For some reason, I am more tired than usual this morning and it seems that I am not able to wake myself up and actually do work. So far I have gotten iced coffee (which never tastes as good as regular coffee but it is too hot to purchase a very hot drink), checked email five thousand times, and started writing this post. I did change something in a program that I am working on, but that change caused it to explode with bugs which I am too tired/lazy to look at it. What a good use of a Friday morning.
In other news, the people at the Freakonomics blog have had their say on Swoopo, which is cool since I wrote about it a couple days before they did. A commenter (the only one) on my last post advertised this site called pricedrip.com as an alternative to Swoopo. The mechanism for this site is that you place a single bid, naming a price that you would like to pay for the item and then the price is lowered periodically until someone’s bid is met. Of course, the site charges you for each bid that you placed, but it would seem that an individual would be less likely to place more than a couple of bids per item with this method. This type of auction is known as a Dutch auction, where the auctioneer starts high and then lowers the price until someone is willing to buy it, with the difference that you must pay for your bid on this site. Admittedly, I haven’t really watched any of these auctions, but I guess the real question is how far does the price drop below the market price (the best price that you could purchase it for on the internet). It seems like you have a decent chance of getting a marginal savings on a product. As long as your savings is greater than the price of the bid, you would probably make money. Of course, if everyone uses that same strategy, then everyone gets screwed because everyone will have to bid higher to get that discount, thus decreasing the value of the discount. The thing that makes this less “evil” than Swoopo is that you do not gain any more information over the course of the auction that would cause you to change your bid. Say your bid is X and the auctioneer announces the current price is Y such that X < Y. If no one buys it at Y and the price moves to Y-1, then you gained no information because you implicitly assumed that no one was going to buy it at Y (if you had assumed that, you would have never spent money placing a losing bid of X). If someone does win it at Y, you gained information, but the auction is over so that information is useless now. Since you gain no new information, it never really makes any sense to place more than one bid.
You might conclude that if everyone else is bidding such as you are, that is trying to get a small discount from the item, then they might outbid you on it, so you should increase your bid by a little bit to win the item. Of course then they might have the same idea and increase their bid, and so on and so forth. Eventually you might reach a conclusion that you should bid more than the market value of the item to win the auction, but that is irrational. While that might happen at Swoopo (you would continue bidding so as to not lose the investment that you had already put in), that wont happen here since all that calculation would go in before you place your bid. Once you realize that the only winning strategy is to overbid the item, you would just go to Amazon and buy it and thus never lose any money on the auction. Thus my claim is that this site is probably less evil than Swoopo. I think some work has been done on auctions where there is an entry fee to the auction (not per bid), but considering it never makes any sense to bid more than once in this auction, the same strategies can be applied. I will try to look into that.
I started reading “Linked” by Barabasi which is a popSci book about the science of networks. Barabasi is a pretty big guy in the field but based on some videos that I have seen with him, I think he is prone to overestimating the importance of network theory. Still, it should be an interesting read and is only about 200 pages. I kind of want to do some type of empirical study in network theory (maybe a project like this or this), but I have yet to come up with a good, feasible idea (good means it has to be interesting and its conclusions have to be worth the effort and feasbile means that data has to be relatively easy collect). Oh well, back to doing some work (more likely, reading random news stories before lunch).
