3

We get a lot of spam answers (and I've just seen a spam edit) with links to domains selling dubious software. There are a few websites that claim to offer tutorials for recovering data from a broken phone, but the "tutorial" is an advert for PC software.

The links are added indiscriminately by newly registered or anonymous users, often to inappropriate questions, and also sometimes to questions where they would be a reasonable answer. The text themselves is often spammy-looking, repeating the same phrasing several times or pasting text from their site, for SEO purposes. Even when the answers are sometimes relevant, I don't think we ought to encourage commercial users to follow this scatter-shot approach and rely on our users spending time deciding which posts might be appropriate.

The websites in question look suspicious because (1) they don't give any contact details or official information on the company selling the software, not even a country of origin; (2) they offer a PC software download for an Android problem; (3) they have heavy use of SEO keywords, often with a separate URL per device; (4) often they hide their WHOIS information to make sure you can't find out who they are. These are not properties of honest software businesses.

Domains I've seen people try to link to include:

  • istonsoft.com (hidden whois information)
  • swesoft.com (based in CN)
  • android-data-recovery-software.com (same owner as swesoft)
  • z0download.com (hidden whois information; here's a huge edit that added this link)
  • yodot.com (hidden whois information; here's an answer)
  • android-mobile-manager.com (hidden whois information)

We have a zillion questions on recovering deleted files, and AFAICT they all attract this kind of spammy answer/edit. Certainly, several have been protected for this reason. I'm proposing to reduce the number of spam answers/edits that get through review, and to reduce the load on reviewers, by stopping our repeat offenders earlier.

  • From Meta.SO Join the crusade against r0uter-switch.com Content Spam Seems that it's possible. I've seen some of those isotonsoft vandal edits this week, too. – GAThrawn Dec 5 '13 at 13:20
  • 2
    So just as a PSA of sorts, another behavior these sites have begun using to try to fly under the radar is posting a seemingly innocuous answer and then returning to edit the link in later in a somewhat subtle, inline manner (sometimes on the order of several weeks after the answer was originally posted, with the OP of the answer performing the edit so it won't be reviewed). If you see any of these links, definitely don't hesitate to flag them, because I actively search for and destroy accounts that exist to promote them. The sheer volume makes it pretty obvious that they're spam. – eldarerathis Dec 5 '13 at 14:33
  • @eldarerathis I've seen a few "first posts" lately that appear to be machine-generated (using some Dissociated-Text-like algorithm) that are probably for that purpose. I've been flagging them as spam even when they contain no spam link. Other reviewers, keep your eyes open! – Dan Hulme Dec 5 '13 at 14:48
  • On Music our spammers are much dumber -- they've been trying anonymous suggested edits, which never go through. Kind of hilarious how dumb they are. – Matthew Read Dec 6 '13 at 15:51
2

Yeah, we can blacklist specific URLs or domains.

But it's unlikely to help here.

Blacklisted terms are fairly easy to work around; if spam is being posted or edited in manually, then it's not too much more difficult for the spammer to use a URL shortener or alternate domain to circumvent the restriction.

For those 4 domains, I've found only 6 posts and 2 suggested edits that match any of them (that includes past revisions of posts and rejected edits). Combined with the similarities you've observed, this suggests spammers are already fairly comfortable using "throw away" domains to avoid detection.

We've taken steps recently to beef up our spam detection systems; the best thing you can do is to flag posts that've been created by spammers, and mark any spam-creating suggested edits appropriately as well. All of this feeds into a database that can block future attempts without the need for further intervention.

  • There were two more in the review queue today. As I showed above, some of these are getting through review, so the existing system doesn't seem to be working well enough. – Dan Hulme Dec 6 '13 at 9:31

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .