Game theory is a mathematical discipline that investigates the interaction of multiple, interest driven and rational parties. In other words: Most of our business and social interactions. In this talk we will define some basic game theory terms, talk about some of the more iconic games that have been developed by the discipline and see how they apply to most of our product strategy decisions. We’ll talk about Prisoner’s Dilemma, Rock Paper Scissors and the Game of Chicken – describe business scenarios where they’re applicable and come up with the best solutions, together!
3. What?
• A primer on basic game theory concepts
– Basic
– We try to avoid math
– If you know a lot about Game Theory already, you
may get bored
• A discussion on applying game theoretic
concepts to product strategy
4. Why?
• I think Game Theory is Fun!
• Interesting and illuminating perspective on
behavior of both people and firms
• Analytical toolset for competitive scenarios
5. Who?
• From Israel
• Student of Behavior and Business
• GTM Consultant for multiple startups
• Previously PdM and Strategist in Dell and
EMC and a variety of software development
roles
• Winner of the non existent “crappiest slide in
PCA15” award
6. Agenda
• What is Game Theory?
• A Few Games and Business Analogies
– Rock, Paper, Scissors:
• Zero-sum, symmetric game
– Prisoner’s Dilemma:
• Pricing, market entry, volume
• Iterated Prisoner’s Dilemma: TFT
• Collusion/Cooperation in IPD – cartels
– Chicken and the Value of Commitments
• Investment, market entry
7. What is Game Theory?
• Mathematical discipline designed to answer questions
like:
– You are playing Rock Paper Scissors. Which do you choose?
– Can some games be “solved”? (Tic Tac Toe, Chess, Go?)
– Should we build more nuclear weapons in the cold war or
less? How do we ensure that a disarmament agreement is
enforceable?
• Developed mathematically by Von Neumann (of the
Manhattan Project fame), Nash (Of “A Beautiful Mind”
fame), and Morgenstern (not as famous, but definitely
deserves props) during 40s and 50s.
8. So, what does that have to do with
me?
• You sell a product that is very similar to your
competitor. Do you:
– Price low to steal his customers?
– Price the same?
– Price higher?
• You need to invest in a factory to create your product
• You open a gas station – should you locate it across the
street from your competition or the other side of
town?
– Who can give an example of something more tech-y that is
gas-station-esque?
9. Types of Games
• Sequential vs. Simultaneous
– Sequential: Chess, Checkers, Tic-Tac-Toe
• Solvable
• Game tree solution
– Simultaneous: Rock Paper Scissors
• Solvable
• Equilibria solutions
• Most business problems are simultaneous (or very close to
it) rather than sequential
• Our focus will be simultaneous, 2 player games
10. Brain Teaser – Cut the Cake
• Two twin brothers have gotten a cake from
their parents for their birthday.
• What is the best way for them to split it so
they are both happy?
11. Rock, Paper, Scissors
This is a game
Rock Paper Scissors
Rock 0 , 0 -1 , 1 1 , -1
Paper 1 , -1 0 , 0 -1 , 1
Scissors -1, 1 1 , -1 0 , 0
So how do I read it?
• The first number in each
pair is the payout for the
row player – the second is
for the column
So, who cares?
• This is a boring game – but
it’s here as an example for
more interesting ones
• Still – interesting to note:
– Symmetric
– Zero Sum
12. Rock, Paper, Scissors
Rock Paper Scissors
Rock 0 , 0 -1 , 1 1 , -1
Paper 1 , -1 0 , 0 -1 , 1
Scissors -1, 1 1 , -1 0 , 0
So, again, I ask, who cares?
• RPS is a classic example for
a zero sum, symmetric
game
– Zero sum: Winning is done at
the expense of the loser
– Symmetric: Identical options
with identical payoffs exist to
all players
13. Let’s Play!
Find a partner – and play Rock, Paper, Scissors!
Rock Paper Scissors
Rock 0 , 0 -1 , 1 1 , -1
Paper 1 , -1 0 , 0 -1 , 1
Scissors -1, 1 1 , -1 0 , 0
14. What have we learned?
• Does RPS have much in common with real
business problems?
• What’s the best strategy for playing RPS?
15. Prisoner’s Dilemma
• A classic!
• Probably most famous game in Game Theory
• We are going to spend some time here, so get
comfy
16. So, what’s the big deal?
(positive
payoffs are
desirable)
Shut
up
Snitch!
Shut
up
3 , 3 0 , 5
Snitch! 5 , 0 1 , 1
• Two burglars are busted
and put in separate
rooms
• Police interrogate and
promise the first one to
snitch a deal
• Now, things get
interesting.
• What would you do?
17. Equilibrium
(positive
payoffs are
desirable)
Shut
up
Snitch!
Shut
up
3 , 3 0 , 5
Snitch! 5 , 0 1 , 1
• The problem should be
clear
• Everyone would be better
off shutting up – but each
individual would rather
snitch
• Therefore, the Nash
Equilibrium (and therefore
the result of the game if
played by rational, payoff
maximizing players) is the
sub optimal bottom right
corner
18. What does PD apply to?
• Can you think of any real life scenario in which
PD applies to?
19. A ton of them!
• Pricing/Promotions: We are selling a similar
product; should I price low (and then steal share)
or high?
– http://www.businessweek.com/articles/2013-11-
19/best-buy-and-the-holiday-retail-prisoners-dilemma
• “Feature Wars”
• In fact, most non-zero sum games in which
cooperation yields more benefits than competing
can be modeled like a PD game
20. So what do we do?
• The bad news: The solution to a standard,
“pure” Prisoner’s Dilemma game is always
mutual defection. Sorry. It’s a cruel world.
• The good news: Reality is infinitely more
complex; there are factors we can explore and
ways to “change the game”
– IPD
– Collusion (don’t do that!)
– Change the Game
21. Let’s try something
• Pick your gaming partner
• Play the game!
(positive
payoffs are
desirable)
Shut
up
Snitch!
Shut
up
3 , 3 0 , 5
Snitch! 5 , 0 1 , 1
22. Show of Hands
• Who picked snitching? Why?
• Who picked shutting up? Why?
23. And now, a Twist!
• Play the game again.
• And again.
• And again.
• Until I tell you to stop.
24. Iterated Prisoner’s Dilemma
• An interesting result emerges when playing PD
multiple times. Why?
• What if I told you that you had exactly 100 turns
to play?
• Strategies: TFT, TF2TT
• Axelrod’s Successful Strategy:
– Clear
– Provocable
– Nice
– Forgiving
25. Collusion
Careful! I’m not a lawyer, but usually, this is illegal
Excerpt from Cramton & Schwartz 2002 – see table 1 at
http://drum.lib.umd.edu/bitstream/1903/7061/1/cramton-schwartz-collusive-bidding.pdf
(can’t put table here for copyright reasons)
26. Change the Game
• What are other options to change the game?
• What if you locked the loot from the burglary
with two keys?
• Examples from a Product Strategy perspective:
– Differentiation
– Consortiums (pooled standard ownership for IP)
– Punitive clauses in contracts
– Price matching clauses (retroactive ones as well)
– Most favored customer clauses
27. Let’s Play Chicken!
Swerve Drive on!
Swerve Lose (-1) ,
Lose (-1)
Lose (-1),
Win (1)
Drive on! Win (1), Lose
(-1)
Splat! (-3) ,
Splat! (-3)
• Anti-cooperation game
• What are business
analogies?
• How can you win?
28. The Factory Game (A version of
Chicken)
• What’s the Nash
equilibrium?
• How do you win?
Build Not Build
Build Competition
: 1, 1
Monopoly: 2,
0
Not
Build
Monopoly:
0, 2
Nothing: 0, 0
29. Some interesting questions if we have
time left
• You need to bid on ad keywords. Are you
better off having keywords that are identical,
or different from your competition?
• Do you think it’s better to ape your
competitors features or remove them
entirely?
• When you build a store location – would you
rather build it next to your competitor, or
somewhere else?
30. Hotelling’s Problem – or why
Differentiation doesn’t always Work
The setting – the beach; the adversaries – two hot
dog cart vendors.
• Where should they place their carts?
• What is the socially optimal placement?
• What is the actual equilibrium?
• Why do I care?
Differentiation only matters when it’s difficult or costly to copy
31. Do We Have Time for One More?
• This is a game called “Keynesian Beauty
Contest”
• The story is that of a beauty contest: Given
faces published in a newspaper, pick the
prettiest face. The winner is the one who
picked the most popular face.
• We’ll model this here in a variant.
32. Pick a card, any card
• The options are numbers from 1-10
• The winner is the person who guesses closest
to 2/3 of the average
• Ready? Go!
Either scan the QR code with a QR
code scanning up – go to :
http://shortn.me/kFiK
Or text: GAMES to (952) 649-5350
36. Further Reading
• Art of Strategy: Dixit & Nalebuff
• Evolution of Cooperation: Axelrod
• Prisoner’s Dilemma: Poundstone
• Coursera: Game Theory, Advanced Game
Theory
Notas do Editor
No Nash equilibrium or minimax theory (unless you really want to); no game trees; no dominant and dominated strategies; no games with imperfect information – just the basics.
Here’s a hint – not many business problems are zero sum ; there is almost always a way to grow the pie.
Assuming perfectly rational players – mixed 1/3 1/3 1/3 strategy is Nash equiblirium
Yes, has to be symmetric
Business analogies – (factory next slide)
Commitment signaling (option limiting)
Commitment signaling (Build a larger factory than you have to – means that you can undercut price and try to maintain money)
What are some examples that don’t involve physical goods?
Website launch in a locale
Setting up a large sales office
Investing heavily and publicly in a product feature that both companies want
Visibility is key for signaling to work