Vickrey Auctions and their Application to Modern Computer Networks and Auctioning
With such an unprecedented semester, and with my unsurprising levels of disorganization, it would be no surprise to anyone close to me that I decided to add a 4000 level class on the last day possible to add classes. While this made it a rather painful process to catch up on 2 weeks of missed class, I miraculously ended up actually enjoying the content I was learning for once. ECE 4450: Computer Networks and Telecommunications deals with, you guessed it, computer networks and telecommunications. The class itself offers a very interesting discourse on the actual systems and protocols the Internet depends on, but also inadvertently discusses many things we cover in 2040! While doing research for this post, I was particularly curious about second-price sealed-bid auctions, as I had never heard of them before, and also because HW3 relied very heavily on them so I figured they were at least somewhat important to the course.
Although sealed-price second-bid auctions are known to have occurred as early as the 18th century (apparently stamp collectors frequently used such a system), the rigorous mathematical and statistical study of these particular auctions is credited to William Vickrey, so another common name for these auctions is a Vickrey auction. Aside from auctions, much of Vickrey’s work is relevant to the content covered in 2040; for example, Vickrey postulated that pricing on congested trains and toll roads should increase during peak-use periods, something we realized when considering Nash equilibria on congested traffic networks. Vickrey’s work proved fundamental to a lot of network theory, and he was awarded the Nobel prize in 1996 for his work on sealed-bid second-price auctions. In an unfortunate turn of events, Vickrey died of a heart attack three days after being named, so he never lived to see the prize.
An amazing connection between Vickrey auctions and both my 2040 and 4450 classes is the application of Vickrey’s theory to computer networks and advertisement auctions hosted by companies like Google. Central to the theory that a Vickery auction’s only dominant strategy is to bid truthfully is the Vickrey-Clarke-Groves mechanism, the formal proof of this strategy. Network routing in the Internet’s lower layers employs the VCG mechanism to determine the most optimal distribution of traffic among nodes in the network, and it happens to be that this strategy is always optimal. In terms of advertisement auctions, I was able to find an actual patent for a particular method of advertising auction used by many ad companies that builds upon Vickrey auctions. Google’s very own system for ad auctions initially was just a Vickrey auction. In recent years, they have changed the system considerably but the idea, at its core, is still rooted in the fundamentals of Vickrey’s work.
Sources: https://www.britannica.com/biography/William-Vickrey#ref85027
https://patents.google.com/patent/US20050289043A1/en
https://www.aeaweb.org/articles?id=10.1257/jep.14.3.183