Others have summarized it well but I’ll add my perspective also.
I2P is a decentralized network of peers. All traffic gets routed through multiple peers before reaching its destination. Traffic is also encrypted by multiple layers of encryption and each connected peer can only decrypt one layer, that layer will only contain further routing info so that peer knows the next place to hand off your data. The encryption gets stripped layer by layer until it finally reaches its destination.
What this ultimately means is that by interacting with a website or service through I2P it is virtually impossible to identify any information the user is sending or receiving and it is also impossible to tell where the connections are going or coming from.
To make things even more interesting all I2P routers by default also contribute resources back into the network so while your I2P Router is handling your communication connections, it’s also volunteering to be a connection node in someone else’s connection. This adds further security because now you’ve got many Peer to Peer connections going in and out of your network, all encrypted so any prying eyes will have an exceedingly difficult time trying to make sense of any of your internet traffic.
Tor relies on the good faith of its community to contribute resources to the network and it’s not very well incentivised causing its resources to be far more limited and bogged down. For this reason, Tor cannot sustain heavy torrent traffic without easily being overwhelmed. In I2P, every user is a contributor so the more people who use I2P, the faster and better it becomes.
The big advantage Tor has over I2P are outproxies and the beautifully prepacked Tor Browser. Tor has a lot more influence and money backing it so there are some large and well protected entities that can afford all the legal trouble that outproxies can bring. Unfortunately there just isn’t enough money or legal support in the I2P community to reliably support outproxies even though I2P already has full support for it.
I2P is a fantastic tool for private communication across the Internet and the true ELI5 here is I2P natively supports “anonymous” torrenting (even encouraged it as it strengthens the network further) and will do so privately and securely without any need for a VPN. Adding support to QBittorrent makes it even easier to access I2P torrenting with very minimal set up required.
The only catch here is you can’t go around downloading any old torrent from the Internet on I2P, someone needs to actually be seeding that torrent on the I2P network for you to get any data. There are fully functioning tracker sites exclusively within I2P that have a huge catalog of content but all is not lost for “clearnet” torrents either. Software like BiglyBT and now QBittorrent, allows users to “bridge” or “cross seed” torrents across the two networks, that way content is shared no matter what network you’re a part of.
BiglyBT has been doing this for a while now but I’m so happy to see QBittorrent finally embracing this as well.
To summarise, I2P is similar to Tor, except that every client also serves as a node, and there are no exit nodes, so you can only access data shared by other I2P clients
Took a quick look at the first few messages and the links: seems like BiglyBT is banned by a lot of private trackers because it’s possible to mod it to spoof the numbers required to stay a member in the private tracker, while also being able to create a torrent file that allows others using the mod to utilize the private tracker without permission. Not sure if any of that functionality has to do with I2P.
From my experience, some popular I2P torrents have gotten up to 1 MB/s download, but I usually average around 200 KB/s. While it is not blazing fast, it does provide a good deal of anonymity for everyone involved with the torrent.
Also, you can lower the anonymity and increase speeds by reducing the number of hops from 3 to 2 or 1. You can choose how “anonymous” you’d like to be while torrenting, at the cost of speed.
What stops governments/LEO/copyright dragons from spinning up thousands of the fastest/most accessable i2p nodes so that clients connect to them first, then these hosts log the traffic paths to identify origin/destination?
All traffic gets routed through multiple peers before reaching its destination
I wonder why the whole internet is not designed like this.
if we designed basic protocols like TCP and UDP with user privacy (in practice hiding his IP-Adress) as the most important point we would have a more secure internet I currently think… or am I wrong here?
Others have summarized it well but I’ll add my perspective also.
I2P is a decentralized network of peers. All traffic gets routed through multiple peers before reaching its destination. Traffic is also encrypted by multiple layers of encryption and each connected peer can only decrypt one layer, that layer will only contain further routing info so that peer knows the next place to hand off your data. The encryption gets stripped layer by layer until it finally reaches its destination.
What this ultimately means is that by interacting with a website or service through I2P it is virtually impossible to identify any information the user is sending or receiving and it is also impossible to tell where the connections are going or coming from.
To make things even more interesting all I2P routers by default also contribute resources back into the network so while your I2P Router is handling your communication connections, it’s also volunteering to be a connection node in someone else’s connection. This adds further security because now you’ve got many Peer to Peer connections going in and out of your network, all encrypted so any prying eyes will have an exceedingly difficult time trying to make sense of any of your internet traffic.
Tor relies on the good faith of its community to contribute resources to the network and it’s not very well incentivised causing its resources to be far more limited and bogged down. For this reason, Tor cannot sustain heavy torrent traffic without easily being overwhelmed. In I2P, every user is a contributor so the more people who use I2P, the faster and better it becomes.
The big advantage Tor has over I2P are outproxies and the beautifully prepacked Tor Browser. Tor has a lot more influence and money backing it so there are some large and well protected entities that can afford all the legal trouble that outproxies can bring. Unfortunately there just isn’t enough money or legal support in the I2P community to reliably support outproxies even though I2P already has full support for it.
I2P is a fantastic tool for private communication across the Internet and the true ELI5 here is I2P natively supports “anonymous” torrenting (even encouraged it as it strengthens the network further) and will do so privately and securely without any need for a VPN. Adding support to QBittorrent makes it even easier to access I2P torrenting with very minimal set up required.
The only catch here is you can’t go around downloading any old torrent from the Internet on I2P, someone needs to actually be seeding that torrent on the I2P network for you to get any data. There are fully functioning tracker sites exclusively within I2P that have a huge catalog of content but all is not lost for “clearnet” torrents either. Software like BiglyBT and now QBittorrent, allows users to “bridge” or “cross seed” torrents across the two networks, that way content is shared no matter what network you’re a part of.
BiglyBT has been doing this for a while now but I’m so happy to see QBittorrent finally embracing this as well.
EDIT: https://geti2p.net/en/
So its basically the cooler Tor? Thanks for explaining
Cooler and less cool because no exit nodes. It’s fully separate from normal internet.
There are exit nodes, called out proxies but routers have to set them up themselves
The most known outproxy: https://stormycloud.org/i2p-outproxy/
To summarise, I2P is similar to Tor, except that every client also serves as a node, and there are no exit nodes, so you can only access data shared by other I2P clients
Is my summary correct?
You can have exit nodes, called out proxies on I2P they just have to be manually setup
So it’s like Kazaa, kinda? Are we basically going back to the decentralised P2P days? :-D
Most private trackers ban BiglyBT, hope the same doesn’t happen to QBittorrent
@eya
Why do they ban it?
Issue #1551
That links to a link to a forum with a ton of replies? Can you just copy paste the reason?
Took a quick look at the first few messages and the links: seems like BiglyBT is banned by a lot of private trackers because it’s possible to mod it to spoof the numbers required to stay a member in the private tracker, while also being able to create a torrent file that allows others using the mod to utilize the private tracker without permission. Not sure if any of that functionality has to do with I2P.
Would that not be possible in every other client? Or is it just the easiest to do with BiglyBT?
You can mod any open source client to cheat but most of them don’t give you buttons to do it.
Thanks for such a detailed answer! How does the I2P speeds compare to running torrents over VPN? I assume its a lot slower?
I2P is separate from normal internet. It’s not an alternative to VPN.
Yeah thats fine, I was just wondering how the speed varied from one solution to the other.
From my experience, some popular I2P torrents have gotten up to 1 MB/s download, but I usually average around 200 KB/s. While it is not blazing fast, it does provide a good deal of anonymity for everyone involved with the torrent.
Also, you can lower the anonymity and increase speeds by reducing the number of hops from 3 to 2 or 1. You can choose how “anonymous” you’d like to be while torrenting, at the cost of speed.
Very cool, thanks for sharing.
How are nodes chosen by the client?
What stops governments/LEO/copyright dragons from spinning up thousands of the fastest/most accessable i2p nodes so that clients connect to them first, then these hosts log the traffic paths to identify origin/destination?
deleted by creator
https://geti2p.net/en/docs/how/threat-model#sybil
Good write-up from the I2P team on this topic. Page includes other attack vectors as well.