Time to switch to i2P?

[theblacksquid]

http://www.coindesk.com/day-reckoning-dark-markets-hundreds-illicit-domains/

With the US gov't reaching nout with its fat, well-funded arms out to squash out the TOR darknet network, maybe its time that we switch to i2P for a greater amount of anonymity?

[ruthless09]

I don't think this would have much of a difference of ability of the government being able to de-anonimize any site that is trying to keep their identity secret. 

[RappelzReborn]

What does  i2P refer to exactly ?

[shorena]

What does  i2P refer to exactly ?

 https://geti2p.net/en/

[kwukduck]

I don't think this would have much of a difference of ability of the government being able to de-anonimize any site that is trying to keep their identity secret. 

It does, to some extent, the security model of i2p is quite a bit more advanced than that of tor, i would say 'next level'.
The things that make me say toe is inherently insecure were addressed in i2p many years ago already.

There have been a few successful attacks against i2p services in the past, all very well documented, iirc none were to blame on i2p itself. (Please correct me if I'm wrong, it's been a while)

I2p is generally faster in regards to bandwidth, more resilient against hidden service attacks and has a much lower latency. I've ran a VoIP server on i2p for a while, which worked great, try that on tor... You'd probably even have to modify your software to be able to work with tor, not so the case with i2p.

One of the biggest 'issues' with i2p is that it's basically a closed darkness, unlike tor which provides 'synonymized' access to the normal web for most of it's (or at least that would be my guess) users. So the goal of the two are somewhat different. I did notice the outproxy on i2p was working again about a week ago, but that's not the networks main goal AFAIK.

Best thing about roughly 90% less trolls than tor.

[theymos]

I2P is very similar to Tor technologically. If the Feds are using technical attacks against Tor, then the same attacks will probably also work against I2P. In fact, some attacks are easier against I2P because it has far fewer users and its network isn't carefully managed in the same way that Tor's network is.

I don't believe that any anonymity network in existence today is safe enough to directly run an illegal website on, unfortunately.

[Dabs]

Did those onion sites get attacked technologically? Or did they get attacked because of user/admin error?

What could they have done different that would have prevented discovery?

[Soros Shorts]

I don't believe that any anonymity network in existence today is safe enough to directly run an illegal website on, unfortunately.

You'd definitely need multiple layers of protection, and not just technological. If it were me, at the very least I would also like to have some passive eyes in the physical environment hosting the hidden services. E.g. an employee at the hosting provider in my payroll, preferably someone in security or compliance roles, whose task would be to discretely inform me if the feds came and started poking around.

[justusranvier]

I2P is very similar to Tor technologically. If the Feds are using technical attacks against Tor, then the same attacks will probably also work against I2P. In fact, some attacks are easier against I2P because it has far fewer users and its network isn't carefully managed in the same way that Tor's network is.

What I2P has going for it is a better theoretical basis, and a focus on hidden services rather than proxying to the clearnet.

The only thing Tor has going for it is more users - if I2P has the same number of nodes I'd expect their hidden services to be more secure than Tor's hidden services.

[santaClause]

I don't believe that any anonymity network in existence today is safe enough to directly run an illegal website on, unfortunately.

You'd definitely need multiple layers of protection, and not just technological. If it were me, at the very least I would also like to have some passive eyes in the physical environment hosting the hidden services. E.g. an employee at the hosting provider in my payroll, preferably someone in security or compliance roles, whose task would be to discretely inform me if the feds came and started poking around.

The problem with having "eyes" in the physical environment is that this will expose your identity somewhat as well as the fact that you are hosing something that is illegal when the hosing provider may not otherwise notice the illegality of what you are hosing. Plus you would need to trust the person you are using as your "eyes" 

[Welsh]

I2P just requires a more nodes to be up and running, that would mean it should be more secure than Tor, but of course if you want more nodes more people have to take the plunge initially when it's not so secure. Although, it's still going to be vulnerable to the attacks which Tor has undergone the past few years. Although, I believe it can be more secure than Tor with more people creating nodes.

At the moment Tor is probably more secure due to more nodes, but I2P why not be in the main scope at the moment because of the lack of users, thus this would make it more secure because of the lack of attacks, although there have been quite large attacks on both.

[ThiagoCMC]

You guys needs to know more about CJDNS and Hyperboria.   

[BTCmoons]

Did those onion sites get attacked technologically? Or did they get attacked because of user/admin error?

What could they have done different that would have prevented discovery?

Well both attacks by the government on both SR's were user/admin error (although SR2 made much worse errors then SR1).

I think the main issue is somewhat technological as there are very few large onion sites so any kind of DDoS attack on an onion site would make it easy for anyone with the ability to monitor overall tor traffic to be able to see where a lot of tor traffic is going when they are getting DDoS'ed

[theymos]

The main problem with I2P and Tor is that they only try to protect you against mostly-passive attackers who have absolutely no idea of where you might actually be on the Internet. The Tor threat model says (and this is also true of I2P):

By observing both ends, passive attackers can confirm a suspicion that Alice is talking to Bob if the timing and volume patterns of the traffic on the connection are distinct enough; active attackers can induce timing signatures on the traffic to force distinct patterns. Rather than focusing on these traffic confirmation attacks, we aim to prevent traffic analysis attacks, where the adversary uses traffic patterns to learn which points in the network he should attack.

But attackers looking for the real IP of a target hidden service can significantly narrow the set of possible targets by enumerating all active Tor/I2P users (using widespread traffic analysis or by having a lot of nodes on the network), and then they can further narrow it by doing intersection attacks. Once they've narrowed it down to a few hundred possibilities, they can try timing attacks against each one to get solid proof that they're the target.

(I wonder if the hidden services that were not taken down in the recent bust have anything in common. Are they in a particular country that's unfriendly to NSA demands? Do they use a fixed set of trusted entry guards? Probably we won't find out, unfortunately.)

I just don't think that low-latency client<->server networks can be secure. What we need are distributed data stores like Freenet so that the originator/owner of content doesn't need to always be online and moreover has plausible deniability even if they are under active surveillance. However, I really doubt that any existing anonymous data store could actually stand up to targeted traffic analysis of the content originator. Freenet seems to be put together in an especially haphazard way, without much theoretical basis for its claimed anonymity.

I like a lot of what I've read about GNUnet. I think that a good path forward for anonymous networks would be:
- Make the GNUnet software user-friendly.
- Create message board and Web functionality (like FProxy) on top of GNUnet.
- Make GNUnet work over I2P.
- Increase the popularity of GNUnet+I2P so that attackers can't just do traffic analysis of every single user.

[justusranvier]

The main problem with I2P and Tor is that they only try to protect you against mostly-passive attackers who have absolutely no idea of where you might actually be on the Internet. The Tor threat model says (and this is also true of I2P):

By observing both ends, passive attackers can confirm a suspicion that Alice is talking to Bob if the timing and volume patterns of the traffic on the connection are distinct enough; active attackers can induce timing signatures on the traffic to force distinct patterns. Rather than focusing on these traffic confirmation attacks, we aim to prevent traffic analysis attacks, where the adversary uses traffic patterns to learn which points in the network he should attack.

But attackers looking for the real IP of a target hidden service can significantly narrow the set of possible targets by enumerating all active Tor/I2P users (using widespread traffic analysis or by having a lot of nodes on the network), and then they can further narrow it by doing intersection attacks. Once they've narrowed it down to a few hundred possibilities, they can try timing attacks against each one to get solid proof that they're the target.

(I wonder if the hidden services that were not taken down in the recent bust have anything in common. Are they in a particular country that's unfriendly to NSA demands? Do they use a fixed set of trusted entry guards? Probably we won't find out, unfortunately.)

I just don't think that low-latency client<->server networks can be secure. What we need are distributed data stores like Freenet so that the originator/owner of content doesn't need to always be online and moreover has plausible deniability even if they are under active surveillance. However, I really doubt that any existing anonymous data store could actually stand up to targeted traffic analysis of the content originator. Freenet seems to be put together in an especially haphazard way, without much theoretical basis for its claimed anonymity.

I like a lot of what I've read about GNUnet. I think that a good path forward for anonymous networks would be:
- Make the GNUnet software user-friendly.
- Create message board and Web functionality (like FProxy) on top of GNUnet.
- Make GNUnet work over I2P.
- Increase the popularity of GNUnet+I2P so that attackers can't just do traffic analysis of every single user.

There's an solution to traffic pattern attacks - it's just really expensive.

They way you solve traffic pattern analysis is to make your protocol consume a constant amount of bandwidth all the time, regardless of whether anything is actually going on or not.

[bluemountain]

But attackers looking for the real IP of a target hidden service can significantly narrow the set of possible targets by enumerating all active Tor/I2P users (using widespread traffic analysis or by having a lot of nodes on the network), and then they can further narrow it by doing intersection attacks. Once they've narrowed it down to a few hundred possibilities, they can try timing attacks against each one to get solid proof that they're the target.

I would say a solution to this would be to have a lot more tor/onion sites that are legitimate and receive a lot of traffic. This would make a timing attack much more difficult as there would be more traffic to analyze which makes each data point less significant.

(I wonder if the hidden services that were not taken down in the recent bust have anything in common. Are they in a particular country that's unfriendly to NSA demands? Do they use a fixed set of trusted entry guards? Probably we won't find out, unfortunately.)

I am not 100% sure if this is technologically possible but maybe they were set up in a way so that only "x" percent of traffic will go to a specific server with each request being routed to a server at random. Another possibility is that whoever runs the sites that were not taken down were much better at fighting DDoS/timing attacks by shutting down/going offline whenever there is an increase in traffic above "x" percent.

I just don't think that low-latency client<->server networks can be secure. What we need are distributed data stores like Freenet so that the originator/owner of content doesn't need to always be online and moreover has plausible deniability even if they are under active surveillance. However, I really doubt that any existing anonymous data store could actually stand up to targeted traffic analysis of the content originator. Freenet seems to be put together in an especially haphazard way, without much theoretical basis for its claimed anonymity.

This sounds a lot like storJ to me

[inBitweTrust]

The great thing about TOR is it has a lot of support and infrastructure. This is also what makes it so dangerous as a certain percentage of exit nodes, bridges, and relays are controlled and owned by the NSA/GCHQ. What we need to do is increase the amount of high speed nodes and especially exit relays.

With security it is impossible to become 100% secure but you can certainly make it impractical and costly to attack.

Shutting down SR1 and SR2 was probably a very costly exercise and individually investigating dealers on a decentralized platform where all escrow funds were held in a mutisig account that couldn't be seized would be an effort in futility. The "war on drugs" is mostly funded by asset forfeiture and the theft of both the dealers and clients money. What happens when those funds are held in method where they cannot be taken?

[cellard]

I tried tor once and it was pretty boring so slow and it was a pain in the ass to search things so I gave up.
Is i2p faster and has it like a search page or you have to hop around sites?

[Dabs]

Well both attacks by the government on both SR's were user/admin error (although SR2 made much worse errors then SR1).

Ahhh.. Well, there you go.

SR3, don't host it in the United States. Morons. hehehe.

As for physical security, ... there are lots of methods, and although expensive you can host it yourself. Have any of the SR1 and SR2 operators seen "See More Buds" ? That talks a bit about physical security of grow houses.

I've never used SR1 or SR2 or any of the others that died, so I don't know if the user interface would be affected if I had done things differently or took care of things on my end, or set up shop in some remote mountain with walls like UBL (but UBL did not have internet, bummer.)

[inBitweTrust]

I tried tor once and it was pretty boring so slow and it was a pain in the ass to search things so I gave up.
Is i2p faster and has it like a search page or you have to hop around sites?

Tor is better at security just from the fact that there is more oversight, more development, and an order of magnitude more nodes. I2P has properties that make it better for torrenting files.

Here is some more info :
https://gnunet.org/sites/default/files/herrmann2011mt.pdf

If an attacker has enough of the network they can effectively DeAnonymize the user on Tor with the entry and exit nodes.
The solution is simply to grow the amount of node relays , but especially exit nodes as trusted ones are in short supply.