You'll find it on eBay!

5 July 2010
Man fined over fake eBay auctions by Dan Whitworth of the BBC

eBay spokesperson Vanessa Canzenni denies that not enough is being done to prevent [shill-bidding].

[…]

[eBay user Rezza Faizee, having noted that shill-bidding were a significant problem, said] I honestly don't know what you can do to tackle the problem, I honestly don't.

Catching shill-bidders on eBay used to be one of my hobbies. I would regularly stumble-upon suspicious confluences, start examining auction and bidder histories, and from them often assemble proof that there had been shill-bidding, which proof I would then send to eBay and to the victims. I'm sure that I wasn't the only person engaging in this sort of detection.

But eBay began choking-off the data available to us. With decreasing information, it became ever harder to make the case. It became impossible even to see some of the confluences that would have triggered suspicion in the first place.

For an honest auction firm, there may be an optimal amount of shill-bidding to allow, simply because of enforcement costs. (A perfectly secure trading environment would be prohibitively expensive.) But for a dishonest firm the question is of balancing the gain that otherwise comes from allowing ending prices (and hence fees) to be thus increased, against the alienation of users who consequently reduce their spending. Access to information which both empowers volunteers to catch shill-bidders and alerts users more generally to the occurrence of shill-bidding is, as such, not in the perceived interest of a dishonest firm.

BTW, the changes that reduced our abilities to spot shill-bidders, and which made it more typically impossible for us to prove a case of shill-bidding (as well as other changes that enabled eBay to be more easily used by thieves) were primarily effected while Margaret Cushing (Meg) Whitman, now the Republican Party nominee for governor of California, was eBay's President and CEO.

Tags: , , ,

3 Responses to You'll find it on eBay!

  • Speaking of politics... okay, how about rather than comment from an entry you made over a year ago, I'll do it here. I remember you calling to attention a simple linear regression chart concerning the reporting of vote talley from the election in Iran. Here in Mexico, we just had an election, as you may well know. I decided that, since the Gevernment was kind enough to provide the data on line to the internet masses, I should try to do my own chart based on their data. As you may or may not know, I am a writer, sometimes journalistically for the money, but I would appreciate the quick evaluation from someone who, quite honestly, knows what they are doing much more than I do. I don't feel comfortable writing about this otherwise.

    First, their data can be found here:

    http://www.prepbc.com/tendencia_ayuntamientos.cfm?ayuntamiento_id=4

    I began at 8PM and took data from every twenty minutes until there was no more movement (the count was complete, I presume).

    Second, here is a screen capture of my spreadsheet:

    http://farm5.static.flickr.com/4117/4767197509_3d3f015034_b.jpg

    No idea what I'm doing wrong, or perhaps, what I've done properly. Your input would be generous and very much appreciated.

    • Daniel says:

      I don't see anything wrong in what you've done here.

      But it's important to be careful about interpretation. I don't know how the votes were collected and counted in Tijuana. If they were first basically dumped into a big bin, shuffled, and then counted, then a high r2 might not be quite so remarkable (though it's still hard to buy). If, on the other hand, the votes from some parts of the city (with one socioëconomic character) were counted before those from other parts of the city (with different socioëconomic character), then a high r2 would be literally incredible.

      (BTW, I get notification when comments are made to old entries.)

      • Thank you. I had thought of that as well, that perhaps should everything simply be counted after being clumped together, it could account for the high r squared value. And certainly, before writing a story that would carry such obvious magnitude (especially since I'm an American living in Mexico), I plan on interviewing someone responsible for the data to ensure I'm not making a mistake. However, Tijuana is a large city, it takes time to get the ballots there. And, I took the liberty of performing the same action in Mexicali, Tecate, and Ensenada, where certain polling places are certainly far away (one or two hours in some cases) from wherever the votes are counted. And, the r-squared values are all above .999 in those cases as well.

        And, your statement is the obvious one, because more so than in the U.S., the people of Mexico tend to vote according to their economic status, and that division is spread far more than in the U.S. and the classes are very much separated regionally.

        Thanks very much, and I'll keep you posted as to where this goes (and certainly link the article should I write it).

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.