Hacker News new | ask | show | jobs
by C4stor 2517 days ago
This style of auction moves the complexity away from the buyer.

In a traditional one-round sealed first price auction, you're trying to guess what everybody else is bidding and bid a penny more. It's incredibly difficult for a buyer to accurately predict that, it leads to bubbles and crashes disconnected from the sold item value since it barely accounts for anything in the reasoning, but it's easy to resolve.

With a second price auction, the buyer only need to know its true price for self, and bid that, which is actually a lot more simple, and allows to bid depending on the actual provided value of items, making sure the prices don't err too far away from the value.

So, it's actually very practical for one turn auctions, which in turn are very practical when you don't want to spend the time to run a multi turn auction with progressive bids.
1 comments
hdkrgr 2517 days ago
This is really only true in combinatorial auctions when the set of items is small. In the divorce example (split n items between the two) from the other comment, VCG would require both partners to list their truthful valuation for all 2^n bundles of items.

Since this isn't really feasible in practice, VCG isn't used even in settings, where the complexity of the mechanism would be warranted. (Think cellular frequency auctions, where CCAs or SMRAs are used.)

link
C4stor 2517 days ago
Well, I've only known it used in the adtech space, but I'll really miss it. It made writing a bidder on the open rtb exchanges a doable things without too much tech.

With exchanges going back to first price, I think the barrier to entry will be a lot higher for new players. It's a sad state of affair that second price auction was ruined by fraudulent clearers, where the second price for your bid was always your bid minus 1 cent...

I attended a conference given by the ex scientific director of Criteo demonstrating that mathematically, if players clear bids honestly, second price auction is the only economically viable choice in the long term. It was quite convincing. But the hypothese didn't hold up I suppose.

link