Network markets for digital goods Free riding and competition Andreas U. Schmidt Fraunhofer...

10
Network markets for digital goods Free riding and competition Andreas U. Schmidt Fraunhofer Institute for Secure Information Technology SIT, Darmstadt, Germany ontained is for educational purpose only. All trademarks are property of their respective holders.

Transcript of Network markets for digital goods Free riding and competition Andreas U. Schmidt Fraunhofer...

Network markets for digital goods

Free riding and competition

Andreas U. SchmidtFraunhofer Institute for Secure Information Technology SIT, Darmstadt, Germany

The material contained is for educational purpose only. All trademarks are property of their respective holders.

HICSS-41, Andreas U. Schmidt, Network Markets 2

Markets for digital goods: copyright waring or economy

digital goods

transferable &non-rival

durabilityno wear & tear

creation can be costly

free ridingpirate sites

Bittorrent

file sharing nets

warez & cracker sites

p2p nets

copy protection

DRMimpedefair use

DMCA

WIPO treaty

EU directive

national laws

online music

video on demand

ring-tonessoftware asa service

criminalise the clientele

restore features of physical goods

some majors defect

taxation

cultureflat-rate

GEMA network marketsfor digital goods

super-distribution remuneration

SnocapMashBoxx PeerImpact

Potato

HICSS-41, Andreas U. Schmidt, Network Markets 3

Economic qualities of network markets

Ponzi schemes

Peter-and-Paul scams

chain letters

pyramid selling

Inventory loading

externalise distribution cost

late entrantsare penalised

snowball systems

externalise risk

favour recruititionover resales

digitalgoods

if there is a free version and it is common knowledge that some (late-comers) pay the rent, then an inductive argument shows that the market is empty

as with pure chain letters, the success of network markets for digital goods may depend on bounded rationality of buyers

what revenues from resale can an agent expect ?

HICSS-41, Andreas U. Schmidt, Network Markets 4

The core model for expected resale revenues

a flux model

buyers enter continuously

remaining revenues go to a collector

agents compete with each other for new buyers

There may be transaction costs

over time until the market is saturated

pay a certain price to the reseller

satisfies a conservation law / zero-sum condition

reparameterisationby market saturation

mild error behaviourw.r.t. discrete model

can be extended to multiplerewarding levels (by a Markovproperty)

the map from price to incentive is invertible

dynamical forward pricing

mechanism design

fair reward / incentive schedules

competition against free-riders?

enabling

scale free!incentive is independent ofabsolute market size

HICSS-41, Andreas U. Schmidt, Network Markets 5

Examplesearly adopters are favoured

nonzero price at s=0 entailsartefactual singularity

early subscriber discount

rebate for late adopters mitigates the penalty

letting π(1)=0 effectively closes the market

invitation to enter during an initial period

taxation by the collectordoes not hurt the incentivetoo much

multiple levels benefit all agents

late-comers are penalised

HICSS-41, Andreas U. Schmidt, Network Markets 6

Modeling duopoly competition

which externalities influence buyer decision?

subtracting the utilities of A and B and summation yields the bias of agents to buy A

exogenous factors

a. price b. popularity

in a duopoly network market

endogenous factors

2. genuine multiplier externality,tuned with parameter ε

monopoly resalerevenues

estimated probability that others buy, based on popularity alonebounded rationality

1. reward expectation with bounded rationality

HICSS-41, Andreas U. Schmidt, Network Markets 7

From bias to probabilities to dynamicsgiven the bias Δ to buy A, how to calculate actual probabilities?

choose a ‘natural’ distribution of the subjective utility of both goods

ΔB A

separate them by the bias

calculate the dynamics

HICSS-41, Andreas U. Schmidt, Network Markets 8

Competition with free riders – typical dynamics

price is a spike peaking at m, leading to different incentive schedules

free-riders: pA=pB, πB=0

shares, turnovers

and collector’s shares

can be observed

initial invitation to enteryields strong initial growth

more extended and amplified by multiplier effect as m increases

while large m are optimal w.r.t shares, turnovers suffer from the long rebate period

substantial growth at high s due to the multiplier effect mitigates this

HICSS-41, Andreas U. Schmidt, Network Markets 9

Competing against free riders

m=0.1 m=0.5 m=0.9

HICSS-41, Andreas U. Schmidt, Network Markets 10

Open issues and further work

ABCE

put network marketing of digital goods with dynamical forward pricing to the real-world test!

waiting costs

s=1 singularity

dynamical forward pricing strategies and implementation

restrictions on resale

explicitly scale-dependent effects

rushing /sniping ?

informationavailability

market homogeneity

enable marketingfor resellers