|What's the best way to combat directory spam?|
| 12:43 pm on Jul 16, 2005 (gmt 0)|
As my small directory becomes more popular, I've noticed that the majority of submitters can't have bothered to read the submission guidelines. So I'm looking at ways to stem the tide of spam.
But it's time for more effective measures, so my question is, what should I do next? Captchas? Setting cookies on the submission guidelines page? Or are there any other solutions that have worked, and if so, which ones are best?
| 3:42 pm on Jul 17, 2005 (gmt 0)|
Most of people will not care about submission guide! Even for DMOZ, how many read the submission guide before submit a site?
Maybe you just need a way to prevent auto submission, that should be enough.
| 3:48 pm on Jul 17, 2005 (gmt 0)|
Make people pay to be listed.
| 4:05 pm on Jul 17, 2005 (gmt 0)|
i had terrible problems with automated directory spam, one short term thing that works is to move the submission page, i think some spammers get the url and automate subs without going through the process you think they are.
you can log the ip of the submitter and ban further subs from that ip (this has drawbacks too in all kinds of ways)
i have now stretched the sub process over several pages, require cookies and have implimented an email confirmation system for them to confirm their entry by clicking the email link when they get it.
| 8:32 pm on Jul 17, 2005 (gmt 0)|
|Make people pay to be listed. |
I agree. I have paid, reciprocal, and free listings. I almost never get spam submissions to the paid and reciprocal, but constantly get them to the free submission.
| 8:45 pm on Jul 17, 2005 (gmt 0)|
Assuming your directory is offering free listings, then always take a credit card number and ensure the submitter knows they may be charged a "review and rejection" fee.
Simply: if they are eligible to be listed, they get in for free. If they are not, you charge the fee.
That way, you can get rich on other people's spam.
| 2:22 am on Jul 18, 2005 (gmt 0)|
I like victor's idea. I haven't tried it (the ODP social contract precludes this kind of experimentation), but ... I have a feeling it would cause a lot of um, unpleasant interaction with spammers -- they'd want to appeal, then they'd want to argue, then they'd get nasty.
So I was trying to think of some way to tweak that idea to discourage that kind of behavior, and I think the trick is to decouple "paying" and "getting a listing".
Something like: always charge, for a listing or for no listing, WITH the option, at the reviewer's sole discretion, of foregoing the charge for really good sites OR for "almost good enough" sites.
So you'd have four "quality grades":
(1) Editor's pick site -- listed, charge waived. (top 10%, editor's sole option "Thanks for the website suggestion, which is a major contribution to the quality of my index; or which I've confirmed is a worthy non-profit/public-service/information-rich site")
(2) Normal commercial site -- listed and charged. ("OK, yeah, there's a site there, and it's OK, I suppose...")
(3) "almost good enough" site -- not listed, not charged -- a rare case, probably. (1% or so, editor's sole option: sites temporarily down, perhaps. "Come back again, if you like, when you've fixed the problem, but I'll REALLY charge you next time.")
(4) "spam" -- not listed but charged. ("Get lost in the howling wilderness, and may jackals suck the marrow out of your skull cavity while you can still scream.")
So nobody with a listing (class 2) is going to whine about the charge, because the option is to drop them down into no charge/no listing. Likewise, nobody with a listing is going to be in a position to complain about the charge, because they agreed to it up front ... and WITH the charge, it will be clear enough that there's nothing to be gained by a complaint -- your action would be to remove the listing and the charge. And people with no listing but a charge ... you've as good as told them they are bottom-feeding scum-sucking plague-feeding spammers and you really don't want to hear from them again.
Again, I have not tried this; I have not tried to build a commercial directory; and if anyone has tried anything like this, I'd be interested in hearing about it. Reality is always more complex than the simplistic models set up by people with no relevant experience.
| 4:29 am on Jul 18, 2005 (gmt 0)|
IMHO, pay for listing is asking for trouble. People will get the notion that since they are paying customers, they should be listed how they want in the category or categories they want, even if the site is dismal and will have essentially no value to your users.
Yahoo bypasses the argument by telling the submitter s/he's paying for consideration for a listing.
| 11:32 am on Jul 18, 2005 (gmt 0)|
I think I'll go with some kind of captcha, since I'm fairly sure that most of the dud submissions come from bots. The only problem now is finding a solution that isn't inaccessible to people with visual problems. There's a W3C page going into detail about this:
I guess I'll just have to think this one over.
| 11:39 am on Jul 18, 2005 (gmt 0)|
Nice idea, billing spammers sounds really cool. Though I can see some disadvantages:
1) If you want to be sure that the person does not cancel the payment, you would have to get the money first, and refund it in case it is a site of class 1. Billing is easy, refunding is a bit more difficult, especially if you want it to be idiot-proof. One could use an online transaction service like paypal though, this way outsourcing the billing to someone else. :-)
2) Personally I would not submit any class 1 site to a directory like this. I don't trust people who want to bill me for nothing. Guessing I am not the only one who thinks like this, you most liekely will end up with a directory full of commercial sites.
3) I am no lawyer, but I guess a lot of classifications will end up at court, bacause submitters and editors have different opinions about site classification.
| 11:49 am on Jul 18, 2005 (gmt 0)|
I have a directory that has had this problem. Some automatic submission tool finds your submission page and start submitting a lot of irrelevant stuff (my directory only accepts danish pages, but still I get a lot of american viagra/vioxx/#*$! enlargement stuff).
But these autosubmitters leave traces - they often use the same email-adress or name, for instance. So what I do is check for this email/name pattern in the script that processes the submission. If the pattern is there, the submission is not saved - but the text on the page is still "Thank you for the submission", so the autosubmitter can't tell that the submission didn't go through. This has worked well to remove some of the most blatant 5-10 spam submissions/day autosubmitters from my link review list.
I could check for IP as well, I guess, but I think these autosubmitters use a lot of IP adresses...
| 8:47 pm on Jul 18, 2005 (gmt 0)|
I will be simply charging for inclusion. No review fee, but still review all sites though, as I think if people are paying - then they should be included.
But, if the sites are 'unrelated' to my theme, this is fully explained before they add their site, so if they choose to ignore the rules, then they lose the cash and their listing. I find this will keep the spammers at bay, while showing a sense of quality and the fact it's a properly run business, rather than another 'free' directory to be taken advantage of.
Charging certainly makes people think twice before submitting. The spammers especially - if they know they'll get charged, it does stop them fast. Listings are removed on non compliance, it's harsh, but the only way to keep the crap at bay. Ofcourse, if a site owner does get really nasty, then a refund will be arranged only on careful analysis of the threat.
"Dear Sir, We find your directory not to be to our liking. Please remove us at once as we never asked to be listed"
Now, we know if someone has listed or not, and in the case above, we won't refund. Then you have the other removal request types:
"We will send the boyz round, if you don't remove us"
hmmmmmm, not to tricky to deal with. "Nope, you cannot have a refund kind sir"
Then we have the genuine, thick but kinda nice removal request.
Dear Mr Directory Owner,
Please kindly remove us from your directory, as you have charged us for a listing
Er, yeah, which was explained to you when you signed up for it. If you wanted it for free, then why enter your credit card details then? We get lots of these, as it seems people doooooo like a freebie, so we provide an excellent freebie - just in case they want one. Humans eh
We even get submissions, that then claim they didn't submit. Huh?
But will most likely remove them and refund, as some requests just aren't worth the hassle for the fee involved.
| 11:11 pm on Jul 18, 2005 (gmt 0)|
my advice is require a backlink it can be on link page that is not even connected to the main site but a spammer will just leave at this point, i personally try to take time when i submit, i have submitted my site to around 500 directories most of which are ones just starting up, but yes i have a link submission thing on my site but all links have to be aprroved by me or an admin or mod, but agian its not a directory though i have gotten spam i just delete it
| 2:41 am on Jul 19, 2005 (gmt 0)|
I believe that once a directory gets enough attention, the autospammers cease to be a problem (simply because they can be handled automatically if necessary), and it starts getting the "million trained monkeys" submittals -- and monkeys CAN be trained to read the odd letters. This IS based on ODP experience.
I suspect that requiring a backlink will filter out many upfront submitters, whereas the spammers will simply add the backlink -- ONLY until their link is added to the directory.
It's a tricky problem, because the directory is "pinned down" -- whatever you do in self-defence, the spammers have as long as it takes, to zero in on your current position.
| 7:24 pm on Jul 19, 2005 (gmt 0)|
One spammer made themselves very obvious. The guy kept on calling his sites 'A directory of', and to deal with the situation, I sent an email threatening to remove his entries.
This worked as I haven't had any more submissions to date. So a simple email can work wonders. After a while the spammers have to reveal themselves and there's always a pattern.
| 10:58 pm on Jul 23, 2005 (gmt 0)|
The required backlink is a problem because quality sites often don't have link exchange pages. It is often the mediocre sites that lead with link exchange pages.