Another question: According to reports Silk Road and DPR made $80 million in commissions. Yet the FBI has only told us about seizing $3.6 million in BitCoins.
That would be because they've managed to acquire somewhat of a reputation both for not handing over depositor's money to any and every government thug that waves some paper at them and for returning that money to said depositor on request. A tough reputation indeed to maintain on 21st century earth.
Exactly. That severely overstates the revenue taken in as SR was in business for quite some time before BTC was worth much. SR started in early 2011 when 1 BTC was worth less than $1. Hosting alone would have taken many tens of thousands of dollars a month a today's exchange rates.
I'd wager Ross only kept enough coins accessible from the machine hosting SL for day-to-day operations, and sent the extra proceeds to an account the production machines didn't have access to, one with the private key either on his personal laptop or in a safety deposit box.
What's unfortunate for DPR is that, even if there is a pirate's booty out there, the whole murder for hire thing and being an accessory, if not a patricipant, in a massive amount of drug trafficking makes it very likely that he will spend a large chunk, if not the rest of his life, in prison, making the 80 mil all but unusable to him.
I would think he could get some very high quality legal representation w/ $80 million lurking out there. Although it might be complicated by the fact that he won't be allowed out on pre-trial release. And let's face it; he could very well be the fall guy for the originators of Silk Road and not the "real" Dread Pirate Roberts at all.
I'm pretty sure it was a reference to Breaking Bad character Walter White burying his $80mm dollars in barrels in the ground somewhere in the deserts of New Mexico.
It seems likely that the seized funds were taken from SR's escrow accounts. It matches up roughly with the amount of money said in the FBI complaint to be stored in escrow. I can only assume that they have no idea where his personal Bitcoin stash is, otherwise they would have hyped up seizing ~80mil.
In the US, lawyers have argued that forcing someone to hand over their encryption keys violates the Fifth Amendment right to protection from self-incrimination.
Is the implication supposed to be that seizure of Bitcoins is protected by the Fifth Amendment? That seems far-fetched.
Typically the 5th amendment argument applies to handing over passwords that are only in your head, not keys that are written down. The government can force you to turn over data you have.
The case for not turning over passwords for bitcoin wallets is probably stronger than for true crypt volumes.
If they arrest you with an encrypted volume on your laptop, they know its your's. The only question is what it is. In a lot of cases, the government knows what it is, they just need to be able to prove it. Courts call this a forgone conclusion and can compel you do use your decryption password( but not tell it to the cops).[0]
On the other hand, handing over bitcoin keys maybe the only way they can learn which bitcoins are yours. This is,arguably, not a forgone conclusion at all. And again, you are not just giving them access to the data, you are effectively being compelled to testify that it is your data.
And the inquisitors may have already proven that you are a witch, so what's a little harm in torturing you to confess.
They may have evidence that this is your wallet, but by supplying them with the password, you are testifying that it is your wallet.
"An act is testimonial when the act entails implicit statements of fact, such as admitting that evidence exists, is authentic, or is within a suspect's control."
"By entering the password Boucher would be disclosing the fact that he knows the password and has control over the files on drive Z."
Compelling Ulbricht to unlock the wallet would be compelling him to testify that the wallet is his. You may think this is a silly chain of reasoning, but I'm sure you will forgive me if I will continue to side with existing federal precedent. http://www.volokh.com/files/Boucher.pdf
Ultimately, Burcher was orderd to decrypt his drive because the government both knew it was his and knew part of what was on it.[0]
In Ulbricht's case there is a compelling argument that the bitcoin's contained in the wallet on his computer(assuming it was on his computer and not the server) are his. The question is what does the government know about those coins. Clearly, if they know Ulbricht's bitcoin id/public key, then they can look up his transactions in the block chain. If they don't, I'm not sure what they could possibly know.
Of course, given Ulbricht's apparently atrocious OPSEC, it seems likely his password is Ludwig von Mises and the FBI won't need his help at all.
So in the case of bitcoins ran through a laundry such as what Silk Road ran, they might have almost no idea who's money you have. ( though if Silk Road was dumb enough to just launder money used on its site, it's all drug money, so that's moot).
What happens if you have a bunch of illegally-acquired gold stored away in an unknown location, but the government has good reason to believe that you posses it? Are you forced to reveal the location of the gold? What are the repercussions of not handing over that information?
I can see a consistent worldview where the government cannot ask you to decrypt something as part of an _investigation_, per the Fifth Amendment, but once you're convicted, can compel you to surrender something protected by an encryption key (as part of a proceeds-of-crime rule or something, or as part of seizing assets).
If the feds took an encrypted wallet file and are preventing the accused from accessing it, but haven't decrypted it, whether it's "seized" seems somewhat a matter of semantics.
(IANAL and I have no idea whether this meshes with the actual legal state of things in the US.)
It would be difficult to be sure that there is only one person with the key. If the defendant has a co-collaborator who also has access to the keys and is outside the reach of the seizing authority (or unknown to them), they could just transfer the Bitcoins to another new address and out of seizure.
The seizing authority in general has no way to tell in advance if such a co-collaborator exists, and assuming the transfer is to a new address with no historical transactions, it would be extremely difficult to identify who the collaborator was after the fact.
The second point is specific to a digital asset which can be duplicated (in this case we are also talking about any passwords).
Bitcoin, by definition, allows the second point, but requires a transaction. So the FBI should have made a transaction immediately (or have the means of invalidating any transaction of those bitcoin).
This is novel and critical to the story, but the Guardian article seemed to tangentially miss it.
If they have a near global view of the network, they can do traffic correlation without having to 'ask' hosting providers - and putting a large signal through would certainly help that traffic correlation to occur.
It is possible that DDOS attacks are not actively initiated by the NSA, but that they greatly help them complete their traffic correlation analysis network to an acceptable level of certainty by lighting up the circuits from the rendezvous points to the hidden service.
That, or just look through old forum posts to see if the announcement of a bitcoin drug marketplace looking to hire php programmers was signed rossulbricht at gmail dot com.
That doesn't explain how they were able to find the server hosting the hidden service by July the 23rd. The complaint relies on a lot of circumstantial evidence to tie the server to the person, and that evidence wouldn't easily have allowed them to find the server (e.g. that someone accessed a VPN that in turn accessed the Silk Road server from an Internet cafe close to where Ross Ulbricht lived).
If there is stronger evidence that they could have followed from the person to the server, they would probably have put that in the complaint as it would be more direct and therefore more incriminating.
Therefore, it seems likely that they didn't find the server by following it from the person. There are several possibilities here:
1. Perhaps the NSA helped with finding the server by traffic analysis to follow the circuit back to the hidden service IP?
2. Perhaps the server was rented from / on the advice of an informant and setup only shortly before it was imaged. Reading between the lines, redandwhite was a law enforcement plant on Silk Road (possibly Canadian, given that is what the article claims) that DPR trusted to some extent and possibly took advice about how to get a host for the server. The complaint conceals where the server was found - perhaps for this reason.
3. Perhaps a foreign government surveillance programme routinely searches virtual servers, and this resulted in the server being found.
1. could you please try to briefly explain how would such traffic analysis work? People with deeper knowledge of networking always mention traffic analysis without further explanation... but to us curious amateurs this does not really explain anything.
The fact that they had the server image related to specific date - July 23rd - does this say anything about how they obtained the image? Can we infer that they did not locate or hack the real server (because they would continue to have access after that date) and that maybe somehow they got hands on some backup image or something? Can we infer anything else?
To understand how a traffic analysis attach works, you first need to understand how hidden services work.
The way Tor hidden services work is that the hidden service builds a route (called a circuit) through the network of Tor nodes through to a node which it calls a 'rendezvous point' (RP). It then repeats this process to establish multiple circuits, one to each RP. The IPs of the RPs are public information, but no node (not even the RP) except the hidden service knows the route the circuit takes to the RP - only how to get to the next hop. End-users build their own circuit with Tor to the RP, so the RP or the hidden service also don't know the identity of the end-user.
An attacker wants to find out the route a circuit takes from an RP to the private node, to find the private node.
Tor does not send masking traffic - it only sends data when someone is actually asking something. So when someone sends data through the RP, the RP receives that data, unwraps one layer of the 'onion' packet and sends the rest along the circuit. The next node does the same. Therefore, this creates a pattern that is observable to someone who is sniffing the networks.
An attacker with access to all the networks between the nodes cannot see the contents of the packets, since it is encrypted, but they can look for the pattern in the traffic to see where the traffic is flowing. If no one was using Tor at all except for one packet being sent along one circuit, it would be trivial to see the path a packet follows, since you would see the RP sending the packet, then a slightly smaller packet going to the next node, and so on to the destination. In practice, lots of people are using Tor, so you need to do statistical analysis and build up confidence over time about a circuit.
If there is an unusual amount of data traveling over the circuit you care about (e.g. due to a DDOS), this again makes it easier - just look for all the most active nodes. If you can send a pulse of exceptionally high traffic, you can watch the time evolution of the pulse through the system and see where it ends up. Even without a high traffic pulse, you can compute probabilities that observed traffic outputs from a node were caused by a given input (based on the timing, size, etc...) and over time, combine a lot of individual points of weak evidence to get strong evidence about a circuit.
The complaint says that the 23rd was when they asked a foreign government to assist them by imaging the server - so they know where it is physically, and who owns the datacentre.
46 comments
[ 2.6 ms ] story [ 91.2 ms ] threadWhere's the rest of the fortune?
Why is the Cayman Islands, with a population of 56,000, the 5th largest financial center in the world?
Is the implication supposed to be that seizure of Bitcoins is protected by the Fifth Amendment? That seems far-fetched.
The case for not turning over passwords for bitcoin wallets is probably stronger than for true crypt volumes.
If they arrest you with an encrypted volume on your laptop, they know its your's. The only question is what it is. In a lot of cases, the government knows what it is, they just need to be able to prove it. Courts call this a forgone conclusion and can compel you do use your decryption password( but not tell it to the cops).[0]
On the other hand, handing over bitcoin keys maybe the only way they can learn which bitcoins are yours. This is,arguably, not a forgone conclusion at all. And again, you are not just giving them access to the data, you are effectively being compelled to testify that it is your data.
[0]http://www.slate.com/articles/technology/future_tense/2012/0...
Mmm, good point. Although considering bitcoin is a public network, they could have linked a certain address to you already, ala "follow the money"
They may have evidence that this is your wallet, but by supplying them with the password, you are testifying that it is your wallet.
"An act is testimonial when the act entails implicit statements of fact, such as admitting that evidence exists, is authentic, or is within a suspect's control."
"By entering the password Boucher would be disclosing the fact that he knows the password and has control over the files on drive Z."
Compelling Ulbricht to unlock the wallet would be compelling him to testify that the wallet is his. You may think this is a silly chain of reasoning, but I'm sure you will forgive me if I will continue to side with existing federal precedent. http://www.volokh.com/files/Boucher.pdf
In Ulbricht's case there is a compelling argument that the bitcoin's contained in the wallet on his computer(assuming it was on his computer and not the server) are his. The question is what does the government know about those coins. Clearly, if they know Ulbricht's bitcoin id/public key, then they can look up his transactions in the block chain. If they don't, I'm not sure what they could possibly know.
Of course, given Ulbricht's apparently atrocious OPSEC, it seems likely his password is Ludwig von Mises and the FBI won't need his help at all.
[0]http://en.wikipedia.org/wiki/In_re_Boucher
If the feds took an encrypted wallet file and are preventing the accused from accessing it, but haven't decrypted it, whether it's "seized" seems somewhat a matter of semantics.
(IANAL and I have no idea whether this meshes with the actual legal state of things in the US.)
The seizing authority in general has no way to tell in advance if such a co-collaborator exists, and assuming the transfer is to a new address with no historical transactions, it would be extremely difficult to identify who the collaborator was after the fact.
1. Taking control of them
2. Removing control of them from anyone else
The second point is specific to a digital asset which can be duplicated (in this case we are also talking about any passwords).
Bitcoin, by definition, allows the second point, but requires a transaction. So the FBI should have made a transaction immediately (or have the means of invalidating any transaction of those bitcoin).
This is novel and critical to the story, but the Guardian article seemed to tangentially miss it.
Are denial of service attacks the answer? DDOS a site and just ask popular hosting providers if they're experiencing any unusual traffic levels?
Seems like such an attack always precedes a takedown.
It is possible that DDOS attacks are not actively initiated by the NSA, but that they greatly help them complete their traffic correlation analysis network to an acceptable level of certainty by lighting up the circuits from the rendezvous points to the hidden service.
If there is stronger evidence that they could have followed from the person to the server, they would probably have put that in the complaint as it would be more direct and therefore more incriminating.
Therefore, it seems likely that they didn't find the server by following it from the person. There are several possibilities here:
1. Perhaps the NSA helped with finding the server by traffic analysis to follow the circuit back to the hidden service IP?
2. Perhaps the server was rented from / on the advice of an informant and setup only shortly before it was imaged. Reading between the lines, redandwhite was a law enforcement plant on Silk Road (possibly Canadian, given that is what the article claims) that DPR trusted to some extent and possibly took advice about how to get a host for the server. The complaint conceals where the server was found - perhaps for this reason.
3. Perhaps a foreign government surveillance programme routinely searches virtual servers, and this resulted in the server being found.
The fact that they had the server image related to specific date - July 23rd - does this say anything about how they obtained the image? Can we infer that they did not locate or hack the real server (because they would continue to have access after that date) and that maybe somehow they got hands on some backup image or something? Can we infer anything else?
The way Tor hidden services work is that the hidden service builds a route (called a circuit) through the network of Tor nodes through to a node which it calls a 'rendezvous point' (RP). It then repeats this process to establish multiple circuits, one to each RP. The IPs of the RPs are public information, but no node (not even the RP) except the hidden service knows the route the circuit takes to the RP - only how to get to the next hop. End-users build their own circuit with Tor to the RP, so the RP or the hidden service also don't know the identity of the end-user.
An attacker wants to find out the route a circuit takes from an RP to the private node, to find the private node.
Tor does not send masking traffic - it only sends data when someone is actually asking something. So when someone sends data through the RP, the RP receives that data, unwraps one layer of the 'onion' packet and sends the rest along the circuit. The next node does the same. Therefore, this creates a pattern that is observable to someone who is sniffing the networks.
An attacker with access to all the networks between the nodes cannot see the contents of the packets, since it is encrypted, but they can look for the pattern in the traffic to see where the traffic is flowing. If no one was using Tor at all except for one packet being sent along one circuit, it would be trivial to see the path a packet follows, since you would see the RP sending the packet, then a slightly smaller packet going to the next node, and so on to the destination. In practice, lots of people are using Tor, so you need to do statistical analysis and build up confidence over time about a circuit.
If there is an unusual amount of data traveling over the circuit you care about (e.g. due to a DDOS), this again makes it easier - just look for all the most active nodes. If you can send a pulse of exceptionally high traffic, you can watch the time evolution of the pulse through the system and see where it ends up. Even without a high traffic pulse, you can compute probabilities that observed traffic outputs from a node were caused by a given input (based on the timing, size, etc...) and over time, combine a lot of individual points of weak evidence to get strong evidence about a circuit.
The complaint says that the 23rd was when they asked a foreign government to assist them by imaging the server - so they know where it is physically, and who owns the datacentre.
1. Ross posted online using his real email address, which was found with subpoenas
2. The Silk Road server had a public IP address used for server admin