Information Sharing Using an Electronic Marketplace - PowerPoint PPT Presentation

1 / 13
About This Presentation
Title:

Information Sharing Using an Electronic Marketplace

Description:

Lack of information sharing played an important part of intelligence failure prior to 9/11 ... examples: eBay, monster.com and match.com ... – PowerPoint PPT presentation

Number of Views:62
Avg rating:3.0/5.0
Slides: 14
Provided by: bbn
Category:

less

Transcript and Presenter's Notes

Title: Information Sharing Using an Electronic Marketplace


1
Information Sharing Usingan Electronic
Marketplace
  • David Montana, Alice Leung and Marshall Brinn
  • BBN Technologies

2
Overview
  • Motivation
  • Background
  • Information sharing market
  • A game Find the Ring
  • Experiments
  • Conclusions

3
Information Sharing Requirements
  • Lack of information sharing played an important
    part of intelligence failure prior to 9/11
  • pieces of information existed, but could not
    piece them together
  • Obstacles to information sharing
  • inertia of separation (cultural and
    technological)
  • threats to security and privacy
  • need to know and paper trail have provided
    safeguards against too-wide dissemination
  • example of problem Robert Hanssen and FBI
    database
  • Required ability to share critical information
    across organizations fluidly as needed to detect
    terrorist plots but not too widely
  • Proposed solution well-regulated and restricted
    electronic market for buying and selling of
    intelligence information
  • incentivize and facilitate sharing while
    restricting information flow

4
Electronic Markets Background
  • Much work and attention focused on electronic
    futures markets as method of predicting future
    events
  • proven effective for true/false and
    multiple-choice type predictions in places such
    as Iowa Electronic Markets and Hollywood Stock
    Exchange
  • lack ability for more free-form type of
    predictions (where cannot enumerate the possible
    outcomes)
  • information aggregation without sharing
  • Electronic goods and services markets provide
    better model for information sharing
  • examples eBay, monster.com and match.com
  • ability to quickly identify complementary needs
    among a pool of thousands of potential buyers and
    sellers, to establish a market, and to easily
    perform any transactions
  • easily auditable trail of transactions

5
Information Sharing Market Framework
  • Participants buyers and sellers of information
    and data services
  • sellers submit sealed bids plus sealed
    descriptions of information
  • buyers submit sealed offers plus sealed
    descriptions of desired information
  • Broker matches buyers and sellers
  • if transaction realized, buyer receives copy of
    information from seller
  • Large reward for piecing together pieces of
    information into coherent whole drives the market

6
Information Sharing Market Issues
  • Advantages over bulletin board or open database
  • incentivizes sharing rather than hoarding of
    information
  • discourages wanton gathering of information
  • Large number of issues, including
  • security with participants having different
    privileges
  • matchmaking how to accomplish without revealing
    too much, particularly about what information is
    being sold
  • reputation discouraging sale of bad information
  • economic (focus for this paper)
  • what unit of payment? (e.g., fixed price vs.
    share of reward)
  • price of a transaction?
  • information about market to participants?
  • information rights (does seller retain ability to
    sell again)?
  • market dynamics?
  • Investigate economic issues using a game that
    includes an information market

7
A Game Find the Ring
  • Suspects R ring members among them
  • Large reward for player that identifies the
    members of the ring
  • Players each starting with M and C clues
  • Clues partial information about which suspects
    in ring
  • Equivalence (A and B either both in or both out)
  • Anti-equivalence (either A or B in but not both)
  • Negative (A not in ring)
  • Play proceeds in rounds, with each round
    consisting of two phases
  • players submit offers
  • sell offers minimum price plus clue
  • buy offers maximum price plus logical
    description of set of clues
  • identification offers guess the ring
  • broker resolves offers
  • penalties (loss of money) for wrong
    identification
  • reward for one (randomly chosen) correct id, and
    game over
  • for each player, find best sell offer for one buy
    offer and split price
  • for each player, find best buy offer for one sell
    offer and split price

8
Find the Ring (cont.)
  • Stalemate if game proceeds N rounds without a
    transaction
  • stalemate is a bad outcome, since ring not found
  • Information available to all players
  • number of suspects, size of ring, number of
    players, etc.
  • summary of game state including leader board
    and number and average price of transactions last
    round
  • Goals
  • individual maximize money at end of game by
    collecting reward and/or by receiving more for
    selling clues than spent buying
  • because reward not always given, not constant-sum
    game
  • game maximize probability that ring found

9
Web Version Screenshot
10
Strategies and Agents
  • Two main components of strategy (analyzed in
    detail in paper)
  • which clues to try to buy
  • important to know but specific to particular game
  • pricing on buy and sell offers
  • focus of economic analysis
  • relevance to information markets in general
  • Implemented software agents to play the game
  • know which clues to buy
  • optimal pricing strategy depends on what other
    players doing
  • provide small number of parameters that allow
    different components (buying, selling, stalemate
    backoff) of pricing strategy to vary between
    agents

11
Experiments
  • Experimental method
  • select particular parameters for the game (number
    of players, number of suspects, size of ring,
    reward, starting money, number of initial clues
    of each type) and generate many different games
    from these parameters
  • try different mixes of agent strategies until
    converge on equilibrium where no player/agent can
    improve by varying strategy
  • measure properties of game/market, particularly
    fraction of time that ring found
  • Vary parameters of game and see how this affects
    outcome

12
Results
  • When sales competitive, prices low and market
    quickly proceeds to solution conversely, when
    lack of sales competition (information monopoly),
    prices high and tendency to get stuck in
    stalemate
  • When players have sufficient money (relative to
    reward), even in competitive case likely to reach
    solution eventually however, when do not have
    sufficient money, tendency to get stuck in
    stalemate
  • Change in broker mechanism will change specifics
    of the pricing strategies but will not change two
    basic conclusions above
  • Conclusion for situation where non-competitive
    sales and reward-to-capital ratio is high, need
    some mechanism for breaking the deadlock, e.g.
    providing a way to split any future reward

13
Conclusions
  • Information sharing market provides a potential
    mechanism for information sharing that
    incentivizes and facilitates cooperation while
    maintaining a restricted and regulated flow of
    information
  • Find the Ring game provides mechanism for
    studying some of the properties of an information
    sharing market in a controlled setting
  • Using software agents and Find the Ring, we
    performed experiments investigating these
    properties and discovered some conclusions that
    can help general design of information sharing
    markets
  • Much work remains to be done before can implement
    real-world information sharing market
Write a Comment
User Comments (0)
About PowerShow.com