Conspiracy Theory, Up Close & Personal

Hal Finney made 33 posts here on Overcoming Bias from ’06 to ’08. I’d known Hal long before that, starting on the Extropians mailing list in the early ‘90s, where Hal was one of the sharpest contributors. We’ve met in person, and Hal has given me thoughtful comments on some of my papers (including on this, this, & this). So I was surprised to learn from this article (key quotes below) that Hal is a plausible candidate for being (or being part of) the secretive Bitcoin founder, “Satoshi Nakamoto”.

Arguments for this conspiracy theory:

  • Hal lives a few miles from the guy Newsweek recently claimed was Nakamoto, and who admitted to being involved somehow.
  • Bitcoin is very carefully thought out and implemented, and Hal is one of the top few people in the open crypto world who have demonstrated this capacity. For example, Hal did most of the work behind PGP 2.0, perhaps the most successful open crypto predecessor to Bitcoin.
  • Hal is on record as the first guy besides Nakamoto to use Bitcoin software, he got the first coin transfer from Nakamoto, and he made some key software improvements.
  • Hal’s writing style is much closer to Nakamoto’s than anyone else who the many reporters digging into this have suspected of being Nakamoto.

The arguments against this conspiracy theory:

  • In a world has seven billion people, the prior on Hal being Nakamoto has be rather low.
  • Hal says he isn’t Nakamoto, and seems sincere.
  • Hal says Nakamoto understands C++ better than he does.
  • Hal’s son showed a reporter some gmails between Hal and Nakamoto. The reporter says:

The notion that Finney alone might have set up the two accounts and created a fake conversation with himself to throw off snoops like me, long before Bitcoin had any measurable value, seemed preposterous.

That last point seems pretty weak. We already know that the Bitcoin founder wants to be hidden. If Hal really created Bitcoin, he is plenty smart enough to think that Bitcoin might succeed, and to think of and implement the idea of creating fake conversations to cover his tracks. In this case Hal would also plausibly lie about his C++ skills, or maybe he got C++ help from someone else. In any case the probability of seeing those things conditional on Hal actually being Nakamoto seem pretty high.

It seems to me that the question comes down to your prior expectation on whether the person who did such a careful expert job on something so hard would be one of the few people in the field most known to be capable of and to have actually done such things, or whether it would be a new largely unknown person. And thinking about it that way I have to put a pretty large weight on it being someone known. And conditional on that it is hard for me not to think that yeah, there’s at least a 15% chance Hal was more involved than he’s said. And if so, my hat’s way off to you Hal!

But I also figure I’m not paying nearly as close attention to this bitcoin stuff as many others. Google doesn’t find me any other discussion of the Hal as Nakamoto theory, but surely if I wait a few weeks others who know more will weigh in, right? And since I can’t think of any actions of mine that depend on this issue, waiting is what I’ll do. Your move, internet.

Added 8a 26Mar: In the comments, Gwern points to further reasonable indicators against the Hal as Nakamoto theory.  I accept his judgement.

Those promised quotes:

March 6th, the day that Newsweek released its bombshell cover story on the man who it claimed had invented Bitcoin: Dorian Prentice Satoshi Nakamoto, a 64-year old ex-engineer and programmer living in the small exurb of Los Angeles known as Temple City. Nakamoto had even seemed to give Newsweek a tacit confirmation of its theory when he told the magazine’s reporter that he was “no longer involved in that,” a quote confirmed in essence by local police who witnessed the interaction. …

Just hours after Newsweek‘s story hit the Web, I received an email from an old cryptography community acquaintance of Finney’s, … [who] pointed out that Hal Finney had lived for almost a decade in Temple City, the same 36,000 person town where Newsweek found Dorian Nakamoto. Finney had lived only a few blocks away from Nakamoto’s family home. …

I collected a 20,000 character sample of Finney’s writing from various forums and mailing lists and sent it to Juola & Associates for analysis…. I received the results from the writing analysis from Juola & Associates. The firm, its chief scientist John Noecker explained in a phone call, had previously tried analyzing candidates for Satoshi Nakamoto named by older investigations performed by the New Yorker, Fast Company, and various Bitcoin enthusiasts. None of the results had been promising enough to publish, according to Noecker. …

Finney contacted Zimmermann and became one of his earliest collaborators. He worked almost a full-time job’s worth of hours developing PGP 2.0, widely considered to be the first truly secure version of the program, and pioneered its “web of trust” model of key-signing, … Though Finney told me that he worked closely with Zimmermann to code “the bulk of the changes” from PGP 1.0 to PGP 2.0, he’s rarely been fully credited for that work. As PGP’s usage spread, Finney was also the first to integrate the encryption software into “remailers”–free services that acted as proxy servers for email, bouncing messages among third parties so that they couldn’t be traced to their source. …

Finney is known to be the second-ever user of Bitcoin after Satoshi Nakamoto himself. He had been one of the first supporters of the idea when Nakamoto floated it on a cryptography mail list, and even received the first Bitcoin test transaction from Nakamoto in early 2009. … Finney downloaded the early Bitcoin code and began running it on an IBM Windows desktop tower machine. He’s widely believed to be the first person other than Nakamoto himself to do so. He kept it running for weeks–just how long, he declined to tell me. With no competition, he was able to mine as much as a hundred coins a day using only his old PC’s off-the-shelf CPU. … His doctor diagnosed him with ALS in August of 2009. …

As Finney’s muscle control declined over the next years, he continued to write code for Bitcoin. At one point he wrote up an improvement to its elliptic-curve cryptography that would speed up its transactions by as much as 20%. Even after he lost the ability to type with both hands, and then to type at all, he continued to use eye-tracking software to write code, including a program called bcflick aimed at better securing Bitcoin wallets. …

Then Jason Finney brought me over to a computer to show me what he’d prepared: It was Finney’s Gmail account, opened to a January 2009 conversation between Finney and someone named Satoshi Nakamoto—clearly, the Satoshi Nakamoto. They had exchanged about fifteen emails, in which Finney described bugs he’d found in the early Bitcoin code and Nakamoto responded with fixes and notes of thanks.

The Nakamoto on the other end of the conversation wrote in fluent, colloquial English, not at all like Dorian Nakamoto. (I would later forward the emails to the same linguistics analysis firm who had earlier analyzed Nakamoto’s writings, Juola & Associates, and they confirmed that the new emails matched the style of the Bitcoin whitepaper more closely than Finney’s writing, and far more closely than Dorian Nakamoto’s.)

Jason Finney also let me see the transaction record for his father’s Bitcoin wallet from 2009, which clearly showed that Finney had received Nakamoto’s ten-bitcoin test transaction on January 11th, 2009. The wallet evidence, along with the Gmail timestamps, would have been very difficult to forge. And the notion that Finney alone might have set up the two accounts and created a fake conversation with himself to throw off snoops like me, long before Bitcoin had any measurable value, seemed preposterous. (more)

GD Star Rating
loading...
Tagged as: ,
Trackback URL:
  • http://www.gwern.net/ gwern

    FWIW, I’ve spent a lot of time looking into Satoshi, and I tend to agree that Finney is generally overrated as a candidate. I wouldn’t give him >5% at this point.

    > In a world has seven billion people, the prior on Hal being Nakamoto has be rather low.

    Not really. Remember, we know a fair amount about Nakamoto, so 1/7b is way off as a prior: for example, that he is a fluent English writer (probably a better writer than 95% of the American population) as is Hal, which alone cuts it down from 7 billion people to, what, maybe 15 million people worldwide? Throw in some decent math skills (not up to, say, a math major but still more than the overwhelming majority of people), cuts that down some more; throw in the spare income/wealth for AnonymousSpeech hosting, still some more; throw in C++ competency (how many C/C++ programmers of any competency are there worldwide? some quick googling suggests maybe 20m software developers worldwide, and C/C++ seems to be increasingly rare as people focus on technologies and languages higher up the stack), libertarian beliefs etc etc. Even after accounting for the variables overlapping, you can still rule out like 99% of the population as impossible. So when you look at someone like Finney, a better prior might be something like 1/1m. You can’t simply ignore everything we know about Finney in giving a prior before looking at more specific datapoints which might point at him being Satoshi. (Unless you’re going to start with 1/7b and immediately take into account all those other factors to arrive at a better estimate like 1/1m, of course, which is going to amount to the same thing.)

    • http://overcomingbias.com RobinHanson

      I didn’t say that the prior should be 1/7b. So what arguments against the Finney theory weigh most heavily for you?

      • http://www.gwern.net/ gwern

        Offhand: corresponding with himself is an elaborate deceptive ruse of a sort Satoshi does nowhere else even years later when people were far more interested in his identity than in 2008-2009*; the coding styles look nothing alike (a week or two ago I was comparing RPOW with the original 0.1.0 release, so I came to this conclusion independently) and in particular Finney’s code entirely lacks certain stylistic quirks Satoshi has that have not been publicly described; Satoshi’s daily activities do not sync well with a California schedule; and his overall arc of activity doesn’t seem to fit with Finney’s deterioration. There’s also a few items in my rubric that I haven’t checked for Finney but expect him to not match, like Windows use and ignorance of Latex.

        * and one part of their interactions may have been unnecessarily & quite dangerous to Satoshi’s pseudonymity, if this theory is true

      • sflicht

        Maybe I’m naive but Hal’s personal account of his own journey with bitcoin strikes me as evidence against his being Satoshi. https://bitcointalk.org/index.php?topic=155054.0

  • Wes Felter

    The really good conspiracy theory is that Bitcoin is an improved version of Finney’s RPOW, but I see that gwern is already on the case there.

  • Anonymous

    A better question than who is behind bitcoin is what does it matter? As someone who always tries to stay anonymous in the internet, I can certainly understand why “Satoshi Nakamoto” wants to stay hidden. Since we don’t really need to trust him in order to use bitcoin (since everything is open), I don’t see what can possibly be gained by revealing the identity of this person against his will.

    • http://www.gwern.net/ gwern

      If nothing else, it’d help force the issue of his hoard of bitcoins: he may have up to 10% of all extant bitcoins (depending on your assumptions about mining and how many coins have been lost). That’s a permanent lingering uncertainty, which does not seem helpful to the Bitcoin ecosystem.

  • Tom

    Is providing evidence for your friend being the anonymous mastermind of a massively influential and lucrative enterprise really a good idea? If he wants to stay hidden and you consider yourself his friend, then it seems like the virtue of silence is the right thing to practice here. You even admit to this information not being actionable to you!

  • Marc Geddes

    Good arguments by Gwern, I thought Finney was a good suspect, but now I would tend to look elsewhere. Other prime suspects are Wei Dai and Nick Szabo.
    I believe the net is starting to close in on Wei Dai, since more and more evidence is pointing towards him as Satoshi, if only by a process of elimination.
    The key point here is the tremendous power available to ‘Satoshi’. Money is power, and hidden money is even greater power. Note that the 100s of millions of dollars potentially available to ‘Satoshi’ *is* enough to initiate a secret AGI project capable of rivalling even the biggest corporations or govt. departments.
    ‘Satoshi’ *does* have the power to execute the biggest conspiracy of all time…
    beating out all competitors and getting to real AI first!

    • http://entitledtoanopinion.wordpress.com TGGP

      Do we even know anything about who “Wei Dai” is?

      The most interesting thing about Satoshi’s fortune is that he doesn’t appear to have spent any of it.

      • Marc Geddes

        Wei Dai page:
        http://www.weidai.com/

        Note interest in Crypto and C++. Fits the bill.
        I’m sure Satoshi can find ways to spend the money without being noticed, he wouldn’t even need to actually shift the Bitcoins if he was clever.

      • http://entitledtoanopinion.wordpress.com TGGP

        One of the fundamental facts about Bitcoin is that transactions are tracked in a public log. It’s sort of the point of the whole thing. Granted, I don’t have any in depth knowledge of how it works, but I don’t see how he’d be able to spend them without it being checked. Is he handing over access to his wallet through the sneakernet, with the buyers content merely to have access while also refraining from spending?

    • Doug

      Satoshi has approximately a billion dollars in relatively illiquid wealth. By comparison this is what the US federal government spends in three hours. Even for a smaller country like Germany this represents less than two days of total expenditure.

      What if major governments are bureaucratic and lumbering, and that Satoshi is brilliant and tech savvy? Apple has $147 billion in cash sitting on its balance sheet, Google has $43 billion, Microsoft has $77 billion, Cisco has $50 billion, Oracle has $40 billion, and the list goes on. All of these firms are stocked with at least dozens of developers near Satoshi’s level brilliance, plus they have much more experience in managing gigantic projects.

      It’s a cool idea, but there’s only a sliver of a chance that Satoshi could make a giant a technological advance ahead of the many many resource-rich and technologically savvy entities in the economy.

      • Marc Geddes

        Past a certain point, throwing more money at a problem won’t help. The AGI project is a basic research problem where progress occurs on a more linear path, for this, I guess that $US 100 million is sufficient to progress as fast as is possible.

      • Doug

        Okay, even if that’s true Satoshi may be brilliant doesn’t have any known experience in AI or Machine Learning. Someone like Geoffrey Hinton, who may not have the sheer raw cognitive power of Satoshi, is still much more likely to be the team that cracks AGI. (This is all assuming per your original post that AGI is cracked relatively soon and in a way that we can basically give most of the credit to a single team).

        Hinton’s still quite brilliant and has the advantage of decades of experience working on AI. Plus he already has an assembled team that’s been working on the problem for a long team. Finally as part of Google he has vast networks of resources to draw on, not just money but access to Google’s already built huge organization of developers and scientists. I don’t see what advantage Satoshi would have over Google.

  • Christian Kleineidam

    If I would create something like Bitcoin and wanted to stay annonymous, why would I make the first transaction from my pseudonymous account to an account linked to my real world identity.

    That seems to be extremely careless for someone with that background.