Tee Time with Admiral Poindexter Markets at Microsoft

  • Slides: 8
Download presentation
Tee Time with Admiral Poindexter? (Markets at Microsoft) Todd Proebsting Microsoft

Tee Time with Admiral Poindexter? (Markets at Microsoft) Todd Proebsting Microsoft

Three Four Types of Researchers • Once upon a time I thought there were

Three Four Types of Researchers • Once upon a time I thought there were 3 types of people involved in research: – Priests – Prostitutes – Pragmatists • Now, I believe there is a fourth: – Promoters (Proselytizers? )

Markets Inside Microsoft • 6/03: E-Commerce conference introduction • 7/03: PAM debacle • “Tee

Markets Inside Microsoft • 6/03: E-Commerce conference introduction • 7/03: PAM debacle • “Tee time with Admiral Poindexter, Sir? ” • 9/03: Start market software creation • 2/04: Beta Test #1(Wisconsin Democratic Primary) • ~50 traders, prize lottery • 3/04: Beta Test #2 (DARPA Grand Challenge) • ~300 traders, prize lottery • 8/04: Two real markets (schedule and bug count for small, internal project) • ~25 traders, $50 each • 9/04: Two real markets (schedule for small project) • ~25 traders, $30 each • 2/05: A half-dozen people “really close” to sponsoring markets

Pitch • 10+ Formal presentations, countless informal pitches • Better predictions from proper incentives

Pitch • 10+ Formal presentations, countless informal pitches • Better predictions from proper incentives – Incentive to reveal true beliefs – Incentive to reveal confidence – Incentive to gather information • Bash competing prediction methods • Reality check on prevailing predictions “What does the market know that I don’t know, or what do I know that the market doesn’t know? ”

Negative Reactions to Idea • • • • Insider trading Real money bothers some

Negative Reactions to Idea • • • • Insider trading Real money bothers some Real money is needed Boom/bust cycles (irrationality) Gambling Markets don’t work Distraction from real work Redundant mechanism Unstable market (predictions affect decisions) – E. g. , predicting failure promotes corrections Self-fulfilling market (predictions affect outcome) – E. g. , predicting success increases budget Conflicting incentives – E. g. , make money off of project failure Positive incentives – E. g. , predicting success increases budget, which increases sales, which increase personal gains (bonuses, promotions, …) Needing markets is a sign of a dysfunctional organization Disclosing that the emperor has no clothes damages morale

Positive Reactions to Idea • Market believers – “It was a blast for me

Positive Reactions to Idea • Market believers – “It was a blast for me as well, finally finding someone as (well, really more) passionate as I about these things. I’ve begun pinging people about whether they want to sponsor one. ” • Mischief makers – “You should run a market on XXX---I’ll short ON TIME. ”

First “Real” Market • August 2004: Predict internal product ship date – Official, accepted

First “Real” Market • August 2004: Predict internal product ship date – Official, accepted schedule: mid-November 2004 – 25 traders @ $50, made up of testers, developers, etc. • Securities: Pre-NOV, DEC, JAN, FEB, Post-FEB • Within a few minutes of opening, NOV dropped to $0. 012… • Currently expected to ship in February… • Software functioned well • Only 60% of traders traded • Automated market maker – Great liquidity – Complicated to explain – Satisfying experience to move market

What Next? • Never-ending internal promotion – Repeat customers – Revenue markets – Review-ratings

What Next? • Never-ending internal promotion – Repeat customers – Revenue markets – Review-ratings markets • Software improvements – – Democratize/simplify market creation Conditional markets Book orders Web services (simplify remote trading/analysis)