[Freedombox-discuss] FreedomBox and Bitcoin (and the petition)

[Jonathan Wilkes]

----- Original Message -----
> From: Ted Smith <tedks at riseup.net>
> To: "freedombox-discuss at lists.alioth.debian.org" <freedombox-discuss at lists.alioth.debian.org>
> Cc: 
> Sent: Monday, November 12, 2012 3:28 PM
> Subject: Re: [Freedombox-discuss] FreedomBox and Bitcoin (and the petition)
> 
> On Mon, 2012-11-12 at 11:56 -0800, Jonathan Wilkes wrote:
>>  > 
>>  > The Bitcoin transaction log records transactions between addresses. If
>>  > you never change your Bitcoin address, the transaction log will
>>  > accumulate records of your transactions. 
>>  > 
>>  > Without a very significant amount of work, it is not possible to link 
> a
>>  > Bitcoin address (even in this sense) to a home address, full legal 
> name,
>>  > payment information, etc.. 
>> 
>>  What makes you say it is a "very significant amount of work" to 
> determine
>>  the originating IP address for a bitcoin transcation?  How much did it cost
>>  you to connect to all the Bitcoin nodes in existence?  I assume you tried 
> or
>>  at least have a ballpark figure, since that is the most obvious way to
>>  link a Bitcoin address with an IP address.  (And as we both agreed above,
>>  when the attacker has the IP of the originator of the transaction they are
>>  only 1 step away from gaining info on home address, full legal name,
>>  payment information, etc...)[1]
> 
> I don't have any ballpark figure for that, since I'm using a nominal
> metric for "work" that could be loosely defined as:
> 
>       * Work level 0: you already have the information you're looking
>         for (you are an ISP and you are looking for the home address of
>         a subscriber)
>       * Work level 1: You don't have the information you're looking for,
>         but you can obtain it via a legal-system process (you are the
>         FBI and can ask an ISP for it)
>       * Work level 1 (continued): You don't have the information you're
>         looking for, but you can obtain it via an existing attack on
>         someone who does (you are the Illuminati and can hack the ISP)
>       * Work level 2: You need to write your own software to obtain the
>         next piece of information you're looking for.
> 
> Not all work is created equal. It's easy to connect to all Bitcoin
> nodes, but you'd have to write software to do it first. Writing software
> is hard.

I'm with you so far.

> 
> When the attacker has the IP address of a Bitcoin address, they're one
> step away from getting the underlying home address -- but they have to
> go through either the legal system or an existing attack. That's "one
> step," but it's not an easy step.

It's as easy as whichever of your so-called "work levels" is available to the
attacker.  For example, work level 1 for the FBI or state law enforcement
(or city law enforcement if you live in NY) is extremely easy.

> 
>>  > 
>>  > With very little work (running Tor and using new addresses), you can
>>  > anonymize your Bitcoin participation to the same extent you could
>>  > anything.

Why run Bitcoin over Tor when you already stated that
"without a very significant amount of work, it is not possible to link a
Bitcoin address (even in this sense) to a home address, full legal name,
payment information, etc.. "

(That's not a rhetorical question, btw-- there are plenty of Bitcoin users who
have argued this on the forum, even going further and saying that they don't
want to burden the Tor network with gigs and gigs of blockchain downloading.
In other words, there are a lot of Bitcoiners who really don't understand how
the underlying technology actually works, even though the devs quit using the
word "anonymity" as a feature long ago.)

>> 
>>  I love how spying on the entire Bitcoin network, which has been done,
>>  demoed, and reported to the Bitcoin community by Dan Kaminsky,
>>  constitutes "a significant amount of work" in your informed 
> opinion, yet
>>  downloading a 2gig blockchain over Tor is "very little work".
> 
> Yes, downloading a 2GB file over Tor takes about 10 seconds of human
> effort if Tor and wget are installed, and about 5 minutes of human
> effort if they aren't. If you assume a novice user that knows what Tor
> is, then it'll take as long as it takes to download and run the Tor
> Browser Bundle -- I don't know numbers off the top of my head, but I do
> know it's been studied extensively, and if you're very interested you
> can ask the Tor Project people to put you in contact with the people
> doing usability studies on the TBB.
> 
> I don't know how long Dan spends writing his talks, but I'd throw down a
> bitcoin that it takes longer than that, and is much more mentally
> taxing. 

That's off topic-- we were discussing whether it is easy or hard to link an
IP with a Bitcoin transaction, not how long it took for a researcher to
originally discover the technique and write it into his talk.

> 
> Further, the attacks discussed in that particular talk are pretty easy
> to fix, especially if you control the entire software stack (which FBX
> does).

Then go ahead and submit a patch to Bitcoin.  If it makes it impossible for
Bitcoin to be surveilled in the way that you now understand it can be,
you'll be doing a service to the Bitcoin community.  That's a lot more valuable
than making an uninformed claim about the difficulty of the attack.

Something else you problem aren't aware of-- Blockchain.info already lists
the IP of the Bitcoin node that relayed the transaction to it.  They even have
a whois link to ease the work it would take to browse to Google.

> 
>>  > 
>>  > Further, is the FBX going to tunnel all traffic through some TCP
>>  > mix-net? (I don't think it is.) All privacy is quantitative; there 
> is no
>>  > concept of perfect anonymity. Nothing provides 100% privacy, and the 
> FBX
>>  > isn't looking to do that anyway.
>> 
>>  Nothing is 100%, but that doesn't mean there are some things that are
>>  clearly _not_ anonymous in any way, shape, or form, and using Bitcoin
>>  without going through Tor is one of them.  (Additionally, you might want
>>  to check to make sure that the Bitcoin reference client knows that Tor
>>  is now randomizing the socks port, because it was previously waiting to
>>  see the "magic Tor port number" to turn off listening, and if you 
> don't do
>>  that you have worse than non-anonymity-- you have the false idea of
>>  anonymity, which is right where we started in this thread.)
> 
> Sure -- but putting Bitcoin through Tor is pretty cheap, far cheaper
> than attacking it via any mechanism we've discussed.
> 
> Other things that aren't anonymous in any way, shape, or form include
> all of the semantic web stuff discussed on this list, all of the social
> networking stuff discussed on this list, and XMPP.

And I promise when anyone comes on here and claims that any of those
feature a "certain kind of anonymity" when they don't, I'll challenge them.

> This seems to be in
> coherence with the original idea of the FBX, which was (from my memory)
> "Replace Facebook et all with free alternatives."

That's a separate issue.  My own opinion is that it's better to wait until
multisig transactions are fully implemented in the reference client (and any
of the other clients for that matter) before considering Bitcoin for inclusion
in the FBX project.  If the idea is that "you can have my data when you
pry it from my cold dead hands", you really don't want what to also be
holding a bar of gold when you say it.  That invites abuse, whereas a client
that makes it trivial in your "work level 0" sense to distribute control over
disparate geographical areas at least invites less.

-Jonathan