WEBVTT

NOTE S17 E32: Perry E. Metzger on Bitcoin in 2008, Cypherpunk Culture &amp; AI Optimism. Corrected transcript from bitcoin-takeover.com

1
00:00:45.000 --> 00:01:26.000
<v Vlad Costea>Hello there and welcome to the Bitcoin Takeover Podcast. This is Season 17, Episode 32. This episode is sponsored by SideShift, Cake Wallet, Orange Rock, LayerTwo Labs, and Braiins Hashpower. And we are live with quite a legend. He is the owner of the Cryptography Mailing List where Bitcoin was first announced. His name is Perry Metzger. His Cryptography Mailing List is metzdowd.com, which is still active. You can still find people presenting interesting ideas in there. And he's also the leader, I'm not sure about the position exactly, of an organization that deals with advocacy for AI.

2
00:01:26.000 --> 00:01:46.000
<v Perry E. Metzger>And nowadays, it's called Alliance for the Future. It's in US terms, it's a 501(c)(4), which means we're a non-profit, but we're not tax deductible because we do political advocacy. And I do all sorts of other things too. But yeah, thank you for having me here, Vlad.

3
00:01:46.000 --> 00:01:55.000
<v Vlad Costea>Well, it's quite an honor to prepare for this interview. I've actually watched your conversation with Naomi Brockwell from about eight years ago.

4
00:01:55.000 --> 00:02:00.000
<v Perry E. Metzger>Oh my, I had forgotten that I had even done that. Naomi is an old friend.

5
00:02:00.000 --> 00:02:04.000
<v Vlad Costea>I hope you will not say the same about this one.

6
00:02:04.000 --> 00:02:28.000
<v Perry E. Metzger>Yeah, Naomi is a very old friend. I'm actually on the board of her nonprofit, which is called the Ludlow Institute. It advocates for privacy, which is an increasingly difficult thing to achieve in the modern world. But nevermind that. So what can I do for you today, Vlad?

7
00:02:28.000 --> 00:02:41.000
<v Vlad Costea>Well, let's structure this in a chronological order. And I think the main reason why people are going to tune in is just stories from 2008, when Satoshi emerged on the mailing list.

8
00:02:41.000 --> 00:02:43.000
<v Perry E. Metzger>Oh, the story goes back a lot further than that, though.

9
00:02:43.000 --> 00:02:51.000
<v Vlad Costea>Oh, yeah, we can start with that. And then we can get to the present because I know you're also into open source AI, and you don't like—

10
00:02:51.000 --> 00:02:52.000
<v Perry E. Metzger>All sorts of things.

11
00:02:52.000 --> 00:02:58.000
<v Vlad Costea>Qwen and Kimi getting banned or censored by the American government, apparently. It's kind of a serious conversation.

12
00:02:58.000 --> 00:03:13.000
<v Perry E. Metzger>I don't think that's imminent, but that's another thing we could discuss later on. I'm sure most of your listeners are more interested in, you know, cryptography and cryptocurrencies and privacy and that sort of thing.

13
00:03:13.000 --> 00:03:24.000
<v Vlad Costea>I just wanted to present, like, the timeline where we're starting and where we're going to end. Let's get started with your background. Like, you've been on the internet for, like, what, four decades?

14
00:03:24.000 --> 00:04:18.000
<v Perry E. Metzger>Well, I started using it in 1983, which was just after something called the TCP transition, which is when the ARPANET officially completely cut over to TCP/IP. So I've been on, but, you know, there are people I know who, you know, I'm friends with one of the people that turned on the first IMPs, right, in 1968. You know, so I'm kind of a newcomer compared to people like that. So yeah, I've been on the internet a while. I've watched the thing go from, well, when I started, you know, at my university, there was a sheet of paper in the machine room that had a map of all of the computers on the internet.

15
00:04:18.000 --> 00:04:53.000
<v Perry E. Metzger>You know, like all of them, you know, with, you know, lines connecting them to, you know, to indicate where the leased lines connected all the machines together. So, and today, I don't think that anyone even has a particularly good estimate of how many computers are on the internet, let alone a map of all of the computers on the internet that fits on a single sheet of paper. So it's been a bit of time. I've gotten to see this thing go from something that was used by only a relatively few people to something that's, you know, as common as dirt these days.

16
00:04:53.000 --> 00:05:28.000
<v Perry E. Metzger>You know, I think that people take the internet so much for granted that, you know, that they get rather surprised when it isn't working correctly where they are, but never mind that. Yeah, and so I've been using the internet for most of the time it's existed and, you know, I have a degree in computer science. I also have an unfinished graduate degree in computer science. I'm, you know, ABD, which means that I never finished my doctoral dissertation.

17
00:05:28.000 --> 00:05:32.000
<v Vlad Costea>Oh, I'm also a PhD dropout.

18
00:05:32.000 --> 00:06:12.000
<v Perry E. Metzger>Yes, I used to joke to the other people in the lab, you know, don't be a fool, stay in school. You don't want to have to go back and do your doctoral dissertation when you're an old man, you know, get it out of the way now while you're young, but never mind that. Yes, so yeah, you should ask questions, it'll make it easier. Otherwise, I'll just babble at you incessantly. I can babble for hours, but that won't be interesting to most people. It might be interesting to, you know, people who'd want to do psychological analyses of me, but you know, I don't think there are many of those. Hopefully. Hopefully, yes.

19
00:06:12.000 --> 00:06:34.000
<v Vlad Costea>Okay, so how did you start this cryptography mailing list? Because I've had Zooko on, and he told me that there was a split at one time, where there was this cypherpunk mailing list, which was getting too political, and you decided to create something that was more neutral and centered around ideas, not necessarily ideologies.

20
00:06:34.000 --> 00:07:19.000
<v Perry E. Metzger>Well, it was, it was the, so there was a, well, so all of this goes back even further than that, right? But yes, there was at one point, there was the cypherpunk mailing list, and it got noisier and noisier, and it was hard to discuss technical issues on it. And so I decided at one point that I was going to create a spinoff mailing list, which was the cryptography mailing list. And the purpose of that mailing list was to be a place for people who were interested in cryptographic, in cryptography, you know, privacy, the politics of these things to discuss technical issues, rather than to just endlessly spin, arguing with each other in various ways.

21
00:07:19.000 --> 00:08:00.000
<v Perry E. Metzger>And so I did that, and very rapidly, you know, a lot of people, you know, climbed on board. And, but the, you know, the motivations for all of these things go back way further than that. You know, you might equally ask, well, how did we end up with the cypherpunk's mailing list, or any of those things? You know, I've been interested in cryptography since I was a kid. I say, as a, I think as a young teen, in fact, it may have even been before that, I was wandering through my local library, and I found this book, you know, called The Codebreakers by David Kahn.

22
00:08:00.000 --> 00:08:43.000
<v Perry E. Metzger>It was only many decades later that I found out that Whit Diffie also got interested in cryptography, because he had come across a copy of The Codebreakers. But I met David Kahn, actually, some years ago, and I had the, you know, I was, had a fun conversation with him. Part of it was, you know, both of us being New Yorkers. And, you know, just talking about being New Yorkers. And part of it was also my, you know, asking him if he had had any idea what it was that his little book on the history of cryptography was going to do to the world, which I don't think he had at the time.

23
00:08:43.000 --> 00:09:23.000
<v Perry E. Metzger>But I was interested in cryptography for many, many, many years. But then, early 1990s, someone pointed out a presentation to me that had turned into a paper by a guy named Chuck Hammill called From Crossbows to Cryptography, which is probably the ancestor of all of this stuff that, you know, you're interested in or your audience might be interested in. And it's a really, really cool document. I never met Chuck Hammill, and I think that he's been dead for some years, and I don't know anything else that he did.

24
00:09:23.000 --> 00:09:59.000
<v Perry E. Metzger>But that one paper, and that one presentation that he gave that turned into the paper had an outsized amount of effect on the world. One of the lessons I've learned over the years is you never have any idea which things that you do are going to have lasting impact on the world around you. You can't predict which ones they will be, and you have no idea whether, you know, some loose talk that you might engage in one morning might end up having, you know, all sorts of disastrous or wonderful effects.

25
00:09:59.000 --> 00:11:11.000
<v Perry E. Metzger>You know, our lives are governed by amazing chance. But anyway, a friend of mine named Russell Whitaker, you know, spread around copies of this document written by Chuck Hammill. And Chuck's thesis was that cryptography was sort of the crossbow of the modern world in so far as that without cryptography, people are, you know, how to... well, you should go off and read the thing on your own. It's a very, very interesting document. But the underlying hypothesis was that just as crossbows had brought an end to the age of knights being in a particularly special situation in European military history, because, or, you know, a peasant could bring down a knight given a crossbow, you know, and you didn't need to spend years or decades studying how to use it very carefully, the way you might with a longbow.

26
00:11:11.000 --> 00:11:54.000
<v Perry E. Metzger>So cryptography provided a sort of great equalizer in the relationship between people and those who would violate their privacy or get in the way of their being able to communicate securely. There's more to it than that. I mean, it's a libertarian tract. And, you know, it has the sorts of things in it that you would expect from that. But this got me interested in cryptography. And I also encountered a bunch of people in the early 1990s, like the late Tim May, who I was friends with some years for, you know, for some years, and then we fell out of touch.

27
00:11:54.000 --> 00:12:47.000
<v Perry E. Metzger>And eventually he ended up spending, you know, the last years of his life as something of a recluse and an internet crank. But he'd been a very, very interesting person in the early 1990s. And he and a bunch of his friends in Silicon Valley, you know, had meetups to discuss cryptography and cryptographic technology and, you know, the privacy implications of it. And, you know, and to study various things like, you know, there were early cryptographic currency proposals like, I'm trying to remember the name of the gentleman, you know, a European cryptographer who invented blind signatures. Among other things and proposed using them for all sorts of things, including digital cash.

28
00:12:47.000 --> 00:12:48.000
<v Vlad Costea>You mean David Chaum?

29
00:12:48.000 --> 00:13:28.000
<v Perry E. Metzger>Yes, David Chaum. Chaum is an interesting fellow, had lots of interesting ideas. I think that a lot of the ideas that he had are still underexploited. You know, but anyway, there were lots of ideas floating around about digital cash systems and, you know, systems that would allow you, for example, to check whether, you know, cryptographically, whether a financial institution was solvent. Or what, you know, what sorts of transactions it was conducting, which you would want to know if it was, for example, handling the backing for some sort of digital currency.

30
00:13:28.000 --> 00:14:14.000
<v Perry E. Metzger>There were all sorts of discussions about, well, I skipped something, which is that the cypherpunks mailing list showed up at some point. And this was also the midst of the era where, you know, PGP emerged from out of nowhere and where the U.S. government attempted at various points to crack down on the use of cryptographic technology by private individuals. It seems to have largely given up on that for the moment, although various European governments have not. You know, Europe has the wonderful, you know, what is the chat censorship and monitoring? It's easy.

31
00:14:14.000 --> 00:14:16.000
<v Vlad Costea>It's chat control.

32
00:14:16.000 --> 00:14:26.000
<v Perry E. Metzger>Oh, yes, chat control. Yes, yes. You know, my condolences on living inside the European Union, too. It's fine.

33
00:14:26.000 --> 00:14:30.000
<v Vlad Costea>At least it's not as bad as it is in the U.K.

34
00:14:30.000 --> 00:14:47.000
<v Perry E. Metzger>Well, the U.K. Is a very special place. Hold on for one moment. I need a Kleenex because I have bronchitis and I, you know, I'm going to mute for just a moment so your listeners don't get to hear me honking into the microphone.

35
00:14:47.000 --> 00:14:48.000
<v Vlad Costea>It's all good.

36
00:14:48.000 --> 00:15:33.000
<v Perry E. Metzger>I should probably have gotten a glass of water for myself, but never mind that. Anyway, so, you know, early 1990s, you know, I encountered a bunch of people. Everyone from, you know, everyone from Tim May to, you know, to, you know, to I'm having a senior moment. I know to Phil Zimmermann and other folks like that. I wrote the documentation for the 2.0 release of PGP, for example. I had a lot of interest in. That was huge, right?

37
00:15:33.000 --> 00:15:38.000
<v Vlad Costea>Because Hal Finney had written the trusted setup code.

38
00:15:38.000 --> 00:16:05.000
<v Perry E. Metzger>Hal Finney wrote a lot of the code for that. I wrote the documentation. Phil took a lot of the flak, which was nice. You know, I did not particularly fancy being under investigation by the federal government, but, you know, I have not spoken to Phil Zimmermann in many years now. So I, you know, I have no idea how he's doing these days.

39
00:16:05.000 --> 00:16:18.000
<v Vlad Costea>Well, he still goes to conferences and talks about his experiences. And I do like the metaphor. So in some ways he's very predictable because I've seen him in a couple of places and he makes the same speech.

40
00:16:18.000 --> 00:16:20.000
<v Perry E. Metzger>And he does the same talk repeatedly.

41
00:16:20.000 --> 00:16:39.000
<v Vlad Costea>Yeah. He tells the same story. But I do like the metaphor where he compares the world to this LCD screen where there are some dead pixels. And then you have some authoritarians who want the image to be very bright and clear and crisp. And they don't want these dead pixels to exist at all.

42
00:16:39.000 --> 00:17:05.000
<v Vlad Costea>So they want to replace them with those that are very bright and clear. And he's like, don't let these pixels die. So I do like this part of his messaging. He does seem sometimes like he watches too much television. Last time I met him, like two or three years ago, he was talking about the end of American democracy. I was like, are you sure about that? He was like, yeah, I'm telling you, that's what's going to happen.

43
00:17:05.000 --> 00:17:44.000
<v Perry E. Metzger>Well, you know, I mean, he and I were, we shared an interest in cryptography. So I think that I was the one who came up with the stupid idea of doing cross signatures on PGP keys. I call it a stupid idea because the web of trust stuff turned out to conflate a lot of things that should not have been conflated. The fact that you think that someone actually owns a particular key does not mean that you think that they're necessarily a good trusted introducer for other people, right?

44
00:17:44.000 --> 00:18:20.000
<v Perry E. Metzger>But the early design conflated a bunch of things that should not have been mixed up. I may be responsible for some of those bad design decisions. I take responsibility for a bunch of the stupidity there. I do still have, I think, the oldest active key in some of the PGP key servers, which I've left there for amusement sake, because it's dated from before the actual release of PGP 2.0. And I want to leave the artifact in place, even though it's not of much use for anything.

45
00:18:20.000 --> 00:19:01.000
<v Perry E. Metzger>I don't think I've actually used PGP for encrypting anything in a long time. Although these days, almost all of my chats and all of my communications are encrypted and use very good cryptography. So I suppose we've won in certain respects, right? There was a long time when I would not have imagined a day where I could pick up my phone and I could call random friends and have the conversation end to end encrypted and even have rolling changes in the keys, which is what happens with the Signal protocol, for example. But these days, that is the case. And it's a pretty good thing, I think.

46
00:19:01.000 --> 00:19:47.000
<v Perry E. Metzger>But the Cypherpunks mailing list showed up. And lots and lots of people expressed lots of intemperate views for quite a while on all sorts of topics. It was a fun and loud and raucous time. And eventually I created the cryptography mailing list because the Cypherpunks list was too noisy. And I wanted a place for people to be discussing things like cryptographic communication protocols and ways to improve security. I mean, you should understand, these days we're very used to using cryptography for almost everything, right? Almost every HTTP request that gets made by your web browser these days is SSL protected.

47
00:19:47.000 --> 00:20:30.000
<v Perry E. Metzger>But in the early days of the internet, nothing was encrypted. You could listen in on all communications. These days, we take it for granted, for example, that if you connect to a remote computer and you're a systems administrator or something, most of your listeners won't know what SSH is. But there is a protocol that Dr. Ylönen came up with for allowing people to connect to remote computers and to have their management traffic with the remote computers completely protected by cryptography. And this was not, you know, this was not a thing early on.

48
00:20:30.000 --> 00:21:08.000
<v Perry E. Metzger>Everything you typed over the network went in the clear. If you connected to a remote computer, all of your keystrokes were in the clear. If you downloaded files from remote computer, all of the downloads were in the clear. And now most of the traffic is encrypted. And this is pretty impressive and useful. You know, so in certain sense, like to a large extent, in fact, the Cypherpunk vision for the internet has succeeded in so far as most of the traffic is encrypted these days. Unfortunately, of course, as you know, have you had Naomi Brockwell on?

49
00:21:08.000 --> 00:21:20.000
<v Vlad Costea>Well, to my shame, I haven't. Like, I met her a few times, but I never directly interviewed her because we sort of do kind of the same thing. And it feels awkward. Like, what are we going To ask each other?

50
00:21:20.000 --> 00:21:54.000
<v Perry E. Metzger>But no, she has plenty of stuff to talk about, because she has a great talk that she does on privacy and the erosion of privacy. And I was going to mention that, you know, she does this talk about, you know, how bad things are in terms of the fact that we all wear tracking devices, you know, all these things, all day long, wherever we go. And the tracking devices, you know, tell lots of people where we are, and people sell this information and, you know, collect and store this information forever. And this is not a great thing, right?

51
00:21:54.000 --> 00:22:34.000
<v Perry E. Metzger>So, you know, to some extent, we've gotten rid of the problem that most of our communications were in the clear, but we've developed an incredible host of new problems, you know, with privacy. And, you know, but that's, as I said, that's more Naomi's schtick than my schtick. I don't discuss that stuff particularly often. But anyway, there was this exciting time when everyone was talking about securing the internet, and it wasn't clear what was going to happen. And the United States government proposed things like the Clipper chip, you know, which was going to allow them to listen in on all communications all the time.

52
00:22:34.000 --> 00:22:42.000
<v Perry E. Metzger>Because after all, why wouldn't you listen in on all communications all the time? Just like the European Union is going to do to your chat soon.

53
00:22:42.000 --> 00:22:49.000
<v Vlad Costea>But they did put backdoors in pretty much every CPU that's been launched for the last 20 years or something.

54
00:22:49.000 --> 00:23:37.000
<v Perry E. Metzger>I would disagree with that. First of all, a large fraction of the CPU designs out there are not done, you know, in the United States. And I think that at this point, you know, there are certainly a lot of security problems out there. And there certainly are supply chain attacks that people face. But I don't think that CPUs are, that most CPUs have a lot of backdoors. I worry about things like, for example, the random number generators in Intel hardware, and other things like that. But that's another long story. Anyway, there was this exciting period, you know, where people in the IETF were proposing various ways to encrypt internet traffic.

55
00:23:37.000 --> 00:24:14.000
<v Perry E. Metzger>And, you know, and people in the government were opposing having internet traffic encrypted, which was, you know, which was annoying. And over the long term, we eventually got to the point where things are not so bad in certain respects, and are pretty bad in other respects. But the topic of digital currencies and, you know, and monetary systems that were either privacy preserving or did not involve the government at all was a, you know, was a topic of conversation in all sorts of circles, including among economists for many decades.

56
00:24:14.000 --> 00:24:53.000
<v Perry E. Metzger>You know, Friedrich Hayek wrote an essay decades ago, of course, it had to be decades ago, he's been dead for decades, about the denationalization of money and that sort of thing. So there was this big continuing topic of interest among a lot of people in, you know, in providing privacy preserving transactions over the internet. And, you know, Chaum's proposals got a lot of interest and traction. Other things, of course, got interest and traction. I'm sure you know about Bit Gold and other things like that went around.

57
00:24:53.000 --> 00:25:32.000
<v Perry E. Metzger>And, you know, one day, of course, you know, there was, you know, whoever Satoshi Nakamoto may have been, or is, you know, posted that stuff to the mailing list that I was running. And that was an interesting conversation. I will note that at the time, there wasn't any particular reason to believe that that was going to be a more successful proposal than the previous 30 or 50. You know, it just so happened that it had the right set of characteristics that made it take off, which I would not have expected at the time. I can argue it was also

58
00:25:32.000 --> 00:25:58.000
<v Vlad Costea>The right timing for it, because it was in the middle of the financial crisis. It also had this political message that was inscribed in the Genesis block with the headline from the Times, which was Chancellor on the brink of a second bailout for banks. So you can argue that it sort of set the mission or purpose of this project as much more than let's put money on the Internet, but let's fight central banking.

59
00:25:58.000 --> 00:26:17.000
<v Perry E. Metzger>But that was a topic of conversation for 15 years or longer before that, right? That was not a new topic of conversation among the crowd of people among whom this stuff came up, right? That was a continuous topic of conversation.

60
00:26:17.000 --> 00:26:27.000
<v Vlad Costea>Sorry for interrupting. So if you think about eCash by David Chaum, it was just a way to put US dollars on Internet banks with some privacy on top.

61
00:26:27.000 --> 00:27:12.000
<v Perry E. Metzger>So there was no particular need underneath that for that to be only, you know, US dollars. It could have been gold deposits or bushels of wheat or anything. And if you go back to the discussions people were having, topics like the denationalization of money were on lots of people's minds under those circumstances. The thing that no one believed was possible at the time, and I would have believed it was not possible at the time. I was wrong, of course, but the thing that no one believed was possible at the time was to construct a decentralized fiat currency that was decoupled from any existing fiat currency.

62
00:27:12.000 --> 00:28:01.000
<v Perry E. Metzger>That was a thing that I was completely wrong on. I didn't believe that that was a thing that you could do. And in fact, of course, to some extent what we have right now in Bitcoin or Ethereum or what have you is an imperfect version of those things. Because the price of Bitcoin and Ethereum, you know, float and gyrate very aggressively because they are not used day to day as the unit of account by lots and lots of contracts. I mean, one of the things that keeps, say, the US dollar or the euro or what have you stable is that there are a vast network of contracts in which payment is promised in that given currency.

63
00:28:01.000 --> 00:28:46.000
<v Perry E. Metzger>And there's a gigantic network of obligations that are associated with those currencies. And one of the fundamental things that stabilizes them is, you know, I mean, there is an extent to which the existence of central banks stabilizes them a bit. But I mean, you know, central banks have the habit of printing lots of money at various times and inflating their currencies. But what generally stabilizes these currencies is the huge network of obligations that are denominated. And so long as Bitcoin, Ethereum and the rest of these things do not have such a large network of contracts associated with them, the price will fluctuate pretty wildly because, you know, they're often used as a speculative medium.

64
00:28:46.000 --> 00:29:37.000
<v Perry E. Metzger>But never mind that. At the time, you know, if you're talking about like around 2008, up until that point, I had not believed that it was possible to have a distributed fiat currency. You know, many of the thoughts that people had about various sorts of, you know, say, denationalized currencies or what have you involve people thinking about commodities on deposit with financial institutions, you know, with Chaum-style, you know, tokens issued against them or things like that. And there had been, of course, you know, previous thoughts that people had had, you know, like Hashcash and things like that were discussion points, but they were all fairly imperfect.

65
00:29:37.000 --> 00:30:19.000
<v Perry E. Metzger>The big contribution that, you know, that Bitcoin brought, I think was, coming up with a solution to, you know, a first solution to a distributed Byzantine agreement problem for, you know, for a ledger, which people had not thought about before and at that point I have to confess, I didn't understand, you know, when I read the paper, the significance. I still happen to think, by the way, that proof of work is a bad way to run things because of the economic consequences it has for a cryptocurrency.

66
00:30:19.000 --> 00:30:57.000
<v Perry E. Metzger>It means that it's difficult to reduce transaction costs with time. One of the interesting features you'll note, if you look at the history of currencies over the last few hundred years, the history of monetary systems and financial institutions, is the tremendous plunge in transaction costs that have occurred over time, right? You know, in the mid 19th century, if you wanted to pay someone in a foreign country a large amount of money for a lot of commodity, there was a tremendous amount of expense and machinery involved around it.

67
00:30:57.000 --> 00:31:35.000
<v Perry E. Metzger>And these days, this is an extremely routine thing. You know, if you want to buy, you know, hundreds of tons of steel from some foreign producer, you know, you barely think about it, right? You don't have this incredible, crazy network of financial institution contracts that are necessary in order to keep even the most basic portions of those going, which used to be very, very routine. One of the things that has happened over time is that, you know, we've gotten used to the idea that transactions are extremely cheap to engage in and extremely safe to engage in.

68
00:31:35.000 --> 00:32:20.000
<v Perry E. Metzger>Even when I was a kid, I'm on the older side. I remember a few stores in, you know, where I grew up still had house accounts, right? That if you wanted to buy something on credit, they had a ledger with the amount that various customers owed, and they dealt with, you know, with collections and things like that. This was already on the way out because of the invention of the credit card. And people don't understand, like, why credit cards were such a revolution. Credit cards were such a revolution because you could pay an intermediary to get rid of the risk associated with transactions, with credit transactions with strangers, which was a really, really big deal, right?

69
00:32:20.000 --> 00:32:39.000
<v Perry E. Metzger>You know, I mean, at one time, every, you know, every department store, every corner store, what have you, was dealing with credit risk on a daily basis when what they wanted to be doing was selling you know, like food or pants or something, right?

70
00:32:39.000 --> 00:33:31.000
<v Perry E. Metzger>Transaction costs have plunged over the centuries, like, you know, like, down, down. One of the problems with proof of work systems that has always bothered me is the fact that they add a certain amount of inherent cost to transactions that cannot be removed from the system. Proof of stake systems do not have that problem, by the way, which is one of the reasons I find them somewhat more exciting. But never mind that. You know, the thing that was really, really interesting and kind of incredible about Bitcoin at the time, which as I said, I did not appreciate, was it was a first solution to how you could have a distributed ledger that was maintained by some sort of Byzantine agreement without any sort of centralized trust.

71
00:33:31.000 --> 00:34:06.000
<v Perry E. Metzger>And the other thing that I did not understand at all at the time, you know, I was a complete idiot about, was that I did not understand that it was possible even to have a private fiat currency. I had assumed that those required, you know, that, you know, that you have a government and a central bank, you know, and a monopoly on, you know, a Weberian monopoly on the legitimized initiation of force and all of that in order to maintain. I did not understand that private fiat currencies were a feasible thing at all.

72
00:34:06.000 --> 00:34:47.000
<v Perry E. Metzger>You know, in retrospect, it should have been obvious, but so many things in the world are obvious in retrospect, right? But I don't think that 2008, that it was ripe so much because of the financial crisis. I think one of the big things that had happened at that point was that the Internet had become pervasive, and especially that mobile Internet was pervasive, that there are a lot of, were at that point already a lot of countries where you could expect that people were carrying around a mobile computer with a lot of power in it capable of running the protocols involved.

73
00:34:47.000 --> 00:35:28.000
<v Perry E. Metzger>I mean, people forget that, you know, smartphones are a very recent development. You know, if you're, you know, if you're under, say, 20, you barely remember that there was such a thing as a time before smartphones, you probably don't even. And even if you're, you know, if you're pushing retirement like me, you still have a bit of difficulty remembering how things used to be not that long ago. And when I started, you know, early in my career, there was a point where I worked for a large financial institution, Lehman Brothers, and I wore a pager.

74
00:35:28.000 --> 00:36:08.000
<v Perry E. Metzger>And if something went wrong, I had to run and find a payphone, right, because people didn't even have cell phones at the time. You know, I think the single biggest, you know, thing that made the time ripe was the existence of mobile phones. But I also don't think that the technologies that people had proposed before Bitcoin were particularly good, right? It was the first thing that really gave you a distributed ledger that had no central trust and had reasonable guarantees on the Byzantine agreement.

75
00:36:08.000 --> 00:36:09.000
<v Vlad Costea>Yeah, it's been- Well, that

76
00:36:09.000 --> 00:36:13.000
<v Perry E. Metzger>Was probably longer than you wanted as an answer, wasn't it?

77
00:36:13.000 --> 00:36:21.000
<v Vlad Costea>It was a very good answer. I'm sure that my editor will be able to clip some parts and make something interesting out of it.

78
00:36:21.000 --> 00:36:27.000
<v Perry E. Metzger>Well, I'll try to make sure that I don't make it impossible to clip out interesting segments from the rest of this.

79
00:36:27.000 --> 00:36:37.000
<v Vlad Costea>I mean, it's your challenge. It's up to you how you're going to deal with it. I got to present some sponsors at this point after 30 minutes, but one of them

80
00:36:37.000 --> 00:36:45.000
<v Perry E. Metzger>Sorry, I delayed them for like 10 minutes. No, it's fine. Don't worry. Go and describe your sponsors, yes.

81
00:36:45.000 --> 00:37:25.000
<v Vlad Costea>So I will start with SideShift, which is this place where you can swap your stablecoins for something that cannot be censored or confiscated, which is Bitcoin or Ethereum or whatever it is that you prefer. They have pretty good liquidity and also low fees, but on the other hand, they're not available in the US because of regulations, which are no different from the chat control that we have for some of the same considerations. So I got to give you that back. Like, you have some issues in the US, which we don't have in Europe, at least not yet, but the URL for that one is SideShift.ai, SideShift.

82
00:37:25.000 --> 00:38:02.000
<v Vlad Costea>There's also Braiins Hashpower, which is allowing people to rent proof-of-work hashrate to mine Bitcoin. And next month, there's going to be two more forks of Bitcoin that are coming out. One of them is going to be called, I guess, BIP-110 Knots or something, and the other one is going to be called eCash, a throwback to David Chaum's invention. And if you want to support these networks, you're going to be able to rent hashrate. So check them out. It's hashpower.braiins.com. Also, LayerTwo Labs, they're building the drivechain.

83
00:38:02.000 --> 00:38:33.000
<v Vlad Costea>I think it sort of contradicts what you said about proof-of-work because the drivechain works in a way where you have merged mined sidechains for everything. So let's say that you have a base layer with Bitcoin, and on top of that, you're going to have sidechains for Ethereum, for Zcash, for Namecoin, for pretty much everything that's interesting in the crypto space. The miners are mining all of them at the same time, and you have sort of unlimited scalability because there's always a chain that you can flip or run on top of the stack.

84
00:38:33.000 --> 00:39:10.000
<v Vlad Costea>And for the miners, the cost is just maybe supporting it with a node, which means storing every transaction that's happening inside of the system. And yeah, they're also launching something next month, which is called the eCash fork. Check it out. It's ecash.com. They actually bought the domain. And last but not least, I got to mention Cake Wallet. They're going to give you Mr. Metzger $50 in Zcash if you install their wallet. It's sort of a reward or a bounty. I'm not sure what you call it.

85
00:39:10.000 --> 00:39:13.000
<v Perry E. Metzger>I get paid to be here. I had no idea.

86
00:39:13.000 --> 00:39:28.000
<v Vlad Costea>Sort of, yeah. I mean, I'm pretty sure that your threshold for getting paid is much higher, but... A little bit. Yeah, I know what Mr. Zimmermann charges to speak at conferences. I'm not sure what you charge to speak on podcasts. But anyway...

87
00:39:28.000 --> 00:39:32.000
<v Perry E. Metzger>I don't charge to speak on podcasts. I do it for fun. But go on.

88
00:39:32.000 --> 00:40:05.000
<v Vlad Costea>Last but not least, Orange Rock, which is the place where you can trade without limits, without KYC, without anything. Right now, it's available on iOS as an app. It uses this DeFi integration where you can swap between coins so you don't deal with centralized exchanges. You don't have to KYC AML. It's pretty cool. But for now, it's only available on iOS. I think there is a beta version on Android and they're also going to launch something on desktop. So it's orangerock.com. I'm done with the ads.

89
00:40:05.000 --> 00:40:17.000
<v Vlad Costea>Was there a question that stems from this? Like, yeah, what do you think about this design and proof of work where you have sidechains for everything and there's no real bottleneck or Limitation?

90
00:40:17.000 --> 00:40:56.000
<v Perry E. Metzger>So I think that we still so I have no doubts that in the long run, this technology stack is going to be able to do pretty much anything that one wants to do. But one does have to keep in mind the number, the sheer number of transactions that occur in the normal financial networks on a, you know, on a daily basis. I mean, there are points during the Christmas rush where I can't remember how many millions of transactions an hour the Visa network handles. I used to know.

91
00:40:56.000 --> 00:41:32.000
<v Perry E. Metzger>But it's a truly crazy number. And if you want a decentralized financial network to be able to compete with things like that, it has to be able to handle those levels of load and it has to be able to compete on price at those levels of transaction costs. The thing is that I mean, by the way, the last couple of years, I have been paying relatively little attention to the latest research in the cryptocurrency world, mostly because I've been obsessed with artificial intelligence.

92
00:41:32.000 --> 00:42:07.000
<v Perry E. Metzger>But the last I looked, we were still significantly off of the sorts of loads that you would need to be able to handle in order to be able to do that sort of stuff effectively. And I remember the very early days of Bitcoin, right? There were, you know, I could go out in New York City and I could buy a sandwich with it at lunchtime, right? There were places that were actually just taking Bitcoin on chain, you know, to purchase things. And now that doesn't happen very much.

93
00:42:07.000 --> 00:43:04.000
<v Perry E. Metzger>You know, people tend to reserve on-chain transactions for relatively large settlements. You know, longer run, I mean, probably some combination of various sorts of sidechain technologies and, you know, and various sorts of improvements to blockchains themselves may turn out to be an answer for how you handle decentralized financial systems at extremely high workloads. I think it's kind of, it's incredibly cool that Bitcoin and other blockchains over the course of, you know, over the course of the last, you know, it's now almost a couple of decades, isn't it, have pushed people to work very hard on the technology.

94
00:43:04.000 --> 00:43:51.000
<v Perry E. Metzger>You know, you mentioned Zcash earlier, you know, a bunch of the fundamental technologies behind Zcash, such as non-interactive zero knowledge proofs, you know, would have seemed like magic to people 30 years ago. And they mostly only exist because of the amount of dedicated research that people have done into cryptography and cryptographic protocols. So, I mean, I think that in the long run, the future is very bright for all of this stuff. I always have skepticism about particular individual current technologies, but, you know, I'm a technologist and, you know, with a background in the financial services industry.

95
00:43:51.000 --> 00:44:43.000
<v Perry E. Metzger>So I tend to think in those sorts of terms, you know, there are also all sorts of, you know, weird issues with governance, right? You know, over the long term, for example, over the not very distant future, it may become necessary. I'm taking a step back. I'm very skeptical about how quickly quantum computers are going to come online and how quickly anyone actually needs to worry about people breaking, you know, RSA or elliptic curve systems or what have you with quantum computers. The current world record for actually cleanly factoring a number using a quantum computer is the number 21, which is not a very large number.

96
00:44:43.000 --> 00:45:40.000
<v Perry E. Metzger>You know, the average school child can tell you that it's three times seven, right? It's not very difficult to factor, you know. And in order to be able to break cryptographic systems, we need to be able to factor, say, you know, instead of, you know, instead of five bit numbers, we need to be able to factor things that are more like 4000 bits, right? Nonetheless, there's a very big push that a lot of people have towards post quantum algorithms. And one wonders about the resilience of the governance systems of certain cryptocurrencies in the face of a need to move to post quantum algorithms or to radically different consensus algorithms or all sorts of technological improvements.

97
00:45:40.000 --> 00:45:53.000
<v Perry E. Metzger>You know, I was very interested for quite a while and in fact did some work on Tezos, which I found kind of interesting only because of its governance properties.

98
00:45:53.000 --> 00:46:00.000
<v Vlad Costea>I do recall that from the Naomi Brockwell interview that you mentioned Tezos and that was going to sort of poke fun because it never went anywhere.

99
00:46:00.000 --> 00:46:50.000
<v Perry E. Metzger>It has failed to thrive. That is, I think, the kind way to put it. It has failed to thrive, but it had some very good ideas in it. And one of them was having an on-chain voting mechanism so that the holders of the currency could make decisions on technical changes to the infrastructure. Right now, if there was a need to move Bitcoin from the current signature scheme to another signature scheme, how would that happen? You know, mostly with bloodshed and screaming and arguments and people shaking their fists at each other, not with an obvious governance mechanism that would make it relatively straightforward to come up with a, you know, to agree on some sort of technical change.

100
00:46:50.000 --> 00:47:48.000
<v Perry E. Metzger>This is a problem, right? I remember just like SegWit was a very minor sort of technical change, as I remember, years ago, and that created a storm of controversy and blood and what have you, which it should not have. You know, so I think things that are going to be interesting, you know, in the future of these systems are going to be increased transaction rates, better governance systems, ways of dealing better with the need to make technical changes when technical changes become important, et cetera. But yeah, I mean, the 2008, as I said, I think that things were, that the combination of things was that Bitcoin was the first thing that got proposed that had a reasonable technical basis.

101
00:47:48.000 --> 00:48:27.000
<v Perry E. Metzger>The earlier proposals were just not technically feasible. And it showed up at a point where people had computers in their pockets finally. And having a computer in your pocket makes things very different because you could not walk up to someone else and conduct a transaction with them before that. I don't think it actually had much to do with the financial crisis. As I said, you cite the politics in the thing that's in the first block, but I think that all of that political thinking and discussion was in that community for decades before.

102
00:48:27.000 --> 00:48:48.000
<v Perry E. Metzger>People wanted, you know, various forms of non-central bank based currencies, you know, non-governmental currencies, et cetera. And there was a thirst for it long before 2008. Anyway, I've talked at you too long. You should probably ask questions.

103
00:48:48.000 --> 00:49:23.000
<v Vlad Costea>I was looking at, so for the people listening right now, there are like 200 people who tuned in. I do want to tell you that we're going to get like straight to the point for the stuff that you came to hear about the 31st of October, 2008 Halloween night when Satoshi announced this. I did find this comment from Tim May who said back in 2018 that, frankly, the newness in the Satoshi white paper and then the early uses for things like Silk Road is what drew many people to Bitcoin.

104
00:49:23.000 --> 00:49:43.000
<v Vlad Costea>If the project had been about regulatory compliant banking friendly thing, then interest would have been small. In fact, there were some yawn inducing electronic transfer projects going back a long time. The fact that SET, for Secure Electronic Transfer was one such mind numbingly boring projects.

105
00:49:43.000 --> 00:50:45.000
<v Perry E. Metzger>That was a very different sort of project. It didn't have anything to do with stuff like this. That was mostly a very, very terrible. Idea on how to do secure credit card transactions online. And it turned out that doing credit card transactions over TLS, meaning, you know, HTTPS, you know, SSL, there are many names for it, was perfectly adequate. SET was garbage. But, you know, Tim was right. That the reason that people were interested and found it and found it exciting was, again, that it, you know, that it had something to do with this thirst for currencies that, you know, that met censorship resistance, you know, that were censorship resistant, and that didn't involve central banks.

106
00:50:45.000 --> 00:51:24.000
<v Perry E. Metzger>And that's something that people really were very, very interested in for a long time. And, you know, he puts it more colorfully in the quote that you had there. But people were very, very interested in these things for a long time. I think that, and had many proposals. The reason, as I said, that Bitcoin became the one that took off was, I think, not just that it was at the right time, but that it was the first proposal that actually had strong technical merits, which, by the way, I will again say I did not recognize at the time because it felt like just another one of many, many proposals that had been made at the time.

107
00:51:24.000 --> 00:51:27.000
<v Perry E. Metzger>I was thoroughly incorrect about that.

108
00:51:27.000 --> 00:52:04.000
<v Vlad Costea>Yes. And this is also something that should be noted, that once Satoshi came to the mailing list and he said, hi, I've been working on this peer-to-peer electronic cash system and whatever electronic money. He was talking, first of all, in the context of eCash, because this was kind of the holy grail of trying to come up with decentralized eCash. And he also got a lot of negative reactions, like the first person to reply to Satoshi. Like a few hours later, actually, after he posted the white paper, it was this guy named James A.

109
00:52:04.000 --> 00:52:14.000
<v Vlad Costea>Donald, who I think still does a blog. So he's still active on the Internet and does stuff. And he said, this sounds good, but it doesn't scale. And I think this—

110
00:52:14.000 --> 00:53:15.000
<v Perry E. Metzger>Well, he was correct. It also turned out not to matter, or at least not to matter in context. The scaling issues are real. Many of the criticisms that people had in the first few weeks were completely correct and also turned out not to be relevant. The number of transactions that Bitcoin can do per second is not particularly high. And yet it has survived perfectly well. This is, by the way, I myself criticized the proposal early on for reasons that I think were technically correct and were, in practice, not important. The other thing to remember, I'm going to emphasize for like the 15th time, this was not the first eCash proposal or even distributed eCash proposal that had gone through in the months and years before this.

111
00:53:15.000 --> 00:54:00.000
<v Perry E. Metzger>People had been very, very interested in this in the cypherpunk context and in the context of my mailing list for a very long time. It was a holy grail that people were, you know, sought out, and that most people didn't, that some people did not think was possible in this particular form. Kudos to whomever Satoshi might actually be. Solving this particular set of technical problems was a significant achievement and one that we did not appreciate enough in October, November 2008.

112
00:54:00.000 --> 00:54:05.000
<v Vlad Costea>There was another reply from Ray Dillinger or is it Dillinger?

113
00:54:05.000 --> 00:54:21.000
<v Perry E. Metzger>I don't know. You're mentioning names that I recognize very well, but many of these are people who I've never encountered outside of mailing lists and online for various kinds, including James Donald, including Ray Dillinger, etc.

114
00:54:21.000 --> 00:54:40.000
<v Vlad Costea>Ray Dillinger basically said that there is no demand for bitcoins. He was trying to explain that there is demand for fiat currencies and there are very well established currencies with network effects. And if you're going to introduce something new into the world, it can take decades for it to catch on. And he was right about this also.

115
00:54:40.000 --> 00:55:15.000
<v Perry E. Metzger>He was right and it did not matter. There are a lot of, as I said, many of the criticisms in the days and weeks after were completely correct and also in retrospect irrelevant. But what did we know at the time? If one had perfect capacity to predict the future, the world would be a very different place. I would be a much richer man if I had the capacity to predict the future perfectly. I would have been mining a lot of Bitcoin early on, for example.

116
00:55:15.000 --> 00:55:42.000
<v Vlad Costea>Yeah, I think Ray actually supported the Bitcoin network later with some feedback because he wrote a blog post like 10 years later in which he explained that he sort of supported Satoshi and fixing the code alongside Hal Finney and that he saw this relationship as someone who's older and more experienced dealing with a graduate student.

117
00:55:42.000 --> 00:55:45.000
<v Perry E. Metzger>I don't know about that. I've never read that blog post.

118
00:55:45.000 --> 00:56:01.000
<v Vlad Costea>I can send it to you later. And there was also this guy. No, wait. Before that, Ray Dillinger also said that proof of work computation has no intrinsic value, which, of course, is correct. There are people who assume that Bitcoin is valuable just because it costs this much amount of money.

119
00:56:01.000 --> 00:56:56.000
<v Perry E. Metzger>No, the purpose of the puzzle solving is to support the security of the Byzantine agreement. It has nothing to do with giving the network value. The value is completely detached from the mechanism to protect the Byzantine agreement. The purpose of solving these hash puzzles is to protect the Byzantine agreement. Yeah, but he was correct, right? And there were a lot of perceptions from things like the Bit Gold and Hashcash and other proposals. There are a lot that I don't even remember the names of where people were thinking in terms, incorrectly, of the value of the thing having something to do with various kinds of computation that went in.

120
00:56:56.000 --> 00:57:00.000
<v Perry E. Metzger>But, you know, that's, of course, irrelevant. Anyway, go on.

121
00:57:00.000 --> 00:57:21.000
<v Vlad Costea>Yeah. So just for the record, I'm only reading these comments because they're useful in a historical context. I'm sure that there's going to be some AI scrounging this conversation and looking for data. And it's useful that there is some input from you, because just reading them without someone who has some context has no value, I think.

122
00:57:21.000 --> 00:57:50.000
<v Perry E. Metzger>Maybe. You know, the way that any of us got context at the time was participating, you know, in group discussions over a course of years. There were a lot of discussions over a lot of time. The CypherPunks mailing list and then the Cryptography mailing list had long histories. This happened many, many years into the existence of the thing. But anyway.

123
00:57:50.000 --> 00:58:10.000
<v Vlad Costea>There are two more people that I want to mention, actually three. But I'll start with this guy whose name is John R. Levine. Yes. He argued that there are more bad actors who can participate with hash power and sort of Sybil attack the network.

124
00:58:10.000 --> 00:58:46.000
<v Perry E. Metzger>That would have been an issue very early on. At this point, of course, it's less of an issue, although there are times or have been times when far too much of the of the computational power in the Bitcoin network has been in the hands of single of small groups. But, you know, again, the generally speaking, you know, that's that's a technical comment, right? You have the issue of whether the Byzantine agreement can be maintained. And so he was John Levine is a person who I had almost forgotten was on the Cryptography mailing list.

125
00:58:46.000 --> 00:58:56.000
<v Perry E. Metzger>He was I know him better as being a guy who ran a mailing list for authors of compilers for decades. But never mind that. Go on.

126
00:58:56.000 --> 00:59:05.000
<v Vlad Costea>This is why you're here. You're giving everything more context. And there was another guy from Sun Microsystems named Nicholas Williams.

127
00:59:05.000 --> 00:59:10.000
<v Perry E. Metzger>Oh, Nico Williams. Yes. I still I still talk to Nico occasionally.

128
00:59:10.000 --> 00:59:23.000
<v Vlad Costea>He had possibly the worst of ideas that there should be some sort of identification for the network participants so that when someone tries to cheat the system and double spend, you can know who they are and sort of Well, that's a that

129
00:59:23.000 --> 01:00:06.000
<v Perry E. Metzger>Is a feature from certain sorts of earlier cryptographic currency proposals that was used over and over again. There are a bunch of early cryptocurrency proposals where your anonymity was protected if and only if you did not double spend that if you attempted to double spend, then information about your identity gets revealed, which is a really neat property and not one that you would expect was possible to achieve. The idea in systems like that is that if you double spend, you have to reveal enough information that the person who's double spending is now identifiable.

130
01:00:06.000 --> 01:00:15.000
<v Perry E. Metzger>And if you don't double spend, then you're completely anonymous, which is actually kind of a cool primitive to have in certain systems.

131
01:00:15.000 --> 01:00:40.000
<v Vlad Costea>Yes. And there's also this mention that I want to make that in his blog post, Ray Dillinger said that he performed the security audit and Hal Finney looked at the scripting language while both Ray and Hal analyzed the accounting code. I just wanted this to be part of the episode for some reason. You can find this information. So I put together a comic book.

132
01:00:40.000 --> 01:00:46.000
<v Perry E. Metzger>I have to confess, I've never really read through most of the Bitcoin code because I hate C++.

133
01:00:46.000 --> 01:00:48.000
<v Vlad Costea>But what do you love?

134
01:00:48.000 --> 01:01:33.000
<v Perry E. Metzger>What do I like? These days for systems programming, I tend to use Rust. If I'm working on compiler projects and things like that, I tend to use things like OCaml. But, you know, now everything has changed because, you know, in the last eight months or so, AI coding has taken over. And at the moment, I'm actually constructing my own programming language, which is not a particularly great programming language, but exists for one reason and one reason only, which is to be able to formally verify programs that are written in it very easily. That's a different story. But you look at me with puzzlement suddenly.

135
01:01:33.000 --> 01:01:48.000
<v Vlad Costea>Now, I was reading the chat and people. So I've been doing this contest before where I was asking the guest to write down the number from one to one hundred and the chat has to guess it. And then the winner gets an amount of Zcash. And I was thinking, should I do this again?

136
01:01:48.000 --> 01:01:53.000
<v Perry E. Metzger>Feels like I can find a pen, I'm sure somewhere.

137
01:01:53.000 --> 01:02:16.000
<v Vlad Costea>Just don't reveal the number and give people like 30 minutes to guess or something. I feel like if we start this in the middle of the episode, it actually rewards people who stuck around as opposed to just telling people they have to tune in for this, because usually what happens is that once you reveal the number and you know the winner.

138
01:02:16.000 --> 01:02:26.000
<v Perry E. Metzger>Well, I won't know the winner, but hold on a second. Here, I will. You want me to write down a number between one and one hundred?

139
01:02:26.000 --> 01:02:31.000
<v Vlad Costea>Yes. Don't reveal it. Don't show it to the camera. Don't say it out loud.

140
01:02:31.000 --> 01:02:49.000
<v Perry E. Metzger>I will not say it out loud. I will put it on a piece of paper and then I will I will say what the number is later. OK. And I my pen is half dead. So they so the thing that I wrote down looks awful, but I have it down on a piece of paper.

141
01:02:49.000 --> 01:02:58.000
<v Vlad Costea>That's good. So I think the reward is like point one Zcash. That's like fifty five dollars at today's rate. So you have to guess the number from one to—

142
01:02:58.000 --> 01:03:00.000
<v Perry E. Metzger>his favorite currency. Yes.

143
01:03:00.000 --> 01:03:23.000
<v Vlad Costea>Yeah. Yeah. So people have already started to type numbers. Others are just being haters for some reason, which I don't understand. I love haters. But anyway, for all the wrong reasons, like. Well, anyway, move on. They're saying that Satoshi is most sad and whatever, and we're cooked and we fell for a scam and whatever.

144
01:03:23.000 --> 01:04:01.000
<v Perry E. Metzger>Well, anyway, so early on. Oh, you were asking about programming languages. So, yes, I mean, as I said, these days, what I'm interested in mostly is not particular programming languages, but the possibility finally now that we have the ability now that software has become much less expensive and programmer time has become much less expensive. Finally, perhaps being able to solve a lot of fundamental security problems by formally verifying software, which means mathematically proving that the software is correct. This has been done to some extent already in various contexts.

145
01:04:01.000 --> 01:04:49.000
<v Perry E. Metzger>There is a surprising amount of formally verified software out there. You know, if you're using a Web browser these days, you're probably using a formally verified cryptographic stack or often are using one. But, you know, there's a formally verified C compiler out there and some very formally verified micro kernels. And there's a formally verified ML compiler and what have you. But all of these are very difficult and complicated and nasty things to do by hand using human labor. And I'm rather interested in the possibility of doing a lot of this stuff using AI rather than using human labor, because it may help us climb out of the security hole that we have been in since around 1988.

146
01:04:49.000 --> 01:05:13.000
<v Perry E. Metzger>I can, you know, since the release of the Morris worm till today, we've been in sort of a continuous security crisis. But never mind that unless people are interested in my talking about it more. Are there questions from the people listening? You know, you say that there are that there is a surprising number of them. So it's not surprising. Many of them typing in, you know, many of them typing in numbers between one and 100, apparently.

147
01:05:13.000 --> 01:05:23.000
<v Vlad Costea>Yeah, I guess that's the game right now. But I want to get back one step. Do you remember what you said, like your first reply to Satoshi about Bitcoin?

148
01:05:23.000 --> 01:05:30.000
<v Perry E. Metzger>I don't. I do remember that I shut down the conversation very early because it started spinning.

149
01:05:30.000 --> 01:06:04.000
<v Vlad Costea>So this was on November 7th of 2008, 12:32 EST. So you said a bunch of people seem anxious to branch the discussion of cryptographic cash protocols off into a discussion of the politics of money. I'm a rabid libertarian myself, but this isn't the rabid libertarian mailing list. Please stick to discussing either the protocols themselves or their direct practicality and not the perils of fiat money, taxation, your aunt Mildred's gold coin collection, et cetera. Signed— yes.

150
01:06:04.000 --> 01:06:44.000
<v Perry E. Metzger>Yeah, that sounds like me. So one of the things that I did as the moderator of the mailing list. So the cryptography list, unlike the cypherpunks list, is moderated. All of the postings that go through are approved specifically. And one of the reasons that was done was in order to keep things on topic. And the topic on the cryptography list was not the politics of any of these things, but the technology. And people often liked moving a little bit too much on to discussing politics. And I was usually pretty heavy handed about stopping that.

151
01:06:44.000 --> 01:06:53.000
<v Perry E. Metzger>So I'm not, you know, I'm not surprised that I did that early on. That was the sort of thing that I did in a lot of discussions.

152
01:06:53.000 --> 01:07:14.000
<v Vlad Costea>But in response to that, Zooko jumped in and said, hey, you can come to money politics, peer to peer hackers mailing list, whatever. So he sort of advertised his own list. But this also made him pay attention to Bitcoin, because I think a couple of months later, he became the first person to write a blog post about the launch of Bitcoin.

153
01:07:14.000 --> 01:07:27.000
<v Perry E. Metzger>Might have been. I've known him for many, many years. He's an interesting fellow, but I did not remember that he was the first person to blog about Bitcoin. Could very well be the case. I don't know.

154
01:07:27.000 --> 01:07:30.000
<v Vlad Costea>I think he sort of takes credit for it.

155
01:07:30.000 --> 01:07:31.000
<v Perry E. Metzger>Could be.

156
01:07:31.000 --> 01:07:41.000
<v Vlad Costea>But I do believe that he was because Satoshi listed his blog on Bitcoin.org for a while, like an acknowledgement that there's something going on.

157
01:07:41.000 --> 01:08:03.000
<v Perry E. Metzger>Again, entirely possible. I don't remember, you know, remembering things from from 15 weeks ago is pretty hard for me. You know, from almost 20 years, it gets to be it gets to be harder. I know a lot about the high level context. If you ask me why something went out on a particular date, I won't remember it, obviously.

158
01:08:03.000 --> 01:08:12.000
<v Vlad Costea>Oh, no. But this turned out to be, I guess, such an important event, even though for you, maybe it was just a Tuesday, just another day.

159
01:08:12.000 --> 01:08:22.000
<v Perry E. Metzger>Well, well, Whenever everything important happens on a day that looks like any other day. Right. That's always the case.

160
01:08:22.000 --> 01:08:28.000
<v Vlad Costea>Do you keep logs on the mailing list? Like, are you able to find IP addresses or whatever of the users?

161
01:08:28.000 --> 01:09:20.000
<v Perry E. Metzger>Nope. That was deliberately there's a policy that those are not kept. I have no information on any of that stuff. And I've had more than one reporter, you know, try harassing me for information on stuff like that, only for me to explain that I have no such information. And there was a policy of making sure that all such information was automatically and deliberately disposed of. So I would not be in a position to give it out to anyone. No such information exists. There was a there have been a succession of people who have tried writing books or investigating who Satoshi is, which I think is a stupid thing to be asking in which I never wanted to cooperate with because clearly whoever Satoshi is, Satoshi wanted to be, you know, to maintain their privacy.

162
01:09:20.000 --> 01:09:56.000
<v Perry E. Metzger>And I don't see any reason that I would cooperate with violating people's privacy. But, you know, there was one particular reporter I remember who was very obvious why they wanted to talk to me. And I deliberately tortured them by discussing completely irrelevant and uninteresting topics for several hours before dropping the hint that I had deleted all of the logs and that all of the logs were automatically deleted, in fact, at which point they excused themselves very quickly. I felt good about it. I had kept them from bothering anyone else for several hours.

163
01:09:56.000 --> 01:10:43.000
<v Perry E. Metzger>But anyway, yes, we do not keep logs of any such things. In fact, the only demand that I've had traditionally for subscribers to the cryptography list has been that they use a stable and unique pseudonym if they're going to participate. There have been a lot of people who have tried subscribing as Satoshi Nakamoto over the years, and I kick any such people off. But other than that, pretty much anything goes in terms of how one identifies oneself. I do prefer that people identify themselves with a stable pseudonym, but, you know, who they actually are is not my business.

164
01:10:43.000 --> 01:10:48.000
<v Vlad Costea>So how do you know when the real Satoshi comes back to the mailing list?

165
01:10:48.000 --> 01:11:26.000
<v Perry E. Metzger>If the real Satoshi were to ever come back to the mailing list, he would presumably be in possession of a number of interesting public keys with which he could identify himself. And, you know, if Craig Wright had been in possession of any of those public keys, no one would have questioned any of his claims, but of course he was not in possession of any of those public keys. You know, if and when, I doubt it will happen, but if and when someone appears and they are in possession of one of Satoshi Nakamoto's public keys, then we can discuss whether they are actually Satoshi Nakamoto. Otherwise, I don't want to hear from them or about it.

166
01:11:26.000 --> 01:11:41.000
<v Vlad Costea>That's fair, and I respect your approach. For context, I know that you have been drawing or at least designing comic books to teach young people about computer science. I do not Draw them.

167
01:11:41.000 --> 01:12:37.000
<v Perry E. Metzger>I am incapable of drawing that well. There is, yes, there is a book that I wrote along, you know, along with a co-author and a wonderful illustrator. The illustrator's name is Jerel Dye, you know, and, you know, and the, you know, and we have Penelope Spector, and the book is a, yes, it is a book on computer architecture for middle schoolers, and it is in graphic novel format. And I encourage anyone listening to buy hundreds of copies each. They're wonderful for things like, you know, preventing erosion in your backyard or, you know, if you have a table with a short leg, you know, you can put the book underneath the leg and it will keep the table very flat.

168
01:12:37.000 --> 01:13:06.000
<v Perry E. Metzger>Or you can build a table with the book. You can build a table entirely out of copies of the book. Early on, I was under the mistaken impression that the publisher cared what people did with copies of the books, but it turns out that they only care if you buy it. They don't care if you read it at all. So, you know, buy as many copies as possible. You might want to give copies of the book to, you know, to impressionable, you know, children who are interested in computers. But I mean, that seems like a secondary effect.

169
01:13:06.000 --> 01:13:09.000
<v Vlad Costea>That's far out, you don't do that.

170
01:13:09.000 --> 01:13:49.000
<v Perry E. Metzger>Yes, I wrote a book on computer architecture for kids, although adults probably would get just as much out of it. It was deliberately written so that, you know, someone who's 40 who's interested in how computers actually work can read it and find it interesting to read. It's not only for children, although it's marketed to children. Yes, that's published by Macmillan. You know, it's been fun. That book took only about, I proposed writing that book, and I thought that it would take a couple of months to write it.

171
01:13:49.000 --> 01:14:15.000
<v Perry E. Metzger>And the illustrator and I and my co-author, I think, turned in the manuscript about 10 years late. So, you know, there is hope for almost any project, no matter how long you've been postponing the end of it. But yes, anything else, you know, anything else you wanted to ask about?

172
01:14:15.000 --> 01:14:18.000
<v Vlad Costea>Well, I mentioned the comic book.

173
01:14:18.000 --> 01:14:42.000
<v Perry E. Metzger>By the way, the book is called Science Comics: Computers: How Digital Hardware Works. And as I said, it's published by Macmillan. And you can Google for my name and buy several thousand copies of it. As I said, they're wonderful and durable. You know, you can make all sorts of things out of them. Yes, go on.

174
01:14:42.000 --> 01:14:55.000
<v Vlad Costea>I made my own comic book about the first year of Bitcoin. It's called Bitcoin Year One. It's published online. You can read it, you can print it, you can do whatever you want with it. So it's not published by Macmillan. It's sort of...

175
01:14:55.000 --> 01:15:02.000
<v Perry E. Metzger>But, you know, since it doesn't arrive in physical form, I can't make furniture out of it. I would have to print it out myself first.

176
01:15:02.000 --> 01:15:09.000
<v Vlad Costea>I mean, I can print one for you and send it to you just because you're in it. And I feel like that would be nice.

177
01:15:09.000 --> 01:15:19.000
<v Perry E. Metzger>I'm sorry to hear that. But, you know, I avoid public notoriety usually. But anyway, I'll download it.

178
01:15:19.000 --> 01:15:20.000
<v Vlad Costea>Except for podcasts, right?

179
01:15:20.000 --> 01:15:39.000
<v Perry E. Metzger>Except for podcasts. I do a lot of podcasts. I have been forced in recent years to have more of a public profile than I used to. Because I've been, you know, because I've been doing AI activism or advocacy or what have you. But never mind that.

180
01:15:39.000 --> 01:15:56.000
<v Vlad Costea>No, we're going to get to that. But before that, I'm going to put the link to that comic book in the description if you want to find it. So you can just go to bitcointakeover.com and scroll to the magazine section and you're going to find it for free download. I will have a look.

181
01:15:56.000 --> 01:15:58.000
<v Perry E. Metzger>I assume there's a PDF there I can look at.

182
01:15:58.000 --> 01:16:05.000
<v Vlad Costea>There is, yes. And it's free. Doesn't cost you anything. Just take a look. It's a very short reading.

183
01:16:05.000 --> 01:16:07.000
<v Perry E. Metzger>Not even your soul?

184
01:16:07.000 --> 01:16:15.000
<v Vlad Costea>I mean, I do like souls. I do collect them sometimes. Very good. Some sort of urn.

185
01:16:15.000 --> 01:16:20.000
<v Perry E. Metzger>They work well in an urn, I'm sure. But anyway, moving forward.

186
01:16:20.000 --> 01:16:26.000
<v Vlad Costea>I wanted to ask you about Dowdeswell. Because Metzdowd comes from...

187
01:16:26.000 --> 01:17:19.000
<v Perry E. Metzger>Metzger, Dowdeswell and Company. Exactly. That's just my consulting company. And Dowdeswell is no longer an active partner in it. That's Roland Dowdeswell, who's a very cool chap. He is still in the consulting business. But he worked for a number of years after we founded the company full-time for a few people. And so he left our business. But yeah, computer security and computer systems consulting. I've been supporting myself that way for decades. I think it's now been... I think it's been well over 30 years since I've worked for a company that I didn't start.

188
01:17:19.000 --> 01:17:37.000
<v Vlad Costea>That's cool. I've asked you about this because when I saw Metzdowd, I was thinking, okay, Metz comes from Perry Metzger. But Doud, does it come from Download or something? And then I came across the name of this company. And I was thinking, wait, who's this person?

189
01:17:37.000 --> 01:17:45.000
<v Perry E. Metzger>The second person, as I said, he's a former business partner of mine. Still a very good friend. I talk to him a couple of times a week.

190
01:17:45.000 --> 01:18:04.000
<v Vlad Costea>That's really cool. I also appreciate the fact that you're still maintaining or keeping this mailing list online. You don't have to go to the Internet Archive to find what you're looking for. It's still out there. You just go to metzdowd.com. But I think the URL is slightly different.

191
01:18:04.000 --> 01:18:15.000
<v Perry E. Metzger>No, I mean, if you go there are links to the mailing list archives and to the mailing list.

192
01:18:15.000 --> 01:18:24.000
<v Vlad Costea>So have you been keeping this up on purpose? Of course, it's on purpose.

193
01:18:24.000 --> 01:18:30.000
<v Perry E. Metzger>It's still an active mailing list. I mean, there are people who post to it. Yes.

194
01:18:30.000 --> 01:18:34.000
<v Vlad Costea>Have you seen anything interesting since Bitcoin that got posted there?

195
01:18:34.000 --> 01:19:11.000
<v Perry E. Metzger>Oh, many things. There's been lots and lots of discussion on lots of topics in the years since then. But, you know, I couldn't summarize it very well. And I think that if one is not particularly interested in cryptography or security issues, it might not be very interesting to you. But if you're interested in cryptography, it's a fine mailing list. Internet mailing lists, of course, have kind of fallen out of favor as a way for people to have discussions online. People these days have their Substack and they tweet and do all sorts of other things.

196
01:19:11.000 --> 01:19:33.000
<v Perry E. Metzger>But it's a continuously operating atavism that we've had since the 1990s. So why shut it down? What would the point be? And Tamzen Cannoy does a fine job continuing to moderate it when I don't, which is most of the time.

197
01:19:33.000 --> 01:19:39.000
<v Vlad Costea>That was going to be the follow up question if you still actively moderate the list.

198
01:19:39.000 --> 01:19:55.000
<v Perry E. Metzger>I used to moderate it very actively and I got other people to help with it because I no longer have as much time as I used to. But, you know, in principle, I still can do it. In practice, I don't very often.

199
01:19:55.000 --> 01:20:10.000
<v Vlad Costea>Well, it's really cool because you've been doing this for such a long time and you obviously care about maintaining a good archive and getting people to have conversations there. Yes.

200
01:20:10.000 --> 01:20:40.000
<v Perry E. Metzger>Yeah, I mean, as I said, there's no reason to shut it down so long as people still want to have the thing there. It's gotten a little bit more difficult in recent years because, you know, over time it's more and more trouble maintaining an email server. But, you know, there's no reason to shut it down. So I have not shut it down. I don't intend to. By the way, I'm curious about your shirt. Do you have a computer science background or?

201
01:20:40.000 --> 01:20:51.000
<v Vlad Costea>Well, I did do some computer science in high school, and then I decided that I'm not good enough for something. So I did political science, which is not a real science in university.

202
01:20:51.000 --> 01:21:11.000
<v Perry E. Metzger>Yes, political science, which is, you know, it is. Yes, it is not a science, but you're wearing a shirt with, you know, with source code on it. I imagine it is to something like the is it is it Bitcoin source code?

203
01:21:11.000 --> 01:21:14.000
<v Vlad Costea>No, it's the I love you virus.

204
01:21:14.000 --> 01:21:15.000
<v Perry E. Metzger>Very good.

205
01:21:15.000 --> 01:21:25.000
<v Vlad Costea>So if someone takes a picture of this and they have some sort of AI scanning, maybe that you're going to get the virus. But I think this only works on Windows. So you don't have to worry about it.

206
01:21:25.000 --> 01:21:54.000
<v Perry E. Metzger>Yes, and probably doesn't have to worry about it, even on Windows these days. But never mind that I remember the I love you virus very well. It was it caused a lot of trouble at one of my clients when it hit. But never mind that I spent a couple of days helping them set up various kinds of email filters in order to and cleaning out various things that had been hit by it. But never mind that.

207
01:21:54.000 --> 01:22:02.000
<v Vlad Costea>Well, I do want to mention that this is made by a French fashion house that's trying to do fashion for nerds. And right now— I'm sorry.

208
01:22:02.000 --> 01:22:10.000
<v Perry E. Metzger>A French fashion house that is doing fashion for nerds. Exactly. A somewhat self-defeating task. But I wish them luck.

209
01:22:10.000 --> 01:22:50.000
<v Vlad Costea>I know that they're shutting down, unfortunately, but I think they're going to have a final sale. I wasn't planning to say this, but since you mentioned the T-shirt, like you can go to it's called Maison Maachi. It's M-A-A-C-H-I. And Maison, I guess it's the French word for house. Yes. So check it out. I think that they're going to basically sell their whole stock at a discount in the coming weeks or something because they were not very successful so far. But I've been wearing them. They sent this to me and I've been wearing it for a long time because it feels like it fits, right? It's exactly what someone like myself should be wearing.

210
01:22:50.000 --> 01:22:54.000
<v Perry E. Metzger>The I love you virus. Yes. Yes. Yes. OK.

211
01:22:54.000 --> 01:23:05.000
<v Vlad Costea>And it starts a conversation when I go to a conference. They're like, oh, what are you doing there? There are some instructions and I think this is JavaScript.

212
01:23:05.000 --> 01:23:12.000
<v Perry E. Metzger>It might be I can't make it out well enough, but I would believe you that it's JavaScript.

213
01:23:12.000 --> 01:23:26.000
<v Vlad Costea>Anyway, there's also a T-shirt with the Bitcoin code if you want to buy that. Check it out. Anyway, let's get into AI, because lately I've also been somewhat of an amateur of local models. I've been running Qwen.

214
01:23:26.000 --> 01:23:37.000
<v Perry E. Metzger>Local, local models are wonderful things. It's unfortunate that it's rather expensive to run the best models locally, but I encourage people to try.

215
01:23:37.000 --> 01:24:22.000
<v Vlad Costea>You mean Kimi, right?
<v Perry E. Metzger>Well, Kimi K3 is not out yet, but even like GLM-5.2 is, you know, I think that running it properly, you can probably, if you're a very frugal and careful person, run it locally for like 50 or 70 thousand dollars worth of hardware. But if you really want to run it at a reasonable pace, you know, without horrible quantization, it's in the hundreds of thousands of dollars worth of hardware, which, by the way, by historic standards is not that much, right? I remember when my first real job was at Bell Communications Research and I had a Sun 3/60 on my desk there, you know, like 1987 or so.

216
01:24:22.000 --> 01:25:06.000
<v Perry E. Metzger>And I recently got curious and tried to find out how much would that machine cost in current dollars, you know, something like 90 thousand dollars, you know, so and that was a personal computer. That was just for me. So in some ways, you know, if you're certainly if you're a larger organization, given what the price of programmers is like, running AI locally is pretty economical, but I'm expecting that over time the price of running models locally will plunge. Not this year, but, you know, eventually as more and more hardware becomes available at lower and lower prices.

217
01:25:06.000 --> 01:25:18.000
<v Perry E. Metzger>This may not be something that makes companies like Anthropic particularly happy, but it's an inevitability in the long term. So what do you what do you use Qwen for?

218
01:25:18.000 --> 01:25:40.000
<v Vlad Costea>Well, right now, not for anything special, like I've created some video games, which once again, you can find on bitcointakeover.com. And I've also used it for analyzing text. So I do the transcripts of my interviews and then I ask it to extract the most important talking points to do time stamping.

219
01:25:40.000 --> 01:25:48.000
<v Perry E. Metzger>Do you do transcripts with a model like Whisper or do you use the or do you use some online service?

220
01:25:48.000 --> 01:26:00.000
<v Vlad Costea>So I do have Whisper installed. I'm not always happy with the output. Sometimes I take the transcript that gets generated by YouTube because that one tends to be a bit more accurate sometimes.

221
01:26:00.000 --> 01:26:20.000
<v Perry E. Metzger>Though sometimes it sometimes the YouTube ones have hilarious mistakes in them, which I always wonder about because, you know, Google supposedly has some of the best models in the world and they have they certainly have access to much better speech to text models, and they don't seem to use them for things like YouTube. And I do not understand why.

222
01:26:20.000 --> 01:27:02.000
<v Vlad Costea>I think it got better over the last few months. I did notice a difference. For example, when I was saying Ethereum, it was transcribing a theorem. That's amusing. But now it became much better. And sometimes I'm even surprised that it can. So from the text that I get from this transcription from YouTube, when I put that into Qwen and I ask it to basically summarize or generate transcripts based on that transcript, it even determines who said what. And sometimes I'm surprised. Like, how do you even know? Because you're not listening to the audio.

223
01:27:02.000 --> 01:28:00.000
<v Perry E. Metzger>Diarization. Actually, like models that can do proper diarization are still not as common as one would like them to be. That's what you want if you want to have it divided by speaker and timestamp and all of the rest of that. I have friends who have used models like this for all sorts of fun purposes. I have a friend whose spouse is a reporter, and they built an entire workflow system for them that takes their recordings of their interviews and processes it and produced diarized transcripts of the interviews and then gives them the ability to search or say things like, okay, so who was the person that said this particular thing to me and when did they say it so that they get traceability for their articles and can do fact checking and what have you.

224
01:28:00.000 --> 01:28:47.000
<v Perry E. Metzger>But yeah, so yes, most of the time these days I spend my time on AI issues and AI research. Though at the moment I am not doing work on AI itself but on applying AI, as I said, to doing more or less the computer security work. Although rather more fundamental computer security work than the stuff that I used to do on a common basis. And I also spend, as I said, a lot of my time on AI advocacy these days, partially because I got disgusted and annoyed at various AI doomers on social media and decided at one point that someone had to do something about it.

225
01:28:47.000 --> 01:29:14.000
<v Perry E. Metzger>So I started an advocacy organization to oppose people who hate AI systems and think that they're going to kill everyone. A surprising amount of money behind such people. There are various organizations currently spending hundreds of millions of dollars a year on trying to convince people that AI is going to kill them. But never mind that.

226
01:29:14.000 --> 01:29:22.000
<v Vlad Costea>I would not be surprised if that money is coming from Russia or China. A small amount comes from China.

227
01:29:22.000 --> 01:30:19.000
<v Perry E. Metzger>A surprising amount comes from various internet billionaires who have been convinced of the truth of the idea. Although there appears to be a bunch of money coming from China that's been fueling the anti-data center movement in the United States. That's not entirely organic. Though I think that even if no more money was spent by the Chinese at this point, that's become such a self-sustaining mental illness that it probably would not go away on its own. But yeah, these days I spend more of my time on that than worrying about cryptography. I remember many years ago I had a conversation over dinner with Ralph Merkle, who was one of the people... Pardon?

228
01:30:19.000 --> 01:30:21.000
<v Vlad Costea>Nice. I'm jealous.

229
01:30:21.000 --> 01:30:52.000
<v Perry E. Metzger>Yeah, I used to know him reasonably well. I haven't spoken to him in some years. And I remember more or less saying to him, well, why did you leave working on cryptography? And his answer was more or less, well, you know, after I did the Khufu and Khafre proposals, which was a cypher that he proposed at one point, and someone cryptanalyzed them, you know, within like 48 hours.

230
01:30:52.000 --> 01:31:35.000
<v Perry E. Metzger>I realized that, you know, that the low hanging fruit, the easy stuff was kind of gone from the field. And, you know, I was interested in, you know, and I was interested in other things. For me, it's not that I'm not interested in cryptographic security or cryptocurrencies or what have you anymore, or computer security in general. I'm still interested in them. But the amount of work that one has to do in order to do interesting new things these days is fairly high. And the amount, and my dedication to working on certain problems is not sufficiently high.

231
01:31:35.000 --> 01:32:21.000
<v Perry E. Metzger>So I have moved on, as I said, to worrying about things that are much more in the programming languages world in some sense. Believe it or not, things like formally verifying software is technically like a sub branch of programming languages in computer science, even though it's not right. It's not really about programming languages as such. So I've, in my old age, I've become, you know, much more interested in that stuff. And I spend much more of my time on that stuff than I do on things like cryptocurrencies now. Anyway, do you have any interesting questions from the listeners or uninteresting ones? Or should I reveal the number at this point?

232
01:32:21.000 --> 01:32:27.000
<v Vlad Costea>Oh, no, let's do that. Right at the end of the interview. Yeah, because I want people to stick around and listen.

233
01:32:27.000 --> 01:32:36.000
<v Perry E. Metzger>You want to convince people that they should stick around to the end in order to get their 50 bucks. Sounds good. Oh, yeah. Yes, very good.

234
01:32:36.000 --> 01:33:05.000
<v Vlad Costea>Someone asked, and I'm actually surprised because the chat is flooded, or I remember you using this term from French inundated. And I was surprised. I was like, Oh, my God, this guy is quite a linguist here. But it's being flooded right now with numbers. But there's actually a question from someone who's asking if you remember someone named Yonatan Sompolinsky on the mailing list. I'm afraid I don't.

235
01:33:05.000 --> 01:33:20.000
<v Perry E. Metzger>This doesn't mean that such a person wasn't on the mailing list. It's just that, you know, age and time tend to wipe out one's memories. But I don't remember that name. Apologies.

236
01:33:20.000 --> 01:33:32.000
<v Vlad Costea>No, it's all good. He's just trying to legitimize the cryptocurrency that he likes. So he's like, Oh, so if Satoshi was there, my guy must have been there too.

237
01:33:32.000 --> 01:33:52.000
<v Perry E. Metzger>Entirely possible, but I don't remember. You know, I am not a youngster anymore, unfortunately. It's weird. You know, when I think back, I mean, the first computer I worked on was in the 1970s. It's been quite a long time.

238
01:33:52.000 --> 01:33:56.000
<v Vlad Costea>But what was your first computer? Like, do you remember? Oh, yeah.

239
01:33:56.000 --> 01:34:42.000
<v Perry E. Metzger>Oh, I remember it very well. It was called a PDP-8/e. It was a mini computer made by Digital Equipment Corporation. And the way that one interacted with this thing was with a teletype, which is to say a machine that printed characters that were being sent by the computer onto paper and you typed onto the paper and those characters were sent back up to the computer. You know, this was a rather loud process. I think to my dying day, I'll remember what an ASR-33 teletype sounded like when it was printing 10 characters per second of output from the computer at you.

240
01:34:42.000 --> 01:35:10.000
<v Perry E. Metzger>That thing had four kilowords of memory, which is not enough to store like an icon on my desktop now. And yet I got a lot of work done on the thing, which is kind of amazing. It's been a very long time since then. I do not miss old computers very much, although I occasionally do work on simulation systems for old computers as a hobby. Anyway.

241
01:35:10.000 --> 01:35:14.000
<v Vlad Costea>You probably have very cool parents if they bought you this.

242
01:35:14.000 --> 01:35:36.000
<v Perry E. Metzger>They did not buy that for me. That was at my high school. They could not have afforded that. That was probably an $80,000 item in current dollars. And that would have been beyond their means to buy. And in fact, they never bought me a computer. When I went to university, I bought myself an early IBM PC, which I had saved.

243
01:35:36.000 --> 01:35:39.000
<v Vlad Costea>Which one was it, the XT?

244
01:35:39.000 --> 01:36:14.000
<v Perry E. Metzger>No, it was an original IBM PC. It was the B revision, not the A revision. But I, you know, I saved money all through high school, you know, doing odd jobs and what have you so I could buy that thing. I think it may have been as much as like four or $5,000 in the dollars of the time, which was a lot of money. It had taken me forever to save up that money. And I was very happy with the thing. But very quickly, of course, you know, I studied computer science and what have you.

245
01:36:14.000 --> 01:36:49.000
<v Perry E. Metzger>I ended up working with much more impressive computers and much more interesting computers, not that long afterwards. But yeah, my first machine was a very weak and pathetic looking computer by modern standards indeed. We have much better things now. You know, I mean, I think there are much better things in my microwave oven. There are probably, in fact, I suspect that smart light bulbs have more computing power than the computer that I started with by a lot.

246
01:36:49.000 --> 01:36:53.000
<v Vlad Costea>It is possible because they have a microcontroller. Oh, they all have microcontrollers.

247
01:36:53.000 --> 01:37:21.000
<v Perry E. Metzger>Yes, I have done talks for my book on computers for kids where I lift up a light bulb for the audience and I mentioned this has a computer in it, which is not something you would expect. We throw them around like candy these days. It's kind of remarkable to someone who has been watching this as long as I have. But anyway, are there any other questions from your audience?

248
01:37:21.000 --> 01:37:27.000
<v Vlad Costea>When was the first time you tried Bitcoin? Since you knew about it, since it was announced, but you did not try it.

249
01:37:27.000 --> 01:38:12.000
<v Perry E. Metzger>Very early on, I think I played with it for a few minutes. But, you know, I didn't really do anything with it because there was nothing you could do with it. These days, there's lots and lots of interesting things you can do, if only to play around with it and to look at it. You can run a block explorer and look at the chain. You can watch the transaction flow and everything else. But early on, there was nothing there. So it was pretty boring. So I think I may have used it for all of five minutes, you know, compiled the code and have a look around at it and, you know, walked away.

250
01:38:12.000 --> 01:38:19.000
<v Vlad Costea>When was the moment you came back to it and realized, OK, there was more to it, so maybe I should run it again or receive a transaction?

251
01:38:19.000 --> 01:39:14.000
<v Perry E. Metzger>I haven't. I don't spend a lot of time actually on cryptocurrencies these days. I mean, I have run blockchain software at various points. I have run nodes for various things at various points. But these days, my interest is fairly passive. I read papers occasionally and I keep up with what people are doing and that's about it. I have not written any blockchain software or examined any or what have you in quite a long time. The last time I actually had anything serious, I was in discussions with someone about trying to implement a formally verified version of the Bitcoin software.

252
01:39:14.000 --> 01:39:39.000
<v Perry E. Metzger>And that conversation didn't really go anywhere. Although it would still be a very neat project to do. I wouldn't mind doing that at some point. It would be fun. But I'm actually very boring from a cryptocurrency development and research point of view these days. I have almost no role these days. I think it's been five years since I did lots of such stuff on a regular basis.

253
01:39:39.000 --> 01:39:46.000
<v Vlad Costea>The question was more like when did you return to Bitcoin realizing that there is actually some value to it, not the most recent interaction?

254
01:39:46.000 --> 01:40:24.000
<v Perry E. Metzger>I wouldn't be able to tell you what year. It grew kind of organically and then there finally came a point where I had clients that wanted me to explain what this thing was. And I had to actually read a bunch of the code in the papers and make sure that I understood it. There are relatively subtle portions of how the whole thing works. If someone asks you to explain the whole thing, you have to become familiar. But I wouldn't be able to tell you what year. It was a few years after the launch.

255
01:40:24.000 --> 01:40:32.000
<v Perry E. Metzger>I did not for the first few years spend particularly much time on it. I looked at the thing and walked away for quite some time.

256
01:40:32.000 --> 01:40:42.000
<v Vlad Costea>Based on your description, I presume it was most likely 2013, but I was going to mention this because it was... Sooner or later than 2013.

257
01:40:42.000 --> 01:40:59.000
<v Perry E. Metzger>I wouldn't be able to tell you. Unfortunately, my memories don't come with timestamps. That would be a neat feature, but you know, perhaps in the upgraded software from, you know, what I run in my brain. What I have right now doesn't tell me dates. At least not usually.

258
01:40:59.000 --> 01:41:21.000
<v Vlad Costea>It's all good. My point with this question is that I went through all of my emails and all of my old messages to see if there is anyone who actually tried to get me into Bitcoin and telling me, hey, here's a free coin or whatever, give this a try. That never happened to me. And the fact that you don't mention this, I assume that it never happened to you either. And you're the creator of the mailing list.

259
01:41:21.000 --> 01:42:06.000
<v Perry E. Metzger>Well, no. I mean, I was around so many people who were using it, you know, it eventually became difficult to ignore. But, you know, no one had to reintroduce me to it. I was familiar with what it was. I just hadn't spent any real time thinking about it for a long time. I started having friends who were mining Bitcoin, you know, I mean, some years ago. You know, don't ask me what year and, you know, and, you know, and, you know, who are trying to use it for transactions and what have you. And at that point, it became obvious that, you know, one should pay attention again, but don't ask me the year. I couldn't tell you. And no, no one, no one came up to me and said, you know, here's, you know, here's a Bitcoin.

260
01:42:06.000 --> 01:42:12.000
<v Perry E. Metzger>Do something with it. That experience did not happen to me.

261
01:42:12.000 --> 01:42:16.000
<v Vlad Costea>That's a bit sad, given your role in the history of the project.

262
01:42:16.000 --> 01:42:21.000
<v Perry E. Metzger>Why would anyone have done it? I already knew what it was.

263
01:42:21.000 --> 01:42:27.000
<v Vlad Costea>I don't know. For whatever reason, just to send you a Bitcoin like that was a fake.

264
01:42:27.000 --> 01:42:46.000
<v Perry E. Metzger>No one ever sent me one early on. I guess I had friends who were mining it in, you know, in their bedrooms and such. Which I never did. Not even when I had free electricity in my apartment. But there you go.

265
01:42:46.000 --> 01:42:52.000
<v Vlad Costea>OK, coming back to the AI issue, because I'm not sure we are.

266
01:42:52.000 --> 01:42:57.000
<v Perry E. Metzger>We are getting we are getting close to two hours and I'm going to have to leave a little bit after four o'clock my time.

267
01:42:57.000 --> 01:43:02.000
<v Vlad Costea>What is your current setup and what kind of models do you run?

268
01:43:02.000 --> 01:43:43.000
<v Perry E. Metzger>I tend to run pretty big models. I don't run models locally very much. I do a little bit for experimentation and learning. I occasionally will do things like fine tuning small models in order to play with that. And I've spent a lot of time teaching myself the underlying technology. You know, I've trained a few small things. But when I'm doing coding and stuff, I tend to use Codex CLI and I tend to use GPT-5.6 Sol for I think it's been has it even been one week at this point?

269
01:43:43.000 --> 01:44:24.000
<v Perry E. Metzger>I can't remember. Maybe it's been a week and a half. Time on this stuff goes very, very fast. I tend to use the most cutting edge OpenAI models when I'm doing software engineering. Not out of any particular deep conviction. Mostly because it's the most convenient stuff. I probably will spend more time playing with stuff with the latest Grok model at some point. I'm sure there's going to be a new release of that soon. And at that point, I'll probably spend more time. I've also played a bunch with things like GLM-5.2 and what have you.

270
01:44:24.000 --> 01:45:07.000
<v Perry E. Metzger>But as I said day to day, I tend to use, you know, GPT-5.6 Sol for software engineering work. You know, I spend too much money on that. It's not a cheap license. Or rather not a cheap subscription. You know, I tend to use command line tools with this stuff just because I'm more comfortable with that. I've been using, I've been a command line person since the start of time. And when that became a thing with Claude Code and Codex CLI, I took to it very well. I still don't use the GUI version of Codex, even though it's around.

271
01:45:07.000 --> 01:45:45.000
<v Perry E. Metzger>It never seemed like it was particularly doing anything for me that I wanted. So I didn't. No. And these days, as I said, most of what, I mean, I have a bunch of projects I work on. But the most important project I'm working on at the moment is I'm building a formal verification framework for system software. You know, which you can roughly think of as, you know, things that need reasonable amounts of performance and where it's difficult to have a garbage collector hide the, you know, the problems they face. You might have with memory management.

272
01:45:45.000 --> 01:45:50.000
<v Vlad Costea>I think that Zooko is working on something similar. Maybe you should get in touch.

273
01:45:50.000 --> 01:46:03.000
<v Perry E. Metzger>I do not know. I have not spoken with him in a while. I know that I tend to see him at events. I have not actually called him up in years.

274
01:46:03.000 --> 01:46:22.000
<v Vlad Costea>I'm in Denver this year, and he showed me some benchmarks that he's working on this memory allocator to make use of RAM more efficient. And he was sort of competing with a Microsoft product. And he showed me that his open source thing is much more efficient.

275
01:46:22.000 --> 01:46:48.000
<v Perry E. Metzger>He has worked on interesting projects for many years. I think, for example, that Tahoe-LAFS is an underappreciated project. It never quite took off, but it's a very cool piece of software that provides you with the ability to store data in a distributed and cryptographically private way across a large cluster of computers all over the Internet, which is a fun thing.

276
01:46:48.000 --> 01:46:56.000
<v Vlad Costea>I mean, I do remember there was a TV show called Silicon Valley where they were building something similar.

277
01:46:56.000 --> 01:47:10.000
<v Perry E. Metzger>I and people have told me to watch that a number of times, and I never actually have, which is perhaps a bit surprising. But there you go. Anything else you wanted to talk about today?

278
01:47:10.000 --> 01:47:18.000
<v Vlad Costea>Yes. What are the arguments that you usually use in advocacy for AI? And what kind of arguments do you have to combat?

279
01:47:18.000 --> 01:48:07.000
<v Perry E. Metzger>Well, the arguments that I have to combat all the time are, you know, from, you know, loony people who believe that AI systems, if they are allowed to if we continue developing them, that one morning they are going to, you know, recursively self-improve in a brief period of time and then kill all of us. For, you know, for not particularly well specified reasons. You know, people like Eliezer Yudkowsky, who I have unfortunately known since he was a teenager. You know, I had another mailing list that was pretty famous in its day called Extropians that also has had a disproportionate impact on the world.

280
01:48:07.000 --> 01:48:51.000
<v Perry E. Metzger>And, you know, I find myself combating a lot of things that I said for amusement and humor value decades ago are now come back to me as serious arguments for why AI is dangerous, which I made the comment that you never know what loose talk will do. And I discovered this myself. You know, shit posts that I put up on this mailing list decades ago have turned into an entire belief system about artificial intelligence. But mostly what my organization does is, you know, lobbies in Washington, D.C.

281
01:48:51.000 --> 01:49:26.000
<v Perry E. Metzger>To try to prevent people from interfering too much with AI research and development because I think that AI research and development is a wonderful thing. You know, it is already saving lives. It's improving the world around us very rapidly. It's one of the most exciting technologies in existence. And I know a lot of people that disagree with this very heavily. And, you know, I have to spend a lot of time trying to disagree back with them. But unfortunately, I do have to go in a couple of minutes.

282
01:49:26.000 --> 01:49:38.000
<v Vlad Costea>So, yeah, I think you mentioned something very important, the Extropians mailing list, which I just looked up. I would be really grateful if you could introduce me to Robin Hanson, if you know him well.

283
01:49:38.000 --> 01:50:23.000
<v Perry E. Metzger>I know Robin Hanson extremely well. I have known Robin for decades. You know, but, you know, you could you can ask me offline. I may or may not be able to introduce you. It depends on him. It doesn't depend on me. You know, but yes, Robin is an interesting character. You know, I still remember getting my copy of Extropy magazine, which had his first article on idea futures in it, which was originally discussed on our mailing list, among other things. You know, and that, of course, has led to things like various prediction markets these days.

284
01:50:23.000 --> 01:50:29.000
<v Vlad Costea>Too bad they were used for sports betting and not more interesting stuff.

285
01:50:29.000 --> 01:51:08.000
<v Perry E. Metzger>Robin has this has the idea that they can be used for, you know, as a replacement way of making government decisions, you know, futarchy and what have you. I am not I'm not as sanguine about the possibility, perhaps, as he is. But, you know, it's been interesting watching that idea, you know, continue to perpetuate itself over a long period of time. Anyway, it's been nice talking to you, Vlad.

286
01:51:08.000 --> 01:51:13.000
<v Vlad Costea>Likewise, sir. Please reveal your number and let the people know how they can follow you on the Internet.

287
01:51:13.000 --> 01:52:00.000
<v Perry E. Metzger>The number is 68. I have no idea who might have magic number 68. But I wrote 68 on a piece of paper after I couldn't get myself to pick a number without, you know, picking a stupid number. So I used random.org to generate one on my phone. As for how to follow me on the Internet, I have a Substack that I barely use called diminished capacity. But I may be using that more in the not that distant future. I'm on X a lot as Perry Metzger, you know, and I post way too much there.

288
01:52:00.000 --> 01:52:38.000
<v Perry E. Metzger>And I think those are the most obvious ways to follow the things that I write. Anyway, and of course, I have that book, you know, and, you know, Macmillan will be very happy if you buy copies of Science Comics: Computers. And there might even be, you know, a 12 year old, you know, who will learn a lot about how computers work by reading it. Or maybe they'll get angry with you for giving it to them and never speak to you again. You know, but, you know, they're as I said, the books are really great for erosion control and all sorts of other purposes. Anyway, great talking to you, Vlad.

289
01:52:38.000 --> 01:53:01.000
<v Vlad Costea>Thank you for being here. There's a person. So before we end the interview, I just want to acknowledge that there is a person who guessed 68 and his name is NatoshiSakamoto42000. So I'm gonna message you, Natoshi, and let you know how you can claim your prize. Thank you very much for being here. Thank you for listening. Thanks to the sponsors. And I'll see you in the next one.

290
01:53:01.000 --> 01:53:31.000
<v Perry E. Metzger>Have a good day.
<v Perry E. Metzger>================================================================================ Production notes: CORRECTED. Base: the pipeline's local Whisper medium.en transcript (int16) with speaker diarization (Guest A -&gt; Perry E. Metzger). Compared word by word against a full local small.en pass (int16, VAD, word timestamps) and YouTube auto-captions (zcash-documentary S17E32.en.vtt / bitcoin-forks-research S17E32_perry_metzger), with a three-way diff. Speaker turns checked with a full-episode local TitaNet embedding scan (two reference voices: Vlad, Perry). 26+6 medium.en int16 no-VAD word-timestamp clips for disputed stretches (clips1 + clips2/clips3). Restored "Hello there" opening; fixed overlapping Qwen/Kimi exchange; restored Chaum clarification, Naomi/privacy talk, soul/comic-book exchange, Dowdeswell origin of metzdowd, Kimi K3 / GLM-5.2 local-models passage, and GPT-5.6 Sol (confirmed against OpenAI model naming). Names and jargon checked against the title, feed, and web: Cryptography Mailing List, metzdowd.com, cypherpunks, Satoshi Nakamoto, Hal Finney, Phil Zimmermann, PGP, Tim May, David Chaum / DigiCash, Chuck Hammill (From Crossbows to Cryptography), Whit Diffie, David Kahn (The Codebreakers), Ray Dillinger, Adam Back / Hashcash, Nick Szabo, Wei Dai / b-money, Tatu Ylönen / SSH, Naomi Brockwell / Ludlow Institute, Alliance for the Future, Extropians / Extropy magazine, Robin Hanson / idea futures / futarchy, Eliezer Yudkowsky, Zooko / Tahoe-LAFS, Tezos, SegWit, Qwen, Kimi K3, GLM-5.2, Codex CLI, Claude Code, GPT-5.6 Sol, Science Comics: Computers (Macmillan), SideShift, Cake Wallet, Orange Rock, LayerTwo Labs, Braiins Hashpower, Chat Control. Light clean verbatim. Intro music omitted. Remaining doubtful words marked ....
