I was really curious what the damage appears to be so far, so I wrote a script to analyze around 1.55 million bids and generate recommended refunds. I realize this is very premature because NameJet hasn't even confirmed or denied that these accounts participated in shill bidding or determined how to move forward, but I've heard people throwing out some wild guesses like tens or even hundreds of millions of dollars lost and I really doubted that to be the case. In the interest of transparency I want to discuss my method a bit, although several people have told me this approach is still too kind to NJ.
What I did was I looked for public auctions with reserve met, where one of the suspicious aliases was the runner up. Then I looked for the next legitimate bid that wasn't the winner or the suspicious aliases being looked at. That third-party bid was used as the baseline. If the next bid was made by the winner, that became the amount the auction should have closed at. If the next bid was a suspicious alias, I used one bid increment (based on the price level) above the third-party bid to be the new closing price.
This solution isn't perfect because it still works with the "new" price even if that wouldn't have met reserve. In my view the correct way to handle this scenario is to let the winner have it at the calculated price, even if that was below the reserve. I don't think a shill bidder should have rights to enforce a reserve after the deed. The only other options I can think of would be to treat the reserve as the "new" price, or to offer a full refund and the winner returns the domain which might have been sold on already.
Some people feel this method is still too generous and that all of the shill bids should be removed, because it only credits winners when the runner-up was a shill. Take, for example, a situation where the shiller went back and forth with someone, the legitimate bidder led, and then a third party jumped in to win. My method would say that no refund is due at all, even though the auction may have ended much lower without the shill.
But I stand by this method because trying to figure out where this scenario would have ended is messy at best, and is likely impossible. No way to know where in the bidding one of the two real bidders would have given up, and the only reasonable case you can actually make is that it would have ended in the same place (but that ignores the "social proof" of the shill).
Anyway, let me give you the output from a single auction to make it more clear:
================
filet.com closed on 2017-07-10 at $9800
Refund of $2200 is recommended for juggernaut. ($9800 - $7600)
http://www.namejet.com/Pages/Auctions/StandardDetails.aspx?auctionid=3951668
Reasoning:
juggernaut bid $9800 at 2017-07-10 16:16:00
hkdn bid $9700 at 2017-07-10 16:16:00
hkdn bid $9564 at 2017-07-10 16:11:00
juggernaut bid $9464 at 2017-07-10 16:11:00
hkdn bid $8989 at 2017-07-10 16:10:00
juggernaut bid $8889 at 2017-07-10 16:10:00
hkdn bid $8588 at 2017-07-10 16:05:00
juggernaut bid $8488 at 2017-07-10 16:05:00
hkdn bid $8388 at 2017-07-10 16:05:00
juggernaut bid $8288 at 2017-07-10 16:05:00
hkdn bid $8164 at 2017-07-10 16:00:00
juggernaut bid $8064 at 2017-07-10 16:00:00
hkdn bid $7878 at 2017-07-10 15:55:00
juggernaut bid $7778 at 2017-07-10 15:55:00
hkdn bid $7600 at 2017-07-10 15:43:00 <= Winner would have bid this with no shill.
neally bid $7500 at 2017-07-10 15:43:00 <= Last legitimate third-party bid.
================
This is a prime example of the reserve issue. The only way hkdn could have bid twice at the end was if the reserve was not met yet, so that means the reserve was likely somewhere between $9,700 and $9,799 and the "new" price would be way below the reserve. But I still think the recommendation is good.
Here are the results broken down by suspicious alias, across all five seller accounts:
SEEK: $3,582 in recommended refunds across 16 auctions.
WINNER8888: $19,171 in recommended refunds across 40 auctions.
I'm running this separately because I'm still not sure hkdn is a shill, although all evidence points to it and nothing that I can find points to it being a legitimate bidder with a real identity. But anyway:
HKDN: $839,346 in recommended refunds across 2243 auctions.
The number of auctions affected don't exactly match up to the number of times each alias was a runner up that I mentioned before. The reason is that, let's say HKDN was the runner up but seek was the winner, in that case it wouldn't recommend a refund or be considered an affected auction. Plus I also gathered a little more data.
So that's either $22,753 in refunds if hkdn is legitimate, or $862,099 in refunds if the alias is a shill, using a method that may be generous to NameJet. The first two aliases seem to mostly push the reserve and then walk away, or place a few early bids, maybe to move it up when people sort by bid/price. So while the calculated refunds aren't astronomical if HKDN is real, it is still a big deal because these two aliases still unduly influenced more than 1,250 auctions in some way or another even if it wasn't a run-up at the end. If HKDN is a shill it's a really, really big problem.
Obviously NameJet may find no fault, or they may find fault but not give refunds, or choose a different method for calculating refunds. This report is completely unofficial, may be incomplete, and I have no involvement in this decision so don't take it too seriously. Just wanted to get people's heads out of the stratosphere.
I also looked into "russell" that several people mentioned. I really, really hope you guys are wrong about that because that account has been active since at least 2008, has been involved in nearly 25k auctions (6.7k of which were Oliver's), and placed more than $2 million in back orders and bids. I'll look into this more when I have time and more WHOIS history queries.
If anyone wants the output for their particular alias just shoot me an email to
[email protected] (replace xx).
These views are my own, and I'm not speaking on behalf of any company I work for and the respective company owners are not aware of this post. I have to run out for a few hours so might be slow to reply to any follow-up questions.
