[RC5] Cheaters removed?
Floppus at Chello.nl
Fri Apr 22 08:41:54 EDT 2005
Firstly, only account were removed that sent in MAINLY false blocks, as in
not a few, but hundreds per day.
Secondly, We have 100% certainty these blocks have not been generated with
unaltered clients. Computers with computational errors could not have
generated what we have seen.
Thirdly, Accounts have not been deleted until a decent means of filtering
was created: future blocks from such clients are discarded automagically.
I hope you understand I cannot give any more details, as that would be like
pointing out to a burglar that the key is under the mat.
Filtering the accounts currently blocked would result in almost empty
accounts, but would put a lot of pressure on our servers plus probably a
stats outage of a few days. So everyone would have some inconvenience to
allow a few cheaters to continue where they left off before starting to
cheat: no thanks.
I'll say it once more: I have 100% certainty that all users blocked have
submitted significant numbers of invalid results. Impossible to generate by
accident. I also have 100% certainty nobody was blocked that didn't deserve
to be blocked.
Personally, I'd expected some positive sounds about this all.
Van: rc5-bounces at lists.distributed.net
[mailto:rc5-bounces at lists.distributed.net]Namens Andreas Gudian
Verzonden: vrijdag 22 april 2005 14:07
Aan: 'D.net Discussion'
Onderwerp: AW: [RC5] Cheaters removed?
> Let's suppose somebody is allowed to install d.net client on company's
> computers. Let's suppose this company has 1000 computers of
> which about 100
> are on 24h a day and other are used by workers. Let's suppose
> for simplicity
> all of those computers are using the same email address in their d.net
> configuration. Let's suppose 99.9% of the computers are
> working correctly.
> Yes, it means one computer makes computational errors. It may mean it
> returns d.net packets which are flawed. If those erros are
> detected - great!
> Discard those block (or even inform the user - I'm sure he'd
> be glad to be i
> nformed that one computer is faulty - he may even try to use
> kind of binary
> search to find out which one and replace it, as d.net is not
> the only thing
> that's likely to suffer).
> But don't discard all of the blocks and don't ban the user.
> That would be
> plain stupid.
> Slawek Piotrowski
I agree with that. You also pointed out some mails ago that it is possible
to "attack" accounts by sending cheat-client blocks using that accounts
email-address, only to let the account be deleted from stats. Actually,
somewhat a year ago when we kicked a cheater out of our team (we also had
him removed from stats), he threadened us to do exactly that, to "poisen"
our accounts with cheat-WUs. Just for revenge.
> I can assure you that is not how it's done :D
> For all users removed, there was VERY solid proof. Thorsten can verify
Floppus, how would the procedure be in the described case? Can you
selectively remove credited but invalid work from such accounts? I also
think about those big one-account subteams at DPC or BugTraq.ru, where one
jerk could destroy the whole work of many honest subteam-members.
rc5 mailing list
rc5 at lists.distributed.net
More information about the rc5