Comp/Math 553: Algorithmic Game Theory Lecture 09

Slides:



Advertisements
Similar presentations
Combinatorial Auction
Advertisements

6.896: Topics in Algorithmic Game Theory Lecture 21 Yang Cai.
An Efficient Dynamic Auction for Heterogeneous Commodities (Lawrence M.Ausubel - september 2000) Authors: Oren Rigbi Damian Goren.
6.896: Topics in Algorithmic Game Theory Lecture 20 Yang Cai.
Auction Theory Class 5 – single-parameter implementation and risk aversion 1.
Yang Cai Oct 01, An overview of today’s class Myerson’s Auction Recap Challenge of Multi-Dimensional Settings Unit-Demand Pricing.
Approximating optimal combinatorial auctions for complements using restricted welfare maximization Pingzhong Tang and Tuomas Sandholm Computer Science.
CPS Bayesian games and their use in auctions Vincent Conitzer
Seminar in Auctions and Mechanism Design Based on J. Hartline’s book: Approximation in Economic Design Presented by: Miki Dimenshtein & Noga Levy.
Auctions. Strategic Situation You are bidding for an object in an auction. The object has a value to you of $20. How much should you bid? Depends on auction.
An Approximate Truthful Mechanism for Combinatorial Auctions An Internet Mathematics paper by Aaron Archer, Christos Papadimitriou, Kunal Talwar and Éva.
Game Theory in Wireless and Communication Networks: Theory, Models, and Applications Lecture 6 Auction Theory Zhu Han, Dusit Niyato, Walid Saad, Tamer.
Selling Billions of Dollars Worth of Keywords Presented By: Mitali Dhoble By Benjamin Edelman, Michael Ostrovsky And Michael Schwarz Reference:
Yang Cai Sep 10, An overview of today’s class Case Study: Sponsored Search Auction Myerson’s Lemma Back to Sponsored Search Auction.
Auction Theory Class 3 – optimal auctions 1. Optimal auctions Usually the term optimal auctions stands for revenue maximization. What is maximal revenue?
Optimal auction design Roger Myerson Mathematics of Operations research 1981.
Part 1: Optimal Multi-Item Auctions Constantinos Daskalakis EECS, MIT Reference: Yang Cai, Constantinos Daskalakis and Matt Weinberg: An Algorithmic Characterization.
Seminar In Game Theory Algorithms, TAU, Agenda  Introduction  Computational Complexity  Incentive Compatible Mechanism  LP Relaxation & Walrasian.
Yang Cai Oct 15, Interim Allocation rule aka. “REDUCED FORM” : Variables: Interim Allocation rule aka. “REDUCED FORM” : New Decision Variables j.
6.853: Topics in Algorithmic Game Theory Fall 2011 Matt Weinberg Lecture 24.
Yang Cai Sep 17, An overview of today’s class Expected Revenue = Expected Virtual Welfare 2 Uniform [0,1] Bidders Example Optimal Auction.
Algorithmic Applications of Game Theory Lecture 8 1.
Mechanism Design and the VCG mechanism The concept of a “mechanism”. A general (abstract) solution for welfare maximization: the VCG mechanism. –This is.
Sponsored Search Auctions 1. 2 Traffic estimator.
Yang Cai Sep 24, An overview of today’s class Prior-Independent Auctions & Bulow-Klemperer Theorem General Mechanism Design Problems Vickrey-Clarke-Groves.
Combinatorial Auction. Conbinatorial auction t 1 =20 t 2 =15 t 3 =6 f(t): the set X  F with the highest total value the mechanism decides the set of.
Yang Cai Sep 15, An overview of today’s class Myerson’s Lemma (cont’d) Application of Myerson’s Lemma Revelation Principle Intro to Revenue Maximization.
Yang Cai Sep 8, An overview of the class Broad View: Mechanism Design and Auctions First Price Auction Second Price/Vickrey Auction Case Study:
A Truthful Mechanism for Offline Ad Slot Scheduling Jon Feldman S. Muthukrishnan Eddie Nikolova Martin P á l.
Multi-Unit Auctions with Budget Limits Shahar Dobzinski, Ron Lavi, and Noam Nisan.
VCG Computational game theory Fall 2010 by Inna Kalp and Yosef Heskia.
Auction Seminar Optimal Mechanism Presentation by: Alon Resler Supervised by: Amos Fiat.
6.853: Topics in Algorithmic Game Theory Fall 2011 Constantinos Daskalakis Lecture 21.
Mechanism Design CS 886 Electronic Market Design University of Waterloo.
Auction Theory תכנון מכרזים ומכירות פומביות Topic 7 – VCG mechanisms 1.
Yang Cai Oct 08, An overview of today’s class Basic LP Formulation for Multiple Bidders Succinct LP: Reduced Form of an Auction The Structure of.
Yang Cai Oct 06, An overview of today’s class Unit-Demand Pricing (cont’d) Multi-bidder Multi-item Setting Basic LP formulation.
Steffen Staab 1WeST Web Science & Technologies University of Koblenz ▪ Landau, Germany Network Theory and Dynamic Systems Auctions.
6.853: Topics in Algorithmic Game Theory Fall 2011 Constantinos Daskalakis Lecture 22.
Combinatorial Auction. A single item auction t 1 =10 t 2 =12 t 3 =7 r 1 =11 r 2 =10 Social-choice function: the winner should be the guy having in mind.
Advanced Subjects in GT Prepared by Rina Talisman Introduction Revenue Equivalence The Optimal Auction (Myerson 1981) Auctions.
Lecture 4 on Auctions Multiunit Auctions We begin this lecture by comparing auctions with monopolies. We then discuss different pricing schemes for selling.
Auctions Supplemental Material. In case you haven't noticed: Auctions are Everywhere! eBay → simple auctions for a single item AdWord Auctions → advertisers.
Comp/Math 553: Algorithmic Game Theory Lecture 10
Comp/Math 553: Algorithmic Game Theory Lecture 11
False-name Bids “The effect of false-name bids in combinatorial
Bayesian games and their use in auctions
Comp/Math 553: Algorithmic Game Theory Lecture 08
CPS Mechanism design Michael Albert and Vincent Conitzer
Comp/Math 553: Algorithmic Game Theory Lecture 14
Laddered auction Ashish Goel tanford University
Comp/Math 553: Algorithmic Game Theory Lecture 13
Market Design and Analysis Lecture 4
Authors: Oren Rigbi Damian Goren
Comp/Math 553: Algorithmic Game Theory Lecture 15
Economics and Computation Week #13 Revenue of single Item auctions
Vincent Conitzer Mechanism design Vincent Conitzer
Vincent Conitzer CPS 173 Mechanism design Vincent Conitzer
Preference elicitation/ iterative mechanisms
Introduction to Mechanism Design
Introduction to Mechanism Design
Auctions Lirong Xia. Auctions Lirong Xia Sealed-Bid Auction One item A set of bidders 1,…,n bidder j’s true value vj bid profile b = (b1,…,bn) A sealed-bid.
Combinatorial Auction
CPS Preference elicitation/ iterative mechanisms
Introduction to Mechanism Design
Information, Incentives, and Mechanism Design
Auction Theory תכנון מכרזים ומכירות פומביות
Vincent Conitzer CPS Mechanism design Vincent Conitzer
CPS Bayesian games and their use in auctions
Class 2 – Revenue equivalence
Presentation transcript:

Comp/Math 553: Algorithmic Game Theory Lecture 09 Yang Cai

Menu Vickrey Auction Case Study: Sponsored Search Auction Myerson’s Lemma Back to Sponsored Search Auction

VICKREY auction

Second Price/Vickrey Auction Another idea: Charge the winner the second highest bid! maybe a bit strange But essentially used by Sotheby’s (modulo reserve price).

Second-Price/Vickrey Auction Lemma 1: In a second-price auction, every bidder has a dominant strategy: set her bid bi equal to her private value vi. That is, this strategy maximizes the utility of bidder i, no matter what the other bidders bid. Hence trivial to participate in. (unlike first price auction) Proof: See the board.

Second-Price/Vickrey Auction Lemma 2: In a second-price auction, every truthful bidder is guaranteed non-negative utility. Sometimes called Individual Rationality Proof: See the board.

Second Price/Vickrey Auction [Vickrey ’61 ] The Vickrey auction satisfies the following: [strong incentive guarantees] It is dominant-strategy incentive-compatible (DSIC) and IR (2) [strong performance guarantees] If bidders report truthfully, then the auction maximizes the social welfare Σi vixi, where xi is 1 if i wins and 0 if i loses. (3) [computational efficiency] The auction can be implemented in polynomial (linear) time.

Second Price/Vickrey Auction Questions: What’s so special about 2nd price? How about charging the highest bidder the 3rd price?

What’s next? These three properties: (i) strong incentive guarantees, (ii) strong performance guarantees, (iii) computational efficiency are criteria for a good auction. Our goal in future lectures will be to: Tackle more complex allocation problems Tackle more complex objectives, such as revenue

Case Study: Sponsored Search Auction

Sponsored Search Auction

In 2012, sponsored search generated 43 In 2012, sponsored search generated 43.6 billion dollars for Google, which was 95% of its revenue. In the meantime, the market grows by ~20% per year.

Sponsored Search Auction: Setup 1 i n … Bidders (advertisers) v1 vi vn 1 j k … Slots Auctioneer/Google α1 αj αk k slots for sale. Slot j has click-through-rate (CTR) αj. Bidder i’s value for slot j is αjvi. Two complications: Multiple items Items are non identical In reality more complex model, where different bidders have quality which multiplies the CTR of the slot they are assigned to. These qualities are estimated by Google.

Sponsored Search Auction: Goals Ideally, DSIC and IR; i.e. truthful bidding should be a dominant strategy, never leading to negative utility. (2) Social welfare maximization; i.e. the assignment of bidders to slots should maximize Σvixi, where xi denotes the CTR of the slot to which i is assigned (or 0 if i is not assigned to any slot). Each slot can only be assigned to one bidder, and each bidder gets only one slot. (3) Polynomial running time. Remember millions of these auctions are run every day!

Sponsored Search Auction Two things to consider: who wins what (allocation) and how much to charge (prices)? Tackle these questions one at a time: Assume that bidders will bid truthfully. Then, how do we assign bidders to slots so that properties (2) and (3) hold? Subject to (1), how do we set the prices so that truthful bidding is a dominant strategy?

Sponsored Search Auction Assume that bidders will bid truthfully. Then, how do we assign bidders to slots so that properties (2) and (3) are satisfied? Greedy Alg. ✔ Subject to 1), how do we set the prices so that truthful bidding is a dominant strategy? use Vickrey payments for each slot? (i.e. charge whoever got slot j, the value of whoever got slot j-1, etc.)

Generalized Second Price Auction GSP auction: Suppose there are 2 slots for sale, the first slot has CTR α1=50% and the second slot has CTR α2=30%. Your value vi happens to be 1000 - the last 3 digits of your McGill ID #. You need to submit a bid bi. Allocation rule: allocate the slots greedily according to the bids. If you win slot j, that is, your bid is the j-th highest bid, then your utility is αj(vi –pj+1), where pk is the k-th largest bid. Otherwise, your utility is 0. Your goal is to maximize your utility. Let’s Play it! - What’s your bid if you are playing against the whole class?

Sponsored Search Auction GSP is not truthful! Example: 3 bidders 2 slots. v1=7, v2=6 and v3=1; α1=1 and α2=0.4. Instead of being truthful, it’s better for bidder 1 to bid 5 and win the second slot.

Sponsored Search Auction Assume that bidders will bid truthfully. Then, how do we assign bidders to slots so that properties (2) and (3) are satisfied? Greedy Alg. ✔ Subject to 1), how do we set the prices so that truthful bidding is a dominant strategy? Myerson’s Lemma!

Myerson’s Lemma

Single-dimensional Environment Definition: n bidders Each bidder i has a private value vi, its value “per unit of stuff” that she gets. A feasible set X. Each element of X is an n- dimensional vector (x1, x2, . . . , xn), where xi denotes the “amount of stuff” given to bidder i.

Single-dimensional Environment Examples: Single-item auction: X is the set of 0-1 vectors that have at most one 1 Selling k units of the same item to unit-demand bidders: X are the 0-1 vectors satisfying Σi xi ≤ k. Sponsored search auction: X is the set of n- dimensional vectors corresponding to assignments of bidders to slots. If bidder i is assigned to slot j, then the component xi equals the CTR αj of its slot. What makes a setting single-dimensional is NOT the allocation constraints, but that every bidder’s private information is one scalar.

Sealed-Bid Auction in Single-dim Environments Make Two choices: Allocation rule x : Rn  X Payment rule p : Rn  Rn Sealed-Bid Auction Execution: Collect bids b=(b1 , ..., bn) [allocation] Implement allocation x(b) [payment rule] Charge prices p(b)

Two important definitions Definition 1: (Implementable Allocation Rule) An allocation rule x for a single-dimensional environment is implementable if there is a payment rule p s.t. the sealed-bid auction (x, p) is DSIC. Example: When selling a single indivisible item, the allocation rule that gives the item to the highest bidder is implementable. Pending Question: Is the Greedy allocation rule implementable in Sponsored Search? How about giving the item to the second highest bidder in single-item settings? How about the lowest bidder?

Two important definitions Definition 2: (Monotone Allocation Rule) An allocation rule x for a single-dimensional environment is monotone if for every bidder i and bids b−i by the other bidders, the allocation xi(z, b−i) to i is non-decreasing in i’s bid z. In single-item settings, the allocation rule that gives the item to the highest bidder is monotone, while the allocation rules that give the item to the second highest or the lowest bidder are not monotone. The Greedy allocation rule for Sponsored Search is monotone.

Myerson’s Lemma [Myerson ’81 ] Fix a single-dimensional environment. An allocation rule x is implementable if and only if it is monotone. (b) If x is implementable/monotone, there is an essentially unique payment rule such that the sealed-bid mechanism (x, p) is DSIC, given by the formula: (c) In particular, there is a unique payment function such that the mechanism is DSIC and additionally IR with non-positive transfers (i.e. bi = 0 implies pi(b) = 0, for all b-i). PROOF ON THE BOARD

Myerson’s Lemma Corollaries Corollary: The greedy allocation rule for sponsored search is implementable. Thus, there is a DSIC auction that maximizes social welfare. On the other hand, in single-item settings, allocating to the second-highest bidder or the lowest bidder are both not implementable.