Forum text blacklist to combat spam
Pages: 1 2
Andrew Hodgkinson (6) 465 posts |
Over time the ROOL site sees more and more spam. Community members help out and do an amazing job, but it’s a lot of work. We believe that a blacklist of some common URLs or phrases would help block spam posts. This has now been implemented and the blacklist will get populated over the next few days. There’s a small chance it might inadvertently block something legitimate – if so, I’m afraid you’ll lose your post, unless you have a browser where the “Back” button takes you back to the form with your post text still present. If you’re able to, though, please send the problematic text to info@riscosopen.org and we’ll see what we can do to amend the blacklist data. Thanks! |
Clive Semmens (2335) 3276 posts |
I’d post an enormous thumbs up emoticon if NetSurf would have it, but these words will have to do instead! |
Steve Pampling (1551) 8155 posts |
+1 I’m going to take a small guess and say subjects that are html links are on the list. Here’s hoping Dave H gets more free time. |
Rick Murray (539) 13806 posts |
Yeeehaw!!! 🙂
NetSurf can do that if you right click the button to open the follow up in a new window. If it’s good, then close the original window. Otherwise… ;-) |
Dave Higton (1515) 3497 posts |
I’m very much looking forward to using this blacklist. One way or another, we’ll fix the little bar stewards… |
GavinWraith (26) 1563 posts |
A blacklist has long been needed. I have got to the point where I don’t bother to open entries with daft names (no vowels, too many vowels, … ). But some spam-engines are probably clever enough to use more plausible names. |
Steve Pampling (1551) 8155 posts |
Gwynn. :)
I’m of the opinion that most of our recent influx is driven by 10 digit appendages controlled by grey cells. If the can’t post the words they want in subject (most common) or body then they will seek an easier target. |
Dave Higton (1515) 3497 posts |
Since that’s a Welsh name, it contains two vowels. I’ll raise you: Aoife :-) |
Steve Pampling (1551) 8155 posts |
Old joke about the Welsh not buying the extra letters to do vowels properly…
Exchange student1 working at CAMRA HQ I met at GBBF. Equivalent of my great aunts name ‘Eva’. I take it you’re short of things to delete – Yay! 1 From Tipperary, the place not the pub in my area – which is named after the song since the author lived thereabouts. |
Steve Pampling (1551) 8155 posts |
Apparently not, as the detritus in the General forum demonstrates. Bit of a tweak required. |
Dave Higton (1515) 3497 posts |
Give it time. The blacklist is starting from empty. |
John Williams (567) 768 posts |
May I suggest :// as a first item, then, for the subject field. |
Steve Pampling (1551) 8155 posts |
I couldn’t see in the CVS changes where the subject was picked out as distinct from the body. If it isn’t then dropping in a filter for all links means no links in post body as well. Distinguishing subject link references at least gets rid of the spammers who don’t care whether you read the body text because the subject line is the message |
Dave Higton (1515) 3497 posts |
I don’t want to discuss any details here of what is blacklisted, in case it helps the bar stewards. |
Andy S (2979) 504 posts |
Many thanks for implementing this! I agree the algorithm and keywords probably shouldn’t be discussed on the forum. |
Rick Murray (539) 13806 posts |
Actually, the algorithm is very simple. When a spam is sent to the server, the IP address of the sender is known. This is tossed to a special Google API that matches IP address to known activity (see, those tracking cookies come in useful after all). This known activity may lead to an actual identity. Oh, have I given away the secret? Oh crap… |
Clive Semmens (2335) 3276 posts |
It’s not a mallet, it’s a silver hammer. And the name’s Maxwell. |
Grahame Parish (436) 480 posts |
And it’s kept in Abbey Road? One thing I’ve noticed that is quite common to a lot of these spammers is the numeric suffix on the account name – often the same two digit number. |
Dave Higton (1515) 3497 posts |
Sorry, folks, the bar stewards have started to use link shortnening. This means that it is now blacklisted, which means no-one can post shortened links any more. |
Rick Murray (539) 13806 posts |
Probably a good idea, as who knows what evil can lurk behind shortened links… They’re only useful for size-restricted things like Twitter. |
Steve Pampling (1551) 8155 posts |
Well I was wondering if the restriction included links back to a previous article because that would be a real pain, but |
Andrew Hodgkinson (6) 465 posts |
It’s a simple system due to very limited time implementation, but it ought to at least help. There’s now a separate topic title filter I’ve just added, which prevents pasting in simple links. To try it out, try creating a new topic with the ROOL URL pasted in as the topic title. |
Steve Pampling (1551) 8155 posts |
Well I think the posting over in General may point to a human component reading the hints and then testing the limits (unless that’s you) |
mikko (3145) 123 posts |
Maybe now’s a good time to agree not to discuss blacklisting logic on the forum anymore and to delete any historical posts which have discussed it… |
Rick Murray (539) 13806 posts |
This is, of course, assuming that spammers read this stuff and can speak English. |
Pages: 1 2