5 ms·
The highest bidder pays the second highest bid. Generalized, the k-th highest bidder pays the (k-1)-th bid for some k-th item. It enforces truthful bidding and
by whymauri 7y ago
The highest bidder pays the second highest bid. Generalized, the k-th highest bidder pays the (k-1)-th bid for some k-th item.
It enforces truthful bidding and is revenue optimal. How is it needlessly complex?
- doctorpangloss 7y agoWhile revenue optimal, growth in retail is eventually driven by psychology, not pricing, since if there’s anything our real economy is good at producing, it’s morons.
- whymauri 7y agoThat's the Bulow-Klemperer theorem. You would rather have more growth (bidders) at the expense of an optimal bidding mechanism.
- antpls 7y agoIf actors are humans, then sure. But if actors are rational computers, and it's used billions time a day, you might prefer optimality. Not saying this would be useful for communication between computers, but coming from CS degree, I have the intuition it could have some use cases.
- denormalfloat 7y agoIt's needlessly complex because it doesn't solve any problems I have or any I have heard of. A simpler auction would still work pretty well, and be more easily understood by participants. I ask because the only case I can think of this providing any value, it would be too complex. The example would be splitting up the assets after a divorce, where each spouse bids on the parts of the estate. A VCG auction may be more accurate, but what couple or lawyer would agree to use this system? It's too complex.
- zodiac 7y agoWhat "simpler auction system" would you propose?
- orijing 7y agoYou're describing the generalized second price auction (GSP) which is NOT incentive compatible, which means the optimal strategy isn't to bid truthfully. VCG is.