1234

Damn spam bots (Read 1841 times)

xor


    Eric mentions the "forum admins" above for reporting spam.  These are:

     

    • zoom-zoom
    • Wingz
    • Trent
    • Neil Gunn
    • jlynnbob
    • NaderAlfie
    • Purdey
    • Run To Win
    • MilkTruck (based on what I saw this morning)

     

    This does not include the administrators and owners of User Groups.  But they (currently) don't get spam.

     

    This is covered in Trent's "Spam and RunningAhead" thread which is in Off the Beaten Path for some reason.

     

    Eric: can we have a sticky thread and/or a page somewhere that clearly lists "how to report spam, and to whom?" 

     

      But you could run the analysis as a low-priority background task, right?

       

      Here's what I'm thinking... Comparing posts won't do because humans tend to modify each spam slightly to blend in with the topic.  I could compare two posts for similarities and if above a certain percentage, they would be considered the same.  Even that is not fool proof.

       

      Instead of comparing entire posts, we can focus on the urls.  Every spam will contain urls.  If each post has more than x urls, and if the urls are similar across posts, then we can safely assume they're spam.

       

      I know how to do all this.  I'm just not sure how much resource intensive it would be, even if it's done in the background.  Also not sure what can be done if spammers switch to post one spam per account...

      xor


        RWOL took the url approach w/ spam too... the problem there was that from the user side, the rule set seemed nondeterministic (it was not... it's just that we mere users couldn't figure out what the hell was going on).  Posts with url links might get posted or they might get tossed.  Aside from a couple obvious things (like links to letsrun, which always got a post deleted and, uh, not for spam reasons), it was never clear which urls would be allowed and which ones would be caught as potential spam.  Even questions to the mod were met with "how the hell do I know".

         

        Most of the spam I see is not multiple links, fwiw.

         

          Urls are the only commonality amongst spams.  The spam check needs to take time into consideration.  Posts created within x seconds of each other, with urls pointing to the same domains would be flagged as spam for review.

            Urls are the only commonality amongst spams.  The spam check needs to take time into consideration.  Posts created within x seconds of each other, with urls pointing to the same domains would be flagged as spam for review.

             

            I bet they have bunches of alternate urls as well, including multiple domains.  This problem might require some serious AI, or limiting all users to one post per day.  On the bright side, we would be motivated to make it a good one. Smile

            Well at least someone here is making relevance to the subject.

              Eric: can we have a sticky thread and/or a page somewhere that clearly lists "how to report spam, and to whom?" 

               +1  yes, that would be helpful.

              2014 Goals:

              #1: Do what I can do. <DOING>

              #2: 365 Hours training <NOPE, INJURED>

               

                Eric: can we have a sticky thread and/or a page somewhere that clearly lists "how to report spam, and to whom?" 

                 

                Would a forum help or forum faq page be useful?  How many times have you had to explain how to insert a picture, or to upload one's avatar?

                Well at least someone here is making relevance to the subject.


                Brains aint me forte

                  I don't have anything constructive to add to this thread but to say thanks for working so hard to keep the spammers out.

                   

                  I also have to say that reading all this gave me a serious nerdgasm.  I had a change in career about 8 months ago and I don't get to talk tech much anymore.

                   "Address the process rather than the outcome.
                  Then, the outcome becomes more likely." - Robert Fripp


                  Samurai in Training

                    I've always felt trying to use deterministic methods to stop spam posts in a forum or social environment isn't resolving the root cause.  The root cause is that accounts are being created to start with.

                     

                    CAPTCHAS work for a while, challenge questions work for a while, honey pots work for a while.  It isn't that they don't work, it is they don't work for long.  Bla, bla, bla, bla...

                     

                    Why not have an automated approach to rotating out the method used when creating an account.  Here is how I would go about it...

                     

                    • Unique URL for each account creation session.  Probably easiest if you did it with a unique bogus subdomain. i.e. 4623452xxx.runn...com/newuser or something like it.  This URL can timeout after a designated time.

                    • Randomly apply one of the meriod options out there for spam bots.  CAPTCHA, challenge question, honey pot, timed button (liked that one), etc, etc, etc.   As a new technique becomes available you can simply add it to the toolchest.

                      It's a lot harder if the spam is created by brigades of human operators in China.

                      Well at least someone here is making relevance to the subject.

                        It's a lot harder if the spam is created by brigades of human operators in China.

                         

                        I'm almost certain that's the case with the fullmalls.com spam hitting the site right now. A business news site I edited back in 2010-2011 had a major problem with the same spammers. Their willingness to do email validation made them incredibly hard to suppress...and I don't think we ever succeeded.

                          Oriwoi just laid down a bunch of spam.

                          Age: 46 Weight: 200 Height: 6'2" (Goal weight 195)

                          Current PR's:  Mara 3:48:09; HM 1:43:26; 10K 43:59; 5K 21:27


                          Muddling through

                            Oriwoi just laid down a bunch of spam.

                             

                            That's been reported via the feedback option. 

                            2014 Goals: Run first trail ultra, first 100K, and see what I can do in a 24-Hour race

                              damn chinese spammers

                                It looks like the spammer has automated the spamming of the recent topics list.  I have put in some spam detection code that should thwart the spamming temporarily.  I am also adding spam detection logic in the upcoming release, scheduled for the next couple of weeks.  Let's hope the spammer hasn't found a way around until then.

                                1234