Russell Thomas

@MrMeritology

Natural Philosopher, Security Data Scientist, PhD Candidate: George Mason, Computational Social Science. Tweets are my own.

Silicon Valley, USA
exploringpossibilityspace.blogspot.com
Joined October 2009
316 Photos and videos Photos and videos

Tweets

@MrMeritology is blocked

Are you sure you want to view these Tweets? Viewing Tweets won't unblock @MrMeritology.

  1. 7 hours ago

    Sometimes, adding more players increases stability of game outcomes (long-period cycle in 4-player)

    Embedded image permalink
    Embedded image permalink
    Embedded image permalink
    8:37 PM - 13 Jan 2016 · Details
    1 like
  2. Retweeted
    8 hours ago

    It doesn't come naturally to me or other academics, but finding good abstractions is a valuable skill that I am trying to develop.

    View conversation Hide conversation
    1 retweet 2 likes
  3. Retweeted
    9 hours ago

    New blog post: Can professional certs be bad? Professional Certifications & Behavioural Economics.

    View summary Hide summary
    4 retweets 5 likes
  4. Retweeted
    15 hours ago

    Our salon on civics in a distributed society with and is starting soon. Stream at

    2 retweets 2 likes
  5. Retweeted
    Jan 12

    EMFASE project - calling security experts to complete comprehensibility of risk models survey

    1 retweet 2 likes
  6. Jan 13

    16 defenders (green), 4 attackers (red), 184 2-player games (100X100) 1,434 ticks and no equilibrium in sight

    Embedded image permalink
    12:45 AM - 13 Jan 2016 · Details
    3 likes
  7. Jan 12

    Here, there are 100 possible moves for each player.

    View conversation Hide conversation
  8. Jan 12

    Next step: introduce asymmetric payoffs btwn attack and defense. Probably lognormal. THAT should make for some fun learning dynamics!

    View conversation Hide conversation
    1 like
  9. Jan 12

    So only difference right now btwn attackers and defenders is the topology of network of games.

    View conversation Hide conversation
  10. Jan 12

    BIG BUT: the payoffs are not set to match cyber security. Currently these are bivariate normal distributions, mean = 0, sd = 1.

    View conversation Hide conversation
  11. Jan 12

    Sample run: 2 agents of12. Green=defender Red=attacker Attacker strategies fairly stable vs Defender oscillations

    Embedded image permalink
    11:37 PM - 12 Jan 2016 · Details
    1 retweet 1 like
  12. Jan 12

    Up to half of players can be attackers, and there are parameters to control for "degree of antagonism" btwn attack/defense, defense/defense

    View conversation Hide conversation
  13. Jan 12

    The current parameter range is up to 20 players and up to 400X400 games. It all works, but is CPU bound. Only a few ticks/second.

    View conversation Hide conversation
  14. Jan 12

    Of course, this certainly doesn't LOOK like any cyber security game most people would recognize. UI is more like MRI + EKG

    View conversation Hide conversation
  15. Jan 12

    It's slow as molasses in January, and many features not in yet. Much performance tuning will be needed, (likely) port to native Java + MASON

    View conversation Hide conversation
  16. Jan 12

    An alternate view w/ agent-to-agent interactions (60 2-player games, total)

    Embedded image permalink
    11:11 PM - 12 Jan 2016 · Details
    1 like
  17. Jan 12

    Agent-based Cyber Security Investment Game in Netlogo (8 defenders, 4 attackers, 200X200 games, on-line learning)

    Embedded image permalink
    11:10 PM - 12 Jan 2016 · Details
    2 retweets 7 likes
  19. Retweeted
    Jan 12

    We are at our strongest when we view our nation's challenges and opportunities not as R vs. Ds, but as the future vs. the past

    58 retweets 85 likes
  20. Jan 12

    OK. New rule for next . Everybody has to breath nitrous while watching.

@MrMeritology hasn't tweeted yet.

Loading seems to be taking a while.

Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

    You may also like

    ·