More detailed explaination from Gwren via email

[tsafa1]

This was posted by email in the mute forum by Gwren in response to my question.


Every time you are connected to other nodes you can do several things:
1) Finding ips of other nodes of the ants net
2) Downloading
3) Uploading
4) Routing
5) ...
Surely routing is the most important part because lets you remain anonymous, but also it is the most "expensive" part of the job. So the main idea is... as you can collect ips these ips could be used to create new connections if the olds are stale and also these new connections sometime (randomly) will take you nearer to the chunk source (surely some other time they will lead away from the sources)... so what? The implemented algorithm is stupid but effective: if you have, say, 5 connections your node continuously keeps track of which connections are actually active, active means you are uploading or downloading over those connections, not only routing. Basically this check lets you keep only a small number of unactive connections, that is connections aimed to routing, while the others MUST be active, otherwise, if a limit is passed, one (randomly) of the connections not active will be disabled and a new one will be created with another node of the net, this action is done at stated intervals of 15 minutes, untill the number of active connections remains under the limit.
This algorithm has two "side effects", on one side it takes you nearer to your sources, minimizing the hops number among your and the chunks "spring", on the other side it also takes the sources nearer to the requestors, because if you are not downloading anything and you are sharing some files, a determined number of connections will remain stable, but the remaining connections will constantly change untill they become active, that is untill some uploads become perfromed over those connections.

So what do we ackieve? A simple double effect... Muhammad goes to the mountain and the mountain gets nearer to him :P

[Hornet]

This new idea for Ants is great as it overcomes one its main problems, that is having to proxy files.

When a file is proxied over many hops a large amount of bandwidth is used for network traffic rather than own traffic so downloads are slow.

If you were to download directly then it would be fast but not anonymous as you would know the IP of the user you're downloading from.

By selectively connecting to users who upload/download your files you are increasing the probability of fewer hops or direct connection so less proxy traffic and faster downloads.

However, you still cannot tell if the IP you are connected to is proxing an upload/download or is the file's source/destination.

Once this new idea is developed, refined and tested the download speed and size of the Ants network should increase greatly.

Basically, the nework is learning what type of files you like and gradually moving your connections closer to users with those files. So that you end up in a cluster(s) with other file sharers that share your interests. This is very much like how Winny operates.

Lastly, re CJ's ban
Zeropaid is a pro file sharing site. Nothing stays the same there are always new applications and trends. Due to the actions of the MPAA and RIAA the current focus is on privacy, deniability and distributed storage.

Ants has all of those things.

• It can and does store ed2k links in a way that can't be closed down by the authorities.
• It does allow you to host websites anonymously and
• It does prevent the RIAA/MPAA from suing file sharers by giving users deniability.

Ants is a file sharing program with a forum on ZP so by flaming and consistenly negative posting you are doing the RIAA/MPAA's job for them. ZP is meant to be a haven for file sharers - where we can regroup and look to fight and win future battle despite recent set backs.

I do hope CJ comes back after her 4 day ban as I believe in free speech, pro or anti new developments.

I also hope ZP users try the new incarnation of Ants at some time in the future if only out of curiosity to see if CJ was right or wrong.

Hornet

[risingfoam]

This new idea for Ants is great as it overcomes one its main problems, that is having to proxy files.

When a file is proxied over many hops a large amount of bandwidth is used for network traffic rather than own traffic so downloads are slow.

If you were to download directly then it would be fast but not anonymous as you would know the IP of the user you're downloading from.

By selectively connecting to users who upload/download your files you are increasing the probability of fewer hops or direct connection so less proxy traffic and faster downloads.

However, you still cannot tell if the IP you are connected to is proxing an upload/download or is the file's source/destination.

Once this new idea is developed, refined and tested the download speed and size of the Ants network should increase greatly.

Basically, the nework is learning what type of files you like and gradually moving your connections closer to users with those files. So that you end up in a cluster(s) with other file sharers that share your interests. This is very much like how Winny operates.

Lastly, re CJ's ban
Zeropaid is a pro file sharing site. Nothing stays the same there are always new applications and trends. Due to the actions of the MPAA and RIAA the current focus is on privacy, deniability and distributed storage.

Ants has all of those things.

• It can and does store ed2k links in a way that can't be closed down by the authorities.
• It does allow you to host websites anonymously and
• It does prevent the RIAA/MPAA from suing file sharers by giving users deniability.

Ants is a file sharing program with a forum on ZP so by flaming and consistenly negative posting you are doing the RIAA/MPAA's job for them. ZP is meant to be a haven for file sharers - where we can regroup and look to fight and win future battle despite recent set backs.

I do hope CJ comes back after her 4 day ban as I believe in free speech, pro or anti new developments.

I also hope ZP users try the new incarnation of Ants at some time in the future if only out of curiosity to see if CJ was right or wrong.

Hornet

I dont have nothing against you hornet or ants, in fact this virtual hashing of proxie nodes is not a new idea, i just question certain aspects of the client.

i dont think it was right for someone to continue to spam and call others names just because they disagreed.

i believe in free speech too.
good luck, but like i said i will stick to my opinion and at the same time i am just as curious as everyone else here on the new p2p programs, with respect to how p2p programs will filter out and block riaa, mpaa etc.

the bottom line is i rather have facts and thats what i am about. matter of fact i believe that i able to learn new stuff to., that is why i like to research stuff to my fulliest, i enjoy doing this.

no harm done, like i said i never called u any names in fact like someone else has.
peace
and feel free to talk about ants etc on here, as i wont stand in you way.
in fact i am the one who suggested they make and ants and mute forum.

peace

[CactusChris]

Firstly - there have been several claims with no substantive backup that the ip address and the vip can be identified (cj's was "btw there is a way to obtain the ip address of the actual host sending the data, in fact there are several ways to do it and several tools to get this information.")
Now this is important !! we need to find out why they think this and what they think the answer is. We mostly believe that ANts is secure - so we are the worst arbiters of its security - there are none so blind etc...
When the comment is raised that ANts can be broken we need to do what Gwren did with his competition - and see what we can get out of the claimant - and then evaluate it impartially - for our own good.

Secondly
Now I think that the idea of disconnecting from peers that are just proxying data that is not for me - and randomly connecting to another peer instead has some merit. I think that what I select for download often only comes from a few users - most do not have the same taste in music so their available uls are of no interst to me (but are to others no doubt).

[notbob]

Firstly - there have been several claims with no substantive backup that the ip address and the vip can be identified (cj's was "btw there is a way to obtain the ip address of the actual host sending the data, in fact there are several ways to do it and several tools to get this information.")
Now this is important !! we need to find out why they think this and what they think the answer is. We mostly believe that ANts is secure - so we are the worst arbiters of its security - there are none so blind etc...
When the comment is raised that ANts can be broken we need to do what Gwren did with his competition - and see what we can get out of the claimant - and then evaluate it impartially - for our own good.

Secondly
Now I think that the idea of disconnecting from peers that are just proxying data that is not for me - and randomly connecting to another peer instead has some merit. I think that what I select for download often only comes from a few users - most do not have the same taste in music so their available uls are of no interst to me (but are to others no doubt).

how TCP works

http://www.ipanalyser.co.uk/content/tcp_packet.htm

in order to send and recieve packets, like a letter, each packet has a destination and a return address

once the packet is successfully sent, an ACK is sent which acknowleges the packet was recieved and that you don't need it sent again--this is protocol, and without these steps the system does not work efficiently (no source = no ack, which could mean the same packet is sent over and over)

with packet sniffers you can use the source (return address to track back to the proxy, and in some cases, all the way to the source)

p.s. don't get big headed about cj's ban--it had nothing to do with her views on your pitiful program, but instead from her habit of ressurrecting long dead threads and posting whimsical nonsense

[Hornet]

Very good post Chris.

Good point about CJ - I for one hope that identity comes back.

Now I think that the idea of disconnecting from peers that are just proxying data that is not for me - and randomly connecting to another peer instead has some merit. I think that what I select for download often only comes from a few users - most do not have the same taste in music so their available uls are of no interst to me (but are to others no doubt).

• Idea is not to disconnect form all peers that are just proxying data not for you - It is to disconnect from some - no more than 50% of such peers.
• ANts is an AI agent - the protocol will not make decisions like connecting you to users that share your taste in music. Instead it will try to learn where best to connect to so that you can upload and download with fewer hops. No taste involved :heart
• Currently files in ANts travel between 0 to 10 hops that one file using times 10 bandwidth very often. This slows download speed incredibly. So it makes sense to allow the ANts neightbour discovery protocol to reduce the number of hops. This is still at an early stage of development but it will become clearer as it develops.
• The purpose of ANts is to files share and yet be able to deny that you were the source of a particular file. Reducing the average number of hops taken by data does not impinge on that. You still cannot tell if a file travelled to/from your neighbour or if your neighbour proxied it. I think it evens randomizes download / upload "speeds" to prevent statistical analysis.
• I agree with your implication though that alot of development is still needed for ANts as it still has some memory leaks and CPU use problems. Also, it needs to increase its stability so that it can be run 24/7, 52 weeks of the year in the background as an alternative to the MPAA/RIAA infested internet.

Hornet

[Hornet]

how TCP works

http://www.ipanalyser.co.uk/content/tcp_packet.htm

in order to send and recieve packets, like a letter, each packet has a destination and a return address

once the packet is successfully sent, an ACK is sent which acknowleges the packet was recieved and that you don't need it sent again--this is protocol, and without these steps the system does not work efficiently (no source = no ack, which could mean the same packet is sent over and over)

with packet sniffers you can use the source (return address to track back to the proxy, and in some cases, all the way to the source)

notbob, don't get me wrong, I am not trying to be funny or question your judgement but could you explain further.

• Where would you put the packet sniffer(s) to discover the source of a file?
• Has this technique been used by the RIAA or MPAA?
• In ANts a file is often proxied 10 times accross the world would a packet sniffer have to be located at every ISP worldwide to trace the source (like in a Tom Clancey novel)?
• Are packet sniffers the tool which CJ was thinking of?
• Practically do you think encrypted proxy chains offer more privacy and security than unencrypted direct connections or are you speaking hypothetically stating that packets always leave a record as they are transferred from node. And with enough money, time and politcal co-opeeration you can always find the source of a file?

I am not trying to start an argument/flame war. I just want to understand what you actually think. The questions above just illustrate where I don't understand you.

Hornet

[notbob]

it isn't so much about the packet sniffing, but the packets you collect

lets say you have a encrypted packet--luckily, your client decrypts it for you--so anyone can use the client (or gut it out and take out the decrypting mechanism) and decrypt ants packets

now that they are not encrypted, (if ants encrypts, i think mute and waste do) they can be analyzed with other software to find where they came from (via the "return address") that address can be spoofed of course, but in that case it makes a lot of trouble for someone trying to exchange files

now say you are in law enforcement, and have the ability to do this at the ISP--it would be a simple matter to find who's sending what, and where

this is one of the things crackerjacker "discussed", though most people aren't fluent in cjese

[CactusChris]

Notbob thanks for the explanation but...
The way this proxying thing works is like a chain - the only ip (not vip) address in the message is the next peer in line - that ip is known by all neighbours - but not its vip or the ips it is connected to. At each ip stopover the address is stripped of and the next one inserted. This is what makes the system secure and is in the code for MUTE and ANts - or was 'cos I read it through (once).

However - correct me if i am wrong ;-))

[F0wler]

I'm sure this will give cj time to chill out.

[fnordprefect]

Also, I seriously doubt that all packets in the Ants network are encrypted with the same key. This is the only way that a malicious client could do what Notbob just said barring a brute force attack or a weakness in the encryption algorithm.

So Notbob, which is it:

Ants uses a weak encryption algorithm?
Ants encrypts all packets with the same key?
Ants encrypts packets in such a way that an attacker could brute force the keys?

[crackerjacker]

I'm sure this will give cj time to chill out.

I always be chilling. Was it good for you?
and for your information i am not banned.

[CactusChris]

Hornet said ->ANts is an AI agent - the protocol will not make decisions like connecting you to users that share your taste in music. Instead it will try to learn where best to connect to so that you can upload and download with fewer hops. No taste involved

Sorry - my bad wording - ANts does not know what I like - however what I like is often coming only from a few others who also like the same(ish) things - and not from the majority who have only things I do not like. So the consequence of Gwren's new idea is that I will find myself connected more closely to those peers who have files that i like (over a period of time).

Question - latest version has these mods in? I ran it overnight and was not impresssed (only on one test) on proxy/dl ratios.....I think it was 0.7.8

Regards
Chris

[Hornet]

I don't think that Gwren has fully implemented his idea yet as you are right ANts does seem to be a bit sickly since 0.7.8.

However, I do not fully understand what you mean.

Do you not want to be connected to peers that proxy files for you rather than peers that don't?

Sure the fact that all nodes will be doing this creates clusters but this should not affect your downloading experience :hi

Hornet

[cheapprick]

Why is tsafa1 banned?