views:

29

answers:

2

Heya guys, im currently working on a system that's based around the concept of Digg where as you have a button on an external site and a user who clicks that button gets to post / digg it.

Anyways, Im looking for some information on the types of spam that I will expect as time grows on.

So far I have thought about a fair few factors, such as flag system, X rates by different users before frontage, Domain Monitoring, and some others..

Any ideas on what money hungry site owners would try and attempt ?

Thanks in advanced Robert.

+1  A: 

I expect creating a system like digg's, you will be prone to problems like:

  • Voting from bots to get items higher (prevent with captcha, time delay between vote ups, karma system)
  • Organised groups of users voting eachothers stuff up (to combat this you can try comparing submissions with votes from other users. If User A always votes User B's content up, and vice versa, you have a fixed result)

Plus the more common problems - mass account creations, mass automated submissions, spam being added. These are combatted using the systems you have mentioned, plus a bayesian spam filter.

danielgwood
in regards to the Bots, Im doing a 1 vote per fingerprint system like SO, captcha depening on violations broken over a given time range, in regards to the fixed result, the one fingerprint per vote system should take care of that!
RobertPitt
You could also try trusted users - though this requires an active community or a lot of input from yourself - who carry more weight in your voting/submission approval algorithm.Be careful though, you don't want to end up like Digg, where the overbearing majority of top-rated posts come from a minority of users.
danielgwood
+1  A: 

Instead of telling you what you should watch out for, here's a more proactive approach: take a look at Akismet, CAPTCHA, DNSBL checks, and bandwidth throttling. All these can be done in PHP. This way you can start real-coding and not vapor-coding.

stillstanding
witch is the best DNSBL provider ? and thanks for the info on DNSBL
RobertPitt