Using AI to design the perfect auction

AI has achieved a significant amount in recent years, but perhaps one of the more interesting applications was in the design of the auction of airwaves by researchers from UBC and Stanford University.  The process was described in a recently published paper.

The last few decades have seen a significant shift in the use of bandwidth, as mobile operators gobble up more of the spectrum from television broadcasters.  This is especially so as mobile data usage has mushroomed.

Physical limitations restrict the amount of bandwidth available to carry all data across the air, but traditionally there has never been enough data to brush up against those limits.  That has changed however, and the FCC recently asked the TV stations to auction off any of their unneeded airwaves to free up space for mobile data in a move designed to free up 84 megahertz of spectrum, generating some $17 billion for both the government and the broadcasters.

Smart auctions

The auction was a success because of a smart reverse auction whereby the price wasn’t determined by the highest bidder, but rather by how low a sum the broadcasters were willing to sell their airwaves for.

“Auctions work when we don’t know what things are worth,” the researchers say. “In this case, we were asking television stations to decide how much it was worth for them to stop broadcasting.”

The result was that densely populated areas, such as New York, saw broadcasters make significant sums, unlike sparsely populated areas where much smaller sums changed hands.

“The auction paid billions of dollars to television stations to stop being broadcasters which is good for struggling stations that had no other way to sell broadcast rights,” the team continue. “People will be happier with faster phones and the government achieved its goal to make better use of the spectrum, which strengthens the economy and widens the tax base.”

The system was even capable of factoring in various things such as the number of trades taking place, and the property rights.  For instance, certain broadcasters were happy to sell rights in certain places, but only providing certain conditions were met in each location.

The researchers believe that the system could be useful in other countries where bandwidth is being auctioned off, especially for services such as 5G mobile telephony, but also for other auctions such as farming rights.

It’s a nice example of how computing and AI is being used to handle complex, yet crucial processes.