The great Domain Name Wire survey fraud
Wednesday, February 13th, 2013
Someone went through quite a bit of effort to game this year’s survey.
The 8th annual Domain Name Wire survey has now closed.
I do the survey every year, and I think it provides a pretty good snapshot of what’s going on in the domain industry: how domainers are doing, what they think of various domain companies, etc.
Of course, every year there’s a bit of ballot stuffing. It’s not a scientific survey and anyone can respond. Companies sometimes ask their customers to vote for them, just like they do in any web voting.
It’s no big deal and it’s all in fun.
But this year someone took it to the extreme. It was fairly sophisticated, yet also clumsy.
Someone submitted about 1,500 entries designed to sway the results.
They either used a computer system or something like mechanical turk.
I’m guessing the former for a few reasons.
They quickly submitted surveys from various blocks of IP addresses. The IP addresses were different, but the same blocks and countries. The system they used filled out the surveys in about 90 seconds start to finish and one every three minutes, like clock work.
Each survey had some varying answers, but also some commonalities. For example, on the question where people were asked to submit a list of all parking companies they used over the past year, the first three were identical on almost all of them.
The best parking company answer rotated between two companies.
But the best part was the domain broker part. As I reviewed the results, I saw that two domain brokers rotated pretty much evenly in the results.
Here’s the kicker. When I looked at the first couple hundred automated submissions, here’s what the best domain broker line said:
“Put 50% Broker One, Put 50% Broker Two”
I’m not going to disclose the names of the brokers, because I suspect at least one (or maybe both) were unwitting participants.
The good thing is the automated submissions were so obvious when looked at in totality that I was able to clean them out of the survey. It was a pain, but I was left with about 550 good entries that will be used as the basis for the survey results.
Although I have my suspicions, if anyone knows who was behind this I’d certainly appreciate you dropping a line.
Next year the survey will be handled differently for sure.
For this year, look for the survey results to come out over the next few weeks — with the bogus submissions removed.