TBDev Forum
Programming board
Main Page Forum Rules Privacy Policy Members Contact Us
Welcome Guest (Login | Register)
TBDev Forum TBDev.net General Chit Chat and Feedback Questiion about Mods and Cheating

Questiion about Mods and Cheating

2006-05-16 18:23:00
Group: Member
Posts: 30
Topics participated: 20
Topics started: 3
First post: 2006-03-11 10:30:00
Last post: 2006-06-03 15:19:00
I've successfully installed browser protection mod and announce waittime mod. But I could not install abnormal uploads mod successfully so I skipped for the time being :(. Is it still possible for users to cheat?
 
2006-05-16 19:01:00
Group: Member
Posts: 2312
Topics participated: 1240
Topics started: 124
First post: 2006-02-25 21:12:00
Last post: 2009-01-03 14:56:00
Uh huh.
 
2006-05-19 22:24:00
Group: Member
Posts: 93
Topics participated: 48
Topics started: 10
First post: 2006-03-30 22:48:00
Last post: 2007-04-10 04:35:00
If you need help installing that mode, i can help you,,i am using this mod in my tracker and works good.
 
2006-05-19 22:40:00
Group: Member
Posts: 461
Topics participated: 232
Topics started: 10
First post: 2006-03-21 13:50:00
Last post: 2008-12-23 18:24:00
Members will always be able to cheat, due to the insecurity of the method used to update an announce (query string).

All we can do is highlight possible faked uploads.

Upload Scanner is one method, and the snatched list is another.

There are a number of ways of cheating. Sending a fake upload is just one.

I've been talking with a few other people about how to detect incorrect uploads, and one system that may work, would be a client reporting system.

It works like this...

An upgraded BT standard that sends two sets of data to an announce. The first data is the actual stats for the torrent in question, and the second set of data is the status of all peers connected to that client.

On a torrent with no cheaters, the second set of data would verify that the stats returned by each peer is genuine (give or take a small margin of error).

Introduce a cheater to the swarm, and all the other peers would report that clients' actual status (% complete and %uploaded to other peers), which would differ from what the actual client was reporting.

The second set of data could be sent as a bencoded string, ready for bdecoding and comparing with the current peers on that torrent.

Considering that an announce supplies each peer with a bencoded string of IPs ready for connection, the return data would not be much of a drain and could be processed inline, or during a regular interval (cleanup).

This system could even work using a mix of pre-reporting clients and reporting clients, although the fewer of the newer clients, the less acurate the results.

This method would depend on someone writing a client that would report peer stats to the announce. So far, many BT client writers are not really interested in the problem of cheaters, prefering to optimise and enhance the code purely for distribution purposes.
 
2006-05-20 01:14:00
Group: Member
Posts: 1378
Topics participated: 601
Topics started: 11
First post: 2006-02-25 01:34:00
Last post: 2008-12-22 14:17:00
I dun see ppl dropping their favorite client for a possibly buggier client.

I have thought of different ways.
of stopping the cheaters, which all follow current bit torrent standards.

1) Give the leecher, 1 seed to connect to for approx 30 minutes.
if it's a cheater, they will report an amount uploaded, which the system
can ban and lock out the user. :) Simple.

the other method helps block out the lame ratio apps, and the browser hack.

2) Using current bt protocol, assign a user a tracker id, upon each announce, assign a new tracker id. the users client should report back the current tracker id.
with a ratio cheating program, this technique will catch them, as they can't change the tracker id of thier client, as it's assigned by on each announce.

It wud be so much nicer if leechers reported what they downloaded and from who. instead of what they uploaded. this may cause a strain on the tracker tho, as more info is needed to be processed.
 
2006-05-20 01:47:00
Group: Member
Posts: 461
Topics participated: 232
Topics started: 10
First post: 2006-03-21 13:50:00
Last post: 2008-12-23 18:24:00
Well, my method could be introduced by the client vendors themselves, into the stable versions of their client. Also, Azureus could have a plugin made.

Imagine an announce line like such...

hxxp://tracker.com/announce.php?
passkey=a4316d5ffbb0456ffdfc2e9ba9393fff
&info_hash=P%FE%FF%2Cp%FFm%FFS%19%AD%CF%D4%0D%A7%08qw%CAr
&peer_id=-AZ2203-zXGlZLxlrPmZ
&port=54672
&uploaded=542147
&downloaded=44171264
&left=323808395
&numwant=146
&no_peer_id=1
&compact=1
&key=01pSlBAf
&bencpeers=

The first (event=started) announce would not send a bencpeers, but the second, all the way to the last, would send a bencpeers which can then be compared with the peers list to highlight any deviation.

Essentially, all peers would be acting as torrent police, reporting what they see others do.

Now say a peer 255.255.204.204 is cheating, by reporting false stats. He has hacked his client to make him look a seeder, when in fact, he is still leeching. His peer will report completion, yet the other peers will report him as still leeching.

What about another one. When a member claims they only downloaded the sample. The peers list will still report what they have to offer, and if the percentage is greater than the sample it is supposed to represent, then we know they are cheating.

While not all peers are available to other peers, this system should give enough coverage to be able to pinpoint cheating, as long as it is combined with other checks.

I do like your idea of sending a tracker id with each announce. This would certainly rule out some cheat programs, but I believe more and more people are using Smashers for Azureus, and this is where the real problem lies.

My own copy of Smasher has revealed that this is quite resourceful to the knowledgeable cheater. There are still tell tale signs when specific functions of this mod have been applied, which the mod doesn't take into account, so even the expert cheaters can be caught.
 

Similar threads

Topic Posts Last post
Peers cheating? 6 2006-11-22 05:42:00
Ratio Cheating + IP Block 4 2006-07-27 15:19:00
Experimental Anti-Cheating 11 2007-10-27 14:59:00
mods 1 2007-04-10 19:22:00
all mods 3 2008-02-17 15:00:00