's pretty close.

00:00:41.770 --> 00:00:42.790
<v Aaron van Wirdum>Can we expect you there?

00:00:43.110 --> 00:00:43.590
<v Sjors Provoost>Maybe.

00:00:44.340 --> 00:00:48.760
<v Aaron van Wirdum>We'll see. I'll be there. I'll be there for sure. Sjors, we're going to talk

00:00:48.760 --> 00:00:54.710
<v Aaron van Wirdum>about... We're going to talk about address reuse. So this is not really a

00:00:54.710 --> 00:01:00.410
<v Aaron van Wirdum>new topic. It's not something we haven't discussed at least sort of

00:01:00.410 --> 00:01:05.610
<v Aaron van Wirdum>tangentially before. For example, two episodes ago,

00:01:05.990 --> 00:01:11.370
<v Aaron van Wirdum>Ruben was here to talk about his proposal for, what was it called again? Stealth

00:01:11.370 --> 00:01:14.450
<v Aaron van Wirdum>addresses, privacy addresses, dark addresses.

00:01:14.750 --> 00:01:15.410
<v Sjors Provoost>Silent payments.

00:01:15.490 --> 00:01:19.770
<v Aaron van Wirdum>Silent payments. There you go. Thank you. Um, so yeah but we never really tackled

00:01:19.770 --> 00:01:24.410
<v Aaron van Wirdum>the concept of address reuse in itself and there is sort of a

00:01:25.210 --> 00:01:30.810
<v Aaron van Wirdum>newsworthy reason for us to do this this week but we'll get to that in a bit, right?

00:01:30.810 --> 00:01:31.490
<v Sjors Provoost>That's right.

00:01:31.490 --> 00:01:37.380
<v Aaron van Wirdum>So before we start address reuse it's something that's discouraged in

00:01:37.380 --> 00:01:43.170
<v Aaron van Wirdum>Bitcoin. Sjors, what does it actually mean when you're

00:01:43.170 --> 00:01:48.090
<v Aaron van Wirdum>reusing addresses? And I'm specifically asking about what does this mean sort of

00:01:48.090 --> 00:01:51.930
<v Aaron van Wirdum>under the hood? Because we've discussed addresses before and addresses aren't

00:01:51.930 --> 00:01:57.330
<v Aaron van Wirdum>actually a thing in the Bitcoin blockchain, right? So what does it mean if you're

00:01:57.330 --> 00:01:58.270
<v Aaron van Wirdum>reusing addresses?

00:01:58.490 --> 00:02:02.010
<v Sjors Provoost>Right, because an address is something you give to a wallet. It's not something

00:02:02.010 --> 00:02:07.180
<v Sjors Provoost>that actually exists on the blockchain. An address is an instruction essentially

00:02:07.180 --> 00:02:12.980
<v Sjors Provoost>to the wallet of the payer that says this is how you need to send me coins. And

00:02:12.980 --> 00:02:16.940
<v Sjors Provoost>usually what the address will contain is a public key or a hash of a public key.

00:02:18.120 --> 00:02:23.000
<v Sjors Provoost>And then as a sender, you will put that on the blockchain so the money actually

00:02:23.000 --> 00:02:28.330
<v Sjors Provoost>goes to either a public key or the hash of the public key. And so what does that

00:02:28.330 --> 00:02:32.370
<v Sjors Provoost>mean for address reuse? It just basically means that you're making multiple

00:02:32.370 --> 00:02:38.350
<v Sjors Provoost>transactions where you're sending coins to the same public key or the same

00:02:38.350 --> 00:02:39.590
<v Sjors Provoost>hash of the public key.

00:02:40.070 --> 00:02:45.410
<v Aaron van Wirdum>And is this immediately clear when you are sending the money? Is it clear that

00:02:45.410 --> 00:02:49.690
<v Aaron van Wirdum>it's being sent to the same quote unquote address? Or is it only clear when

00:02:49.690 --> 00:02:51.830
<v Aaron van Wirdum>you're sending money from that address again?

00:02:52.800 --> 00:02:56.400
<v Sjors Provoost>No, it's immediately clear. So you can see on the blockchain for each transaction

00:02:56.400 --> 00:03:01.380
<v Sjors Provoost>what the destination is of that transaction. So you can see that it was sent.

00:03:02.060 --> 00:03:06.000
<v Sjors Provoost>Yeah, you can see that the destination has been used before if you keep an eye on

00:03:06.000 --> 00:03:06.640
<v Sjors Provoost>what's on the blockchain.

00:03:06.960 --> 00:03:11.960
<v Aaron van Wirdum>Right, exactly. So this is, you sometimes see, a very common example of this

00:03:11.960 --> 00:03:16.600
<v Aaron van Wirdum>would be like donation addresses, where someone just posts an address on a

00:03:16.600 --> 00:03:21.260
<v Aaron van Wirdum>website for donations. And then because of that, you can track exactly how much

00:03:21.260 --> 00:03:26.360
<v Aaron van Wirdum>donations someone is getting, right? That's sort of the classic example. But

00:03:26.360 --> 00:03:30.440
<v Aaron van Wirdum>there are more examples. There are, or at least a couple of years ago, there were

00:03:30.440 --> 00:03:34.900
<v Aaron van Wirdum>bad wallets, or at least I would consider them bad wallets that wouldn't generate

00:03:34.900 --> 00:03:38.660
<v Aaron van Wirdum>new addresses automatically. I think these days that's mostly resolved.

00:03:39.020 --> 00:03:42.860
<v Sjors Provoost>Yeah, I mean, there was a trade-off in the beginning, because before we had this

00:03:42.860 --> 00:03:47.580
<v Sjors Provoost>thing called hierarchical deterministic wallets. So right now you write down

00:03:47.580 --> 00:03:51.180
<v Sjors Provoost>these 12 words, and these 12 words can generate an infinite number of addresses,

00:03:51.400 --> 00:03:56.280
<v Sjors Provoost>or not actually infinite, but quite a lot. But before that, it was the case that

00:03:56.280 --> 00:04:00.980
<v Sjors Provoost>every address was generated using a unique, randomly generated private key. And

00:04:00.980 --> 00:04:05.040
<v Sjors Provoost>so in the beginning, you might create a wallet and it might start with 100

00:04:05.040 --> 00:04:08.360
<v Sjors Provoost>addresses in it, which means 100 different private keys and you would have to

00:04:08.360 --> 00:04:12.020
<v Sjors Provoost>back those up. So that's already a challenge because you can't write that down.

00:04:12.160 --> 00:04:16.100
<v Sjors Provoost>It's too much information. So you'd maybe put it on a USB stick. But then if

00:04:16.100 --> 00:04:21.240
<v Sjors Provoost>you've used up all the 100 addresses and you create address number 101 and you

00:04:21.240 --> 00:04:24.500
<v Sjors Provoost>forget to make a backup of that, you just don't have a backup of any coins that

00:04:24.500 --> 00:04:29.790
<v Sjors Provoost>go to the last address. So there was a trade-off between backups and privacy in

00:04:29.790 --> 00:04:34.490
<v Sjors Provoost>the beginning. And that's one of the reasons probably why not all wallets created

00:04:34.490 --> 00:04:35.130
<v Sjors Provoost>fresh addresses.

00:04:36.310 --> 00:04:41.530
<v Aaron van Wirdum>Right. And now that's resolved. Now wallets just use deterministic addresses.

00:04:42.010 --> 00:04:42.390
<v Sjors Provoost>Yep.

00:04:42.670 --> 00:04:46.190
<v Aaron van Wirdum>And you only need one backup. Okay, so...

00:04:46.190 --> 00:04:50.010
<v Sjors Provoost>So there's still plenty of places, especially exchanges, have a habit of giving

00:04:50.010 --> 00:04:53.700
<v Sjors Provoost>you an address where you get the same address every time.

00:04:53.960 --> 00:04:58.440
<v Aaron van Wirdum>Yeah. I mean, there's, you know, once in a while still run into like a web shop

00:04:58.440 --> 00:05:04.180
<v Aaron van Wirdum>that will not generate a new address or there are more examples.

00:05:04.540 --> 00:05:04.660
<v Sjors Provoost>Yeah.

00:05:05.300 --> 00:05:09.300
<v Aaron van Wirdum>Okay. So then the next question, I think there are,

00:05:11.130 --> 00:05:16.590
<v Aaron van Wirdum>why is this a problem? And I'd say we can easily categorize this in sort of two

00:05:16.590 --> 00:05:17.750
<v Aaron van Wirdum>or three categories.

00:05:19.160 --> 00:05:25.080
<v Aaron van Wirdum>One category is privacy and then arguably another category is censorship

00:05:25.080 --> 00:05:25.620
<v Aaron van Wirdum>resistance.

00:05:26.910 --> 00:05:32.130
<v Aaron van Wirdum>There's a good Bitcoin wiki page about address reuse, which sort of lists these

00:05:32.130 --> 00:05:36.790
<v Aaron van Wirdum>problems and examples, but we'll go over it in this podcast as well. I'm not

00:05:36.790 --> 00:05:40.830
<v Aaron van Wirdum>going to read out loud the whole URL, but if you just Google address reuse

00:05:40.830 --> 00:05:46.590
<v Aaron van Wirdum>Bitcoin wiki, you'll find the page. So yeah, privacy censorship resistance. And

00:05:46.590 --> 00:05:51.830
<v Aaron van Wirdum>this wiki gives three pretty good examples of why privacy is a problem. So let's

00:05:51.830 --> 00:05:57.740
<v Aaron van Wirdum>run through this real fast. So one problem if you're reusing addresses and how

00:05:57.740 --> 00:06:02.700
<v Aaron van Wirdum>it harms your privacy is that it reveals your savings. So, you know, if someone

00:06:02.700 --> 00:06:07.660
<v Aaron van Wirdum>pays you, then that person can, for example, easily see how much money is on that

00:06:07.660 --> 00:06:13.460
<v Aaron van Wirdum>address. And, you know, they might start talking to

00:06:13.460 --> 00:06:17.860
<v Aaron van Wirdum>people about this. And before you know it, your kids are kidnapped and they know

00:06:17.860 --> 00:06:23.510
<v Aaron van Wirdum>exactly how much ransom to demand. Right. That's one example. Another

00:06:23.510 --> 00:06:28.520
<v Aaron van Wirdum>example is exchange front running. It's another example that's on the wiki. So

00:06:28.520 --> 00:06:34.360
<v Aaron van Wirdum>you're not generating a new address on your exchange and that address is known to

00:06:34.360 --> 00:06:39.160
<v Aaron van Wirdum>belong to that exchange for blockchain analysis. And then one day you're sending

00:06:39.160 --> 00:06:43.420
<v Aaron van Wirdum>a lot of money to that exchange. So people can see that and people figure you're

00:06:43.420 --> 00:06:48.140
<v Aaron van Wirdum>going to sell a lot of Bitcoin and people might want to front run your sale by...

00:06:48.880 --> 00:06:53.140
<v Aaron van Wirdum>selling their first their Bitcoin first and buying it back cheaper by the time

00:06:53.140 --> 00:06:55.300
<v Aaron van Wirdum>your money finally arrives at the exchange.

00:06:55.300 --> 00:06:55.620
<v Sjors Provoost>Yeah.

00:06:55.620 --> 00:07:01.180
<v Aaron van Wirdum>Oh, yeah. The first example—the third example is so let's say you're being paid

00:07:01.180 --> 00:07:06.540
<v Aaron van Wirdum>your salary on your known address and then one day you get a pay rise.

00:07:07.700 --> 00:07:12.300
<v Aaron van Wirdum>Your landlord might be able to see that and decide it's time to charge a little

00:07:12.300 --> 00:07:15.600
<v Aaron van Wirdum>bit more rent since you can clearly afford that.

00:07:16.000 --> 00:07:18.840
<v Sjors Provoost>Right. So basically it's similar to the first example. And the first one was

00:07:18.840 --> 00:07:21.780
<v Sjors Provoost>mostly about seeing how much savings somebody has. And the other one is how much

00:07:21.780 --> 00:07:24.160
<v Sjors Provoost>income somebody has. It's the same kind of problem.

00:07:24.580 --> 00:07:29.100
<v Aaron van Wirdum>Yeah. So these are just some examples of why it's bad if you don't have any

00:07:29.100 --> 00:07:33.400
<v Aaron van Wirdum>financial privacy. A fourth example. So the censorship resistant example is…

00:07:33.900 --> 00:07:37.280
<v Aaron van Wirdum>Let's say you have...

00:07:37.280 --> 00:07:41.600
<v Sjors Provoost>Maybe we can stick to the first three first. So I think one comment there is that

00:07:41.600 --> 00:07:45.540
<v Sjors Provoost>reusing addresses helps with these issues. It does not completely remove them.

00:07:45.950 --> 00:07:49.930
<v Sjors Provoost>Because, for example, with the savings, you might use a new address every time,

00:07:49.930 --> 00:07:54.750
<v Sjors Provoost>but if all your savings are on the first address, maybe people can still link it

00:07:54.750 --> 00:07:59.910
<v Sjors Provoost>to you. So it's a mitigation. It helps to use multiple addresses, but it

00:07:59.910 --> 00:08:02.690
<v Sjors Provoost>doesn't completely perfectly solve these privacy issues.

00:08:03.050 --> 00:08:08.330
<v Aaron van Wirdum>Yeah, that's definitely true. I would say address reuse is like a bare minimum.

00:08:08.850 --> 00:08:13.360
<v Aaron van Wirdum>That's the very least you should do to protect your privacy. And then there's a

00:08:13.360 --> 00:08:18.080
<v Aaron van Wirdum>lot more you can do after that. But if you're not even reusing, if you're not

00:08:18.080 --> 00:08:22.280
<v Aaron van Wirdum>even, if you're reusing addresses, then

00:08:22.280 --> 00:08:27.730
<v Aaron van Wirdum>there's literally no privacy whatsoever. That's

00:08:27.730 --> 00:08:32.730
<v Aaron van Wirdum>the least you can do. Okay, then there's the censorship resistant

00:08:32.730 --> 00:08:38.710
<v Aaron van Wirdum>argument, which is if you're, let's say you have some kind of

00:08:38.710 --> 00:08:44.350
<v Aaron van Wirdum>political cause and you're receiving donations on an address, then someone

00:08:44.350 --> 00:08:48.570
<v Aaron van Wirdum>might want to send you some Bitcoin from an exchange, but that exchange doesn't

00:08:48.570 --> 00:08:52.790
<v Aaron van Wirdum>like your political cause, so they refuse to send the payments. That's another

00:08:52.790 --> 00:08:55.930
<v Aaron van Wirdum>example that's easily resolved to generate new addresses.

00:08:56.210 --> 00:08:59.770
<v Sjors Provoost>This goes to address blacklisting in any kind of form, right? Somebody might be

00:08:59.770 --> 00:09:05.110
<v Sjors Provoost>ordering miners to block specific addresses. And yeah, reusing addresses makes

00:09:05.110 --> 00:09:08.230
<v Sjors Provoost>that whole cat and mouse game significantly more difficult.

00:09:08.650 --> 00:09:13.230
<v Aaron van Wirdum>Okay, so that's privacy and censorship resistance. And then I think the third

00:09:13.230 --> 00:09:18.690
<v Aaron van Wirdum>bucket of problems is maybe a little bit more interesting. And that's sort of

00:09:18.690 --> 00:09:22.550
<v Aaron van Wirdum>where you were gearing this episode towards. That's security.

00:09:22.990 --> 00:09:23.270
<v Sjors Provoost>Yep.

00:09:23.670 --> 00:09:24.910
<v Aaron van Wirdum>So sure,

00:09:27.370 --> 00:09:31.930
<v Aaron van Wirdum>reusing addresses is kind of bad for security as well, right?

00:09:32.820 --> 00:09:38.460
<v Sjors Provoost>Well, reusing addresses, there are certain security vulnerabilities or potential

00:09:38.460 --> 00:09:42.140
<v Sjors Provoost>vulnerabilities that will get a lot worse when you're using addresses.

00:09:42.540 --> 00:09:43.020
<v Aaron van Wirdum>Right.

00:09:43.300 --> 00:09:48.600
<v Sjors Provoost>So it's not that the reusing of addresses is necessarily super insecure. But

00:09:48.600 --> 00:09:51.560
<v Sjors Provoost>yeah, there are definitely some attacks that you can think of and that have

00:09:51.560 --> 00:09:54.880
<v Sjors Provoost>actually happened that would not happen if you happen to change your addresses.

00:09:55.320 --> 00:09:58.840
<v Aaron van Wirdum>Right. So what are these? What are the attacks that are possible if you're

00:09:58.840 --> 00:09:59.820
<v Aaron van Wirdum>reusing addresses?

00:10:00.140 --> 00:10:03.840
<v Sjors Provoost>Yeah. So the one that's mentioned in the wiki is something called the same K

00:10:03.840 --> 00:10:09.140
<v Sjors Provoost>value in multiple signatures. But basically what that means is that if you have a

00:10:09.140 --> 00:10:14.390
<v Sjors Provoost>private key... And you create a signature with it. And that signature contains

00:10:14.390 --> 00:10:19.230
<v Sjors Provoost>another piece of data, which should be a randomly generated number.

00:10:21.730 --> 00:10:27.650
<v Sjors Provoost>And as long as that randomly generated number is really random, then you can

00:10:27.650 --> 00:10:31.570
<v Sjors Provoost>use the same private key twice and you can sign different messages. Because every

00:10:31.570 --> 00:10:33.010
<v Sjors Provoost>transaction is essentially a message.

00:10:33.490 --> 00:10:34.050
<v Aaron van Wirdum>Right.

00:10:34.310 --> 00:10:38.640
<v Sjors Provoost>So when you're spending money from an address once... You're signing a message

00:10:38.640 --> 00:10:41.860
<v Sjors Provoost>using that private key if you're spending money again from that same address

00:10:41.860 --> 00:10:45.500
<v Sjors Provoost>because there's another coin, perhaps, that was sent to the same address. You're

00:10:45.500 --> 00:10:49.120
<v Sjors Provoost>signing another message with the same private key. And as long as you're doing

00:10:49.120 --> 00:10:52.880
<v Sjors Provoost>the random number generation correctly, that's all fine. There's no problem. But

00:10:52.880 --> 00:10:56.960
<v Sjors Provoost>if there's something wrong in that mechanism and you are not generating a random

00:10:56.960 --> 00:11:02.200
<v Sjors Provoost>number, but, for example, you are using the number 0000, if you do that twice,

00:11:02.560 --> 00:11:06.960
<v Sjors Provoost>now everybody can steal your money, because they can know your private key.

00:11:07.100 --> 00:11:08.420
<v Sjors Provoost>That's just a cryptographic property.

00:11:10.480 --> 00:11:13.960
<v Sjors Provoost>And so the main protection against that is to make sure your wallet doesn't make

00:11:13.960 --> 00:11:18.860
<v Sjors Provoost>that mistake, by far. But if your wallet made that mistake, but it was not

00:11:18.860 --> 00:11:20.380
<v Sjors Provoost>reusing addresses, then you're still okay.

00:11:24.890 --> 00:11:28.810
<v Aaron van Wirdum>Is that really right? Is what you said really right? If you don't use a random

00:11:28.810 --> 00:11:33.450
<v Aaron van Wirdum>address, people can just pick your signature from... No, they can pick your

00:11:33.450 --> 00:11:36.310
<v Aaron van Wirdum>private key from your signature? That doesn't sound right to me.

00:11:36.390 --> 00:11:42.230
<v Sjors Provoost>If you sign two different messages, so two different transactions using the

00:11:42.230 --> 00:11:45.950
<v Sjors Provoost>same private key, and you're doing something really, really, really wrong during

00:11:45.950 --> 00:11:48.750
<v Sjors Provoost>the signing, namely not generating the right random number,

00:11:50.110 --> 00:11:51.330
<v Sjors Provoost>yeah, then people can steal your money.

00:11:53.440 --> 00:11:59.160
<v Sjors Provoost>And that's actually the mechanism that's used by one episode that we talked

00:11:59.160 --> 00:12:03.740
<v Sjors Provoost>about, the Discreet Log contracts. One of the mechanisms to do the Discreet Log

00:12:03.740 --> 00:12:08.840
<v Sjors Provoost>contract is that somebody, like the Oracle says, I'm either going to sign yes or

00:12:08.840 --> 00:12:14.770
<v Sjors Provoost>I'm going to sign no. And if they sign both, you can steal the money. And so

00:12:14.770 --> 00:12:17.830
<v Sjors Provoost>in that case, it's by design that you can steal the money by that mistake.

00:12:17.950 --> 00:12:18.450
<v Aaron van Wirdum>Oh, yeah.

00:12:18.590 --> 00:12:21.650
<v Sjors Provoost>But wallets can make that mistake accidentally, and they have in the past. Now,

00:12:21.690 --> 00:12:25.690
<v Sjors Provoost>there are ways to prevent that that don't even require random data. It's called

00:12:25.690 --> 00:12:31.050
<v Sjors Provoost>the pseudo random data. So you generate a random number, but you use the

00:12:31.050 --> 00:12:34.470
<v Sjors Provoost>transaction itself to generate the random number. And because no two transactions

00:12:34.470 --> 00:12:37.330
<v Sjors Provoost>are identical, it's always going to be a different number.

00:12:38.100 --> 00:12:43.660
<v Aaron van Wirdum>So this is a different problem because one of the problems that I've heard

00:12:43.660 --> 00:12:48.300
<v Aaron van Wirdum>about address reuse and being able to steal...

00:12:48.500 --> 00:12:52.660
<v Sjors Provoost>But I would say it's not really about address reuse, right? It helps. But in this

00:12:52.660 --> 00:12:57.180
<v Sjors Provoost>case, this particular example that I just mentioned could also happen with RBF if

00:12:57.180 --> 00:13:00.700
<v Sjors Provoost>your wallet made that mistake. Because RBF, you would make one transaction with

00:13:00.700 --> 00:13:03.940
<v Sjors Provoost>one satoshi per byte using the private key. And then you'd make another

00:13:03.940 --> 00:13:08.080
<v Sjors Provoost>transaction that bumps the fee to two satoshi per byte. You'd be using the same

00:13:08.080 --> 00:13:11.980
<v Sjors Provoost>private key twice. And if your wallet made that particular mistake, you would

00:13:11.980 --> 00:13:13.600
<v Sjors Provoost>lose your money during the fee bump.

00:13:13.980 --> 00:13:19.440
<v Aaron van Wirdum>So what I was getting at, I thought it was only possible if there were quantum

00:13:19.440 --> 00:13:20.020
<v Aaron van Wirdum>computers.

00:13:20.240 --> 00:13:20.380
<v Sjors Provoost>Nope.

00:13:20.680 --> 00:13:26.480
<v Aaron van Wirdum>So if there were quantum computers tomorrow, then they would be able to extract

00:13:26.480 --> 00:13:31.980
<v Aaron van Wirdum>private keys from signatures. But you're saying this is already possible if

00:13:31.980 --> 00:13:35.240
<v Aaron van Wirdum>there's no random data included in the signature?

00:13:35.500 --> 00:13:41.070
<v Sjors Provoost>Yeah, so the signature, indeed, consists of, yeah. Yeah, that mistake was already

00:13:41.070 --> 00:13:44.130
<v Sjors Provoost>there. It's been exploited in the wild. There were wallets that made that mistake

00:13:44.130 --> 00:13:48.630
<v Sjors Provoost>and coins were stolen. And there's probably still people out there scanning the

00:13:48.630 --> 00:13:53.070
<v Sjors Provoost>blockchain for people making that mistake and then very quickly stealing the money.

00:13:54.730 --> 00:13:57.290
<v Sjors Provoost>It's very different from quantum attacks. We can talk about those though.

00:13:58.300 --> 00:14:02.680
<v Aaron van Wirdum>I mean, I brought it up, so yeah. What about them? Is what I just described

00:14:02.680 --> 00:14:03.200
<v Aaron van Wirdum>correct?

00:14:03.760 --> 00:14:08.560
<v Sjors Provoost>It's not so much from signatures. So what happens with quantum, at least the idea

00:14:08.560 --> 00:14:13.710
<v Sjors Provoost>of quantum computers in theory one day, is that if there's a public key on the

00:14:13.710 --> 00:14:19.330
<v Sjors Provoost>blockchain, visible, then the quantum computer can find the corresponding private

00:14:19.330 --> 00:14:24.120
<v Sjors Provoost>key. And that means that that's not a problem if that public key is never used

00:14:24.120 --> 00:14:28.090
<v Sjors Provoost>again. Hence the suggestion to reuse addresses. Right,

00:14:28.410 --> 00:14:31.650
<v Aaron van Wirdum>Because the public key is only revealed when money is spent.

00:14:32.630 --> 00:14:37.930
<v Sjors Provoost>Yes, at least until Taproot it was. So when this discussion was very

00:14:37.930 --> 00:14:42.490
<v Sjors Provoost>big, what transactions were doing generally is sending money not to a public key,

00:14:42.530 --> 00:14:46.350
<v Sjors Provoost>but to the hash of the public key, which quantum computers probably cannot deal

00:14:46.350 --> 00:14:52.070
<v Sjors Provoost>with, at least not as efficiently. But once you spend it... in order to spend it,

00:14:52.170 --> 00:14:55.270
<v Sjors Provoost>you have to reveal the public key. And so that's when the clock starts ticking

00:14:55.270 --> 00:15:00.010
<v Sjors Provoost>and the quantum computer could try and steal the money. Now, in a scenario where

00:15:00.010 --> 00:15:04.370
<v Sjors Provoost>you don't reuse addresses, there's only a very short period of time, right? It's

00:15:04.370 --> 00:15:07.770
<v Sjors Provoost>between the moment you make the transaction and it goes in the mempool and the

00:15:07.770 --> 00:15:12.330
<v Sjors Provoost>moment it confirms in a block. That's when the quantum computer could steal the

00:15:12.330 --> 00:15:15.430
<v Sjors Provoost>money because it could figure out what the private key is and basically try to

00:15:15.430 --> 00:15:18.510
<v Sjors Provoost>fee bump a transaction that goes through themselves.

00:15:18.930 --> 00:15:19.090
<v Aaron van Wirdum>Yeah.

00:15:19.770 --> 00:15:23.190
<v Sjors Provoost>Which of course could then be stolen by other quantum computers. So, you know, it

00:15:23.190 --> 00:15:28.170
<v Sjors Provoost>creates a bit of a rat race, but this is the problem. However, well, for various

00:15:28.170 --> 00:15:32.830
<v Sjors Provoost>reasons, one being that there's already a lot of coins on the blockchain that

00:15:32.830 --> 00:15:33.990
<v Sjors Provoost>have the public key known.

00:15:35.470 --> 00:15:37.930
<v Sjors Provoost>With Taproot, you're just spending to a public key.

00:15:38.330 --> 00:15:43.350
<v Aaron van Wirdum>Yeah, the reason there's a lot of coins on the blockchain of which the public key

00:15:43.350 --> 00:15:47.150
<v Aaron van Wirdum>is known is, for example, because a lot of people used xpubs.

00:15:48.200 --> 00:15:53.020
<v Sjors Provoost>Yeah, exactly. And then the question is if that xpub is known by anyone, they

00:15:53.020 --> 00:15:56.740
<v Sjors Provoost>know all those public keys. So the wallet providers know it and maybe some

00:15:56.740 --> 00:16:02.130
<v Sjors Provoost>government goes after them. There's all sorts of reasons. But the point is

00:16:02.130 --> 00:16:05.930
<v Sjors Provoost>there's so many coins out there for which the public key is known that if it gets

00:16:05.930 --> 00:16:08.010
<v Sjors Provoost>stolen by a quantum computer, it would just crash the market.

00:16:10.110 --> 00:16:13.990
<v Sjors Provoost>However, what's important to note there is the quantum computer would have

00:16:13.990 --> 00:16:18.190
<v Sjors Provoost>longer, right? Because if the public key is already known, then you can just

00:16:18.190 --> 00:16:22.410
<v Sjors Provoost>start trying to brute force it and take as much time as you need. Versus the

00:16:22.410 --> 00:16:25.570
<v Sjors Provoost>example where we talked about before when you only have a few minutes to do that.

00:16:27.200 --> 00:16:30.640
<v Sjors Provoost>And of course, we know the first coins that were mined, in the very beginning

00:16:30.640 --> 00:16:33.780
<v Sjors Provoost>when coins were mined, they were not sent to a hash at a public key, they were

00:16:33.780 --> 00:16:38.480
<v Sjors Provoost>sent to a public key. So that's already quite a lot of coins. So that's what the

00:16:38.480 --> 00:16:41.950
<v Sjors Provoost>quantum computer does. It has nothing to do with signatures though. It has to do

00:16:41.950 --> 00:16:42.450
<v Sjors Provoost>with spending.

00:16:42.730 --> 00:16:46.350
<v Aaron van Wirdum>Yeah, I know. I know we're taking a bit of a detour, but while we are on this

00:16:46.350 --> 00:16:51.530
<v Aaron van Wirdum>detour, do you think, and a lot of people do, like this is a question I get very

00:16:51.530 --> 00:16:57.410
<v Aaron van Wirdum>often, like is quantum computer, is the potential of quantum computer someone we

00:16:57.410 --> 00:17:02.050
<v Aaron van Wirdum>should start worrying about, should start thinking about, like to what extent,

00:17:02.230 --> 00:17:04.090
<v Aaron van Wirdum>where's your mind on this point?

00:17:04.870 --> 00:17:08.490
<v Sjors Provoost>Well, I think it's good to think about it, but I don't have the productive

00:17:08.490 --> 00:17:12.730
<v Sjors Provoost>capacity to think about it because I don't understand cryptography, let alone

00:17:12.730 --> 00:17:18.350
<v Sjors Provoost>quantum cryptography to enough of a degree that I could come up with a new way to

00:17:18.350 --> 00:17:23.570
<v Sjors Provoost>deal with it. So I try to listen to the few people who do keep an eye on that.

00:17:23.690 --> 00:17:25.870
<v Sjors Provoost>And it's not that many. It's like two or three.

00:17:27.250 --> 00:17:31.910
<v Sjors Provoost>And as soon as they say, hey, look at this possibility, then maybe I'll say,

00:17:32.090 --> 00:17:35.010
<v Sjors Provoost>okay, let's have a look at that. But from what I've heard, for example, from

00:17:35.010 --> 00:17:40.870
<v Sjors Provoost>Peter Wuille, sipa, and some others said, like, there are some mechanisms out

00:17:40.870 --> 00:17:44.510
<v Sjors Provoost>there that would be safe against quantum computers. However, they would make

00:17:44.510 --> 00:17:49.730
<v Sjors Provoost>transactions really, really large. I don't know like 10 kilobytes some like crazy

00:17:49.730 --> 00:17:54.930
<v Sjors Provoost>transaction size and they would be very hard on on blockchain validation so they

00:17:54.930 --> 00:17:59.280
<v Sjors Provoost>would just increase the amount of CPU that's needed. And also they don't have

00:17:59.280 --> 00:18:03.740
<v Sjors Provoost>many of the nice properties that [unclear] has. So, you know, all these things we

00:18:03.740 --> 00:18:08.120
<v Sjors Provoost>talked about that make Taproot so great, signature aggregation, you know, the

00:18:08.120 --> 00:18:13.880
<v Sjors Provoost>ability to combine signatures and to do cool stuff like the discrete log context

00:18:13.880 --> 00:18:17.480
<v Sjors Provoost>we talked about, all that stuff. The question is whether you can still do that

00:18:17.480 --> 00:18:23.180
<v Sjors Provoost>with quantum secure algorithms, and I'm not sure if that's true. So this is a

00:18:23.180 --> 00:18:26.380
<v Sjors Provoost>case where the longer you wait, the more people come up with better

00:18:26.380 --> 00:18:30.920
<v Sjors Provoost>methodologies. And perhaps also the easier it is to do whatever block size

00:18:30.920 --> 00:18:34.130
<v Sjors Provoost>increase is necessary to even fit the stuff, you know?

00:18:34.040 --> 00:18:34.560
<v Aaron van Wirdum>Right.

00:18:35.590 --> 00:18:39.870
<v Sjors Provoost>And the other side of it is like, how long is this going to take? And that could

00:18:39.870 --> 00:18:44.060
<v Sjors Provoost>be, you know, tens of years or millions of years. I have no idea.

00:18:45.330 --> 00:18:51.010
<v Sjors Provoost>There's a lot of hype about quantum computers, which is, you know, a lot of it is

00:18:51.010 --> 00:18:53.860
<v Sjors Provoost>actually just hype. But there are people working on it.

00:18:54.080 --> 00:18:54.420
<v Aaron van Wirdum>Right.

00:18:54.760 --> 00:18:59.000
<v Sjors Provoost>And, you know, there's one optimistic prediction I think I've heard from Stephen

00:18:59.000 --> 00:19:04.860
<v Sjors Provoost>Wolfram, known from Wolfram Alpha. And I believe his take, one of

00:19:04.860 --> 00:19:09.500
<v Sjors Provoost>his theories, and he's not sure about that, so we can't really count on that, is

00:19:09.500 --> 00:19:13.520
<v Sjors Provoost>that the whole idea of speeding things up with quantum computers is like a mirage

00:19:13.520 --> 00:19:19.070
<v Sjors Provoost>in the sense that theoretically it should be faster. And you can see that, that

00:19:19.070 --> 00:19:23.030
<v Sjors Provoost>it's faster. But in order to actually get the measurement out of the quantum

00:19:23.030 --> 00:19:27.650
<v Sjors Provoost>computer to actually use that information, every second you've gained basically

00:19:27.650 --> 00:19:31.770
<v Sjors Provoost>on the way in, you're going to lose on the way out. So the idea there would be

00:19:31.770 --> 00:19:34.510
<v Sjors Provoost>like, you're not actually going to speed up because it's going to get more and

00:19:34.510 --> 00:19:39.350
<v Sjors Provoost>more slow to actually measure it. But that's not proven. And so we can't count on

00:19:39.350 --> 00:19:42.770
<v Sjors Provoost>it. We can't say, well, maybe it's not going to happen. So let's just ignore the problem.

00:19:43.170 --> 00:19:46.060
<v Aaron van Wirdum>Right. Okay, that was a big detour.

00:19:46.320 --> 00:19:51.800
<v Sjors Provoost>So how this relates to address reuse, it used to. In the case of where you would

00:19:51.800 --> 00:19:55.820
<v Sjors Provoost>use the hash, then address reuse would be nice because then your exposure to the

00:19:55.820 --> 00:19:57.900
<v Sjors Provoost>quantum computer is only a few minutes.

00:19:58.320 --> 00:20:01.520
<v Aaron van Wirdum>Well, not reusing addresses would be nice, is what you mean.

00:20:01.540 --> 00:20:04.500
<v Sjors Provoost>Exactly. If you don't reuse addresses, then your exposure to the quantum computer

00:20:04.500 --> 00:20:08.480
<v Sjors Provoost>is only a few minutes in the case where you use hashes instead of public keys.

00:20:09.530 --> 00:20:13.250
<v Sjors Provoost>And that's, of course, a fallback we could always go back to, I guess. If there

00:20:13.250 --> 00:20:16.110
<v Sjors Provoost>is really a quantum computer, you could say, oh, this is not good. We're seeing

00:20:16.110 --> 00:20:20.570
<v Sjors Provoost>Satoshi's coins getting stolen. Well, I guess that's great for whoever steals

00:20:20.570 --> 00:20:24.070
<v Sjors Provoost>them. Probably not so good for the price. But hey, if you want to protect your

00:20:24.070 --> 00:20:26.930
<v Sjors Provoost>coins, you may want to use the hash of a public key for now.

00:20:27.950 --> 00:20:32.270
<v Aaron van Wirdum>I mean, we wouldn't know that they're being stolen even, right? Like, they're

00:20:32.270 --> 00:20:32.950
<v Aaron van Wirdum>just being spent.

00:20:33.970 --> 00:20:39.110
<v Sjors Provoost>Yeah, but the fact that they are suddenly getting spent... very slowly, perhaps,

00:20:39.610 --> 00:20:43.050
<v Sjors Provoost>maybe at an exponentially increasing pace, because the first quantum computer

00:20:43.050 --> 00:20:48.870
<v Sjors Provoost>might take a year to break the first public key of the first

00:20:48.870 --> 00:20:53.300
<v Sjors Provoost>block, maybe the Genesis block, and then maybe the second quantum computer takes

00:20:53.300 --> 00:20:57.700
<v Sjors Provoost>six months to do it, and so the second block gets spent, and then it goes faster

00:20:57.700 --> 00:21:01.760
<v Sjors Provoost>and faster and faster and faster. But at some point, people want to brag about it, right?

00:21:03.650 --> 00:21:07.810
<v Aaron van Wirdum>I mean, it depends how much money you can earn by keeping your mouth shut, maybe.

00:21:08.610 --> 00:21:10.890
<v Sjors Provoost>Yeah, not that much probably in the beginning.

00:21:12.140 --> 00:21:16.180
<v Sjors Provoost>Because if you're stealing, like it takes you a year to steal 50 Bitcoin from the

00:21:16.180 --> 00:21:18.680
<v Sjors Provoost>first block and that's worth what?

00:21:19.580 --> 00:21:21.220
<v Aaron van Wirdum>A million bucks?

00:21:21.440 --> 00:21:23.360
<v Sjors Provoost>No, it's a hundred thousand right now, isn't it?

00:21:24.310 --> 00:21:25.970
<v Aaron van Wirdum>50 is like [unclear] hundred bucks.

00:21:25.820 --> 00:21:26.440
<v Sjors Provoost>Yeah, it's a million.

00:21:26.670 --> 00:21:27.730
<v Aaron van Wirdum>Sorry, did I say a hundred bucks?

00:21:27.830 --> 00:21:28.450
<v Sjors Provoost>No, yeah, you're right.

00:21:28.610 --> 00:21:29.770
<v Aaron van Wirdum>I said a million bucks, right?

00:21:29.870 --> 00:21:30.530
<v Sjors Provoost>Yeah, you said it right.

00:21:30.730 --> 00:21:30.850
<v Aaron van Wirdum>Yeah.

00:21:31.470 --> 00:21:35.390
<v Sjors Provoost>So, okay, great. That's, you know, it's a nice amount of money. You can buy lots

00:21:35.390 --> 00:21:38.810
<v Sjors Provoost>of Frappuccinos, but you probably spend a couple billion on your quantum

00:21:38.810 --> 00:21:42.790
<v Sjors Provoost>computer. So it doesn't really, it doesn't really justify spending that much

00:21:42.790 --> 00:21:47.230
<v Sjors Provoost>money on it. And that depends on how fast it is. If from day one, it's super fast

00:21:47.230 --> 00:21:51.210
<v Sjors Provoost>and you can steal all the Bitcoin, then it might be worth doing and worth giving

00:21:51.210 --> 00:21:51.670
<v Sjors Provoost>you the secret.

00:21:54.340 --> 00:21:58.560
<v Aaron van Wirdum>Sure, I think it's time to end this detour and get back to the main topic of

00:21:58.560 --> 00:22:04.500
<v Aaron van Wirdum>address reuse and the final topic, the final category. Well, still

00:22:04.500 --> 00:22:09.060
<v Aaron van Wirdum>the category of security. And then this is sort of the reason we're talking about

00:22:09.060 --> 00:22:13.940
<v Aaron van Wirdum>all of this is that we found this on the Bitcoin Optech newsletter.

00:22:14.680 --> 00:22:19.800
<v Aaron van Wirdum>There was recently a new... vulnerability found

00:22:19.800 --> 00:22:23.180
<v Aaron van Wirdum>called the— referred to as the Hertzbleed attack.

00:22:23.180 --> 00:22:23.680
<v Sjors Provoost>Yep.

00:22:23.680 --> 00:22:29.420
<v Aaron van Wirdum>And this could affect Bitcoin and especially address reuse is that right?

00:22:29.420 --> 00:22:30.080
<v Sjors Provoost>Yep.

00:22:30.080 --> 00:22:32.180
<v Aaron van Wirdum>So what is the Hertzbleed attack?

00:22:32.180 --> 00:22:38.080
<v Sjors Provoost>So I don't know in detail but basically hertz is the frequency right so a 50

00:22:38.080 --> 00:22:42.720
<v Sjors Provoost>hertz is like a beeping sound, and so computers or CPUs…

00:22:42.720 --> 00:22:47.450
<v Aaron van Wirdum>So it's a pun. We have a pun in our episode again. Well, it refers to the

00:22:47.450 --> 00:22:49.130
<v Aaron van Wirdum>heartbeat attack, right? I would assume.

00:22:49.430 --> 00:22:53.130
<v Sjors Provoost>Yeah, so the pun is quite indirect because the word hertz is the word for

00:22:53.130 --> 00:22:58.950
<v Sjors Provoost>frequency, right? But it's named after a guy named Hertz who happened to be a

00:22:58.950 --> 00:23:03.830
<v Sjors Provoost>researcher who came up with this idea. But the last name of that person is

00:23:03.830 --> 00:23:05.910
<v Sjors Provoost>actually the Jewish word for heart.

00:23:07.800 --> 00:23:09.780
<v Aaron van Wirdum>Oh, it's even better than I thought.

00:23:09.960 --> 00:23:13.780
<v Sjors Provoost>Yes, it is. But I don't know if it's deliberate. I mean, it sounds the same, but

00:23:13.780 --> 00:23:14.540
<v Sjors Provoost>it's one letter difference.

00:23:14.700 --> 00:23:18.020
<v Aaron van Wirdum>I mean, even the logo is sort of similar. Like this is deliberate for sure.

00:23:18.600 --> 00:23:22.900
<v Sjors Provoost>Well, the reference to Hertz is, but in the sense of the frequency, not Herz,

00:23:23.120 --> 00:23:23.800
<v Sjors Provoost>which is heart.

00:23:24.100 --> 00:23:24.700
<v Aaron van Wirdum>Right, right, right.

00:23:25.260 --> 00:23:25.940
<v Sjors Provoost>So anyway.

00:23:26.240 --> 00:23:28.300
<v Aaron van Wirdum>Anyways, sure. So what is the Hertzbleed attack?

00:23:28.380 --> 00:23:32.700
<v Sjors Provoost>So I think the general principle is that… is that of a timing side channel

00:23:32.700 --> 00:23:36.620
<v Sjors Provoost>attack, which we'll explain in more detail. But basically, I believe it has to do

00:23:36.620 --> 00:23:42.280
<v Sjors Provoost>with how CPUs change frequencies depending on what they're doing. But I don't

00:23:42.280 --> 00:23:45.240
<v Sjors Provoost>actually know that that well, so I'll just describe the general problem of these

00:23:45.240 --> 00:23:48.960
<v Sjors Provoost>side channel attacks and how they relate to address reuse.

00:23:49.260 --> 00:23:49.600
<v Aaron van Wirdum>Okay, yeah.

00:23:49.900 --> 00:23:54.480
<v Sjors Provoost>So the idea is when you're actually making a signature, the act of making a

00:23:54.480 --> 00:23:58.860
<v Sjors Provoost>signature involves doing some sort of mathematical operation on the message, so

00:23:58.860 --> 00:24:02.760
<v Sjors Provoost>the transaction, and you're doing that mathematical operation using your private

00:24:02.760 --> 00:24:07.800
<v Sjors Provoost>key. And so how does that work in practice? Well, your computer looks at your

00:24:07.800 --> 00:24:12.500
<v Sjors Provoost>private key and starts, it's 256 bits. It starts at the first bit. That's either

00:24:12.500 --> 00:24:16.540
<v Sjors Provoost>a zero or a one. And then it does one calculation if it's a zero and another

00:24:16.540 --> 00:24:21.920
<v Sjors Provoost>calculation if it's a one. Essentially, it's kind of like, or you could say it's

00:24:21.920 --> 00:24:27.200
<v Sjors Provoost>doing multiplications, for example, or some elliptic curve operation. It does

00:24:27.200 --> 00:24:30.980
<v Sjors Provoost>that on every bit. So the problem there is that when you're doing an operation

00:24:30.980 --> 00:24:35.790
<v Sjors Provoost>with a zero, your CPU is going to behave slightly different than when you're

00:24:35.790 --> 00:24:40.010
<v Sjors Provoost>doing an operation on a one, if you're not careful. So when you're writing a

00:24:40.010 --> 00:24:43.310
<v Sjors Provoost>cryptographic library, you want to make sure, and the reason is, well, I'll

00:24:43.310 --> 00:24:47.660
<v Sjors Provoost>explain that after. You want to make sure that whatever operation happens after a

00:24:47.660 --> 00:24:52.630
<v Sjors Provoost>zero takes exactly the same amount of time and energy and whatever as the

00:24:52.630 --> 00:24:54.170
<v Sjors Provoost>operation you would do when there's a one.

00:24:54.470 --> 00:24:54.610
<v Aaron van Wirdum>Right.

00:24:55.490 --> 00:24:59.270
<v Sjors Provoost>Because otherwise somebody's going to listen to your CPU. You know, I'm saying

00:24:59.270 --> 00:25:03.770
<v Sjors Provoost>listen quite loosely, but it's like you have a stethoscope and you put it on the

00:25:03.770 --> 00:25:07.710
<v Sjors Provoost>CPU and you listen to it and you hear that it's like…

00:25:07.710 --> 00:25:13.650
<v Sjors Provoost>And that tells you, okay, so it's, you know, it's reading zeros, it's

00:25:13.650 --> 00:25:14.090
<v Sjors Provoost>reading ones.

00:25:14.270 --> 00:25:16.510
<v Aaron van Wirdum>I didn't know you speak computer.

00:25:16.690 --> 00:25:20.850
<v Sjors Provoost>I speak computer. So if you can somehow hear in the widest sense of the word

00:25:20.850 --> 00:25:26.530
<v Sjors Provoost>here, that the CPU is processing 256 bits, then you know what the private key is

00:25:26.530 --> 00:25:31.110
<v Sjors Provoost>and you can extract it. And so cryptographers make sure that they write the code

00:25:31.110 --> 00:25:36.010
<v Sjors Provoost>in a way that you don't hear the difference. But unfortunately, then it turns out

00:25:36.010 --> 00:25:41.690
<v Sjors Provoost>all the time there's some problem with, in this case, I guess the CPUs that spoil

00:25:41.690 --> 00:25:44.790
<v Sjors Provoost>the result anyway. It's some sort of behavioral change.

00:25:45.030 --> 00:25:48.290
<v Aaron van Wirdum>Yeah, and it's basically like all CPUs, like Intel.

00:25:49.490 --> 00:25:54.270
<v Aaron van Wirdum>Basically, if you have a CPU, you're affected. That's sort of what Hertzbleed page suggests.

00:25:54.270 --> 00:25:57.690
<v Sjors Provoost>Yeah, but the thing is about these type of attacks is you need to get pretty

00:25:57.690 --> 00:26:02.330
<v Sjors Provoost>close to somebody's CPU to be able to do it, for one thing. Now, there's two ways

00:26:02.330 --> 00:26:05.850
<v Sjors Provoost>to get close to somebody's CPU. One is you can physically visit somebody and put

00:26:05.850 --> 00:26:06.770
<v Sjors Provoost>something in their computer.

00:26:08.580 --> 00:26:14.480
<v Sjors Provoost>The other is that you're running some program on the computer. So you're

00:26:14.480 --> 00:26:18.300
<v Sjors Provoost>running, you basically put some malware on somebody's computer. And this malware

00:26:18.300 --> 00:26:24.020
<v Sjors Provoost>could be a JavaScript advertisement on Google sitting somewhere, right, in a

00:26:24.020 --> 00:26:26.400
<v Sjors Provoost>browser tab that you're not even paying attention to. And this JavaScript

00:26:26.400 --> 00:26:30.320
<v Sjors Provoost>advertisement is very carefully trying to listen to what your computer is doing.

00:26:31.320 --> 00:26:36.140
<v Sjors Provoost>It might be, for example, your JavaScript advertisement might be doing a simple

00:26:36.140 --> 00:26:39.870
<v Sjors Provoost>calculation itself. And it's measuring how long does this simple calculation

00:26:39.870 --> 00:26:43.210
<v Sjors Provoost>take. And one, you know, it finds out, oh, my simple calculation took 10

00:26:43.210 --> 00:26:47.170
<v Sjors Provoost>milliseconds, and then it took 30 milliseconds, and now it took 10 again. Hmm,

00:26:47.250 --> 00:26:52.150
<v Sjors Provoost>the CPU must be really busy with something. And then if that something is a nice

00:26:52.150 --> 00:26:55.810
<v Sjors Provoost>repeating pattern, then eventually it'll be able to detect it. But that gets to

00:26:55.810 --> 00:27:00.770
<v Sjors Provoost>the second problem. It's not a very clear signal. So you probably want to measure

00:27:00.770 --> 00:27:05.900
<v Sjors Provoost>like 100 times or 1,000 times to see if you see a repeating pattern. And if you

00:27:05.900 --> 00:27:10.500
<v Sjors Provoost>see a repeating pattern of 256 bits, then you can guess, oh, that's probably a

00:27:10.500 --> 00:27:14.820
<v Sjors Provoost>private key. Let me give that to my boss and see if they can steal some Bitcoin.

00:27:15.540 --> 00:27:19.660
<v Sjors Provoost>So that's where the address reuse gets in, because if you only use the private

00:27:19.660 --> 00:27:20.260
<v Sjors Provoost>key once,

00:27:21.650 --> 00:27:25.030
<v Sjors Provoost>then you've only got one shot to try and measure it through this indirect method,

00:27:25.030 --> 00:27:28.750
<v Sjors Provoost>and you probably won't get a very good signal-to-noise ratio, and so you probably

00:27:28.750 --> 00:27:30.470
<v Sjors Provoost>won't be able to write out the key.

00:27:30.650 --> 00:27:36.070
<v Aaron van Wirdum>So if you reuse addresses and therefore reuse signatures, you're more

00:27:36.070 --> 00:27:37.130
<v Aaron van Wirdum>clearly...

00:27:37.130 --> 00:27:42.070
<v Sjors Provoost>Yeah, but even then, this would assume you're reusing the same address hundreds

00:27:42.070 --> 00:27:45.770
<v Sjors Provoost>of times a second on the same computer, not over the space of a month.

00:27:45.950 --> 00:27:46.230
<v Aaron van Wirdum>Right.

00:27:46.330 --> 00:27:49.570
<v Sjors Provoost>Or a year, because that malware is not going to be sitting on your... Well, it

00:27:49.570 --> 00:27:53.010
<v Sjors Provoost>might be if it's patient, but probably not.

00:27:54.490 --> 00:27:58.430
<v Aaron van Wirdum>Right. So it's a very theoretical attack again, but it's possible.

00:27:58.430 --> 00:28:02.150
<v Sjors Provoost>I would guess so, but I haven't read it in enough detail. It depends on what

00:28:02.150 --> 00:28:06.670
<v Sjors Provoost>you're doing. There are keys that are used a lot more often. There, it's a bigger problem.

00:28:07.090 --> 00:28:07.310
<v Aaron van Wirdum>Right.

00:28:07.430 --> 00:28:10.390
<v Sjors Provoost>But yeah, address reuse helps in that sense that every key is only used once.

00:28:11.250 --> 00:28:15.630
<v Aaron van Wirdum>Right. The Bitcoin Optech newsletter does point out that this problem was sort of

00:28:15.630 --> 00:28:20.130
<v Aaron van Wirdum>already known within Bitcoin because this is how some of the hardware wallets

00:28:20.130 --> 00:28:21.310
<v Aaron van Wirdum>were attacked in the past.

00:28:21.670 --> 00:28:25.090
<v Sjors Provoost>Well, this general pattern of this timing has been used against hardware wallets.

00:28:25.170 --> 00:28:30.190
<v Sjors Provoost>Yeah, exactly. And there it's a little bit different. So in the case of a

00:28:30.190 --> 00:28:35.250
<v Sjors Provoost>hardware wallet, the scenario you might be worried about is some thief gets your

00:28:35.250 --> 00:28:39.490
<v Sjors Provoost>hardware wallet physically in their hands. And now they can sign the same message

00:28:39.490 --> 00:28:43.370
<v Sjors Provoost>as often as they want. So this is very different from you just casually using

00:28:43.370 --> 00:28:46.830
<v Sjors Provoost>your own wallet and you might only use your address once. But if you've got

00:28:46.830 --> 00:28:49.830
<v Sjors Provoost>somebody's hardware wallet, you can probably make it do something a million times

00:28:49.830 --> 00:28:55.470
<v Sjors Provoost>over and then listen very carefully and you might be able to hear the private key

00:28:55.470 --> 00:28:59.630
<v Sjors Provoost>on the wallet and such attacks. So yeah, that's familiar.

00:29:00.680 --> 00:29:05.600
<v Aaron van Wirdum>Okay, well, in conclusion, I would say the biggest problem with address reuse

00:29:05.600 --> 00:29:11.360
<v Aaron van Wirdum>right now by far is still privacy. And then security is sort of theoretical.

00:29:12.100 --> 00:29:17.120
<v Aaron van Wirdum>There are some potential vulnerabilities, but it's really a privacy problem.

00:29:17.340 --> 00:29:20.760
<v Sjors Provoost>Yeah, I mean, when it comes to security, the address reuse itself is not the

00:29:20.760 --> 00:29:24.740
<v Sjors Provoost>vulnerability. It is something that can help slightly reduce the impact of the

00:29:24.740 --> 00:29:25.220
<v Sjors Provoost>vulnerability.

00:29:26.900 --> 00:29:32.720
<v Aaron van Wirdum>Okay. Yeah, there are some more problems mentioned on the wiki, but I think

00:29:32.720 --> 00:29:38.700
<v Aaron van Wirdum>these are sort of very niche, like barely worth mentioning. Like people

00:29:38.700 --> 00:29:42.620
<v Aaron van Wirdum>might throw away their private key because they assume that they won't be paid on

00:29:42.620 --> 00:29:47.260
<v Aaron van Wirdum>the same address. But then if they are, well, then the money is lost. I guess

00:29:47.260 --> 00:29:51.560
<v Aaron van Wirdum>that's true. But why would you throw away a private key? I don't know. It just

00:29:51.560 --> 00:29:54.560
<v Aaron van Wirdum>seems kind of like grasping at straws for problems.

00:29:54.840 --> 00:29:58.980
<v Sjors Provoost>Well, this is probably more from before the time than the addresses were

00:29:58.980 --> 00:30:03.620
<v Sjors Provoost>deterministic. So right now you still have the master key, but you keep getting a

00:30:03.620 --> 00:30:04.100
<v Sjors Provoost>new address.

00:30:06.330 --> 00:30:12.040
<v Aaron van Wirdum>There's something else that's mentioned is that it could sort of… it disrupts fee

00:30:12.040 --> 00:30:16.300
<v Aaron van Wirdum>estimation a bit. If you have several outputs on the same address and you want to

00:30:16.300 --> 00:30:21.680
<v Aaron van Wirdum>spend one, then the wallet might not be able to correctly calculate how much it

00:30:21.680 --> 00:30:22.880
<v Aaron van Wirdum>is for some reason.

00:30:23.280 --> 00:30:24.780
<v Sjors Provoost>That does not ring a bell.

00:30:25.060 --> 00:30:28.800
<v Aaron van Wirdum>I didn't really understand that argument either, to be honest, and it seems very

00:30:28.800 --> 00:30:34.280
<v Aaron van Wirdum>niche anyways. And then there was some problem that doesn't even really sound

00:30:34.280 --> 00:30:38.080
<v Aaron van Wirdum>like a problem to me, which is just confusion about terminology.

00:30:39.930 --> 00:30:44.690
<v Aaron van Wirdum>I mean, I guess that's been an argument that we shouldn't even use the word

00:30:44.690 --> 00:30:50.330
<v Aaron van Wirdum>address for Bitcoin addresses because an address like an email address or the

00:30:50.330 --> 00:30:54.970
<v Aaron van Wirdum>address where you live does suggest that it's going to be the same thing that you

00:30:54.970 --> 00:30:56.270
<v Aaron van Wirdum>send stuff to every time.

00:30:56.680 --> 00:30:57.040
<v Sjors Provoost>Yeah.

00:30:57.400 --> 00:31:03.360
<v Aaron van Wirdum>So a better term would be Bitcoin invoice or, you know, someone, a

00:31:03.360 --> 00:31:07.460
<v Aaron van Wirdum>Belcher suggested invoice address and then hopefully after a while people will

00:31:07.460 --> 00:31:12.600
<v Aaron van Wirdum>just forget about the address part. But address is not a great name actually for

00:31:12.600 --> 00:31:13.500
<v Aaron van Wirdum>a Bitcoin address.

00:31:14.120 --> 00:31:16.780
<v Sjors Provoost>Yeah, it sounds like one of those things that will just stick around.

00:31:17.310 --> 00:31:20.410
<v Aaron van Wirdum>Yeah, I don't think we're going to get rid of it at this point.

00:31:20.990 --> 00:31:25.790
<v Sjors Provoost>And also, if you try to change the terminology, that's really just difficult.

00:31:26.110 --> 00:31:30.790
<v Sjors Provoost>Generally, it might be easier to just make sure that people understand how things

00:31:30.790 --> 00:31:33.250
<v Sjors Provoost>work and eventually they don't worry too much about the terminology.

00:31:33.470 --> 00:31:33.710
<v Aaron van Wirdum>Right.

00:31:35.570 --> 00:31:40.030
<v Aaron van Wirdum>Anyways, yeah, that's our episode, I think, Sjors. I think we made it to the end.

00:31:40.430 --> 00:31:40.810
<v Sjors Provoost>Excellent.

00:31:41.030 --> 00:31:44.910
<v Aaron van Wirdum>I thought this was going to be an episode of like five minutes, but I think we

00:31:44.910 --> 00:31:48.790
<v Aaron van Wirdum>were able to stretch it to, what was it, half an hour maybe in the end?

00:31:49.650 --> 00:31:51.750
<v Sjors Provoost>I don't know. It depends on what the editor makes of it.

00:31:52.010 --> 00:31:52.350
<v Aaron van Wirdum>True.

00:31:52.450 --> 00:31:55.250
<v Sjors Provoost>So I guess thank you for listening to Bitcoin

00:31:55.250 --> 00:31:56.170
<v Aaron van Wirdum>Explained.

