**today** Lecture: Wed June 21 12:30p, Joshua Thorp, Stephen Guerin: Frogs in a Blender...

[Stephen Guerin]

TITLE: Frogs in a Blender, Turtles in Processing: stirring an agent-based
modeling stew using Blender3D, Processing, and Netlogo on a stadium evacuation
project

SPEAKERS:
Joshua Thorp
Stephen Guerin
Owen Densmore (in spirit from Ireland)

AFFILIATION: RedfishGroup

LOCATION: 624 Agua Fria Conference Room
TIME: Wed June 21, 12:30p

Lunch will be available for purchase

ABSTRACT
Primarily a tech review about our recent experiences melding Blender3D
(http://www.blender3d.org) with Processing (http://www.processing.org) for an
agent-based modeling production pipeline. Netlogo-like language constructs were
developed in Processing for use in the interactive real-time models using 50,000
agents. Blender3D was used for stadium modeling and offline 3D
animation/rendering on a 730 CPU renderfarm.

We'll demonstrate the pipeline in the context of a Homeland Security project
visualizing crowd dynamics arising from simulated suicide bomb attacks at
Pittsburgh's PNC baseball Stadium.

[Stephen Guerin]

** tomorrow**

TITLE: Attention Allocation in Organizational Decision-Making

SPEAKER: Douglas A. Samuelson

AFFILIATION: Homeland Security Institute

LOCATION: 624 Agua Fria Conference Room
TIME: Wed August 9, ** 2:00p ** (non-standard time)

ABSTRACT: Consider how to improve organizational decision-making by streamlining
the process of seeking and allocating the attention of top decision-makers.
These decision-makers try to optimize the value they receive by allocating their
attention, taking uncertainty into account.  In fact, optimizing the benefits of
attention results, for the organization's original problem, in the well-known
"satisficing" behavior described by Herbert Simon.  In practice, the behavior is
often similar to the greedy heuristic for the knapsack problem: a few of the
largest topics and many small topics get addressed, while most middle-sized
topics are neglected until they become major problems.  As in the knapsack
problem, more clearly identifying sizes (time and attention required) and
values, and considering better ways to allocate space (attention available),
produces better results.  By encouraging persons familiar with particular issues
to "bid" for decision-makers' attention, giving short, clear estimates of
importance and complexity of the issue, and by then rewarding helpful initiative
while penalizing overbids, senior decision-makers can substantially decrease the
likelihood of overlooking major problems until they become crises.

Now consider agent-based models of teams of workers, each with a supervisor,
with problems arriving at random by a Poisson process.  A problem requires
certain skills and a certain number of units of effort for each needed skill.
Workers have skills and various numbers of units of work they can accomplish,
per skill area, per time period.  In alternative versions of the model, problems
may arrive at a central point where they are sent to team supervisors, or they
may drift through the organization's space until they encounter a team, or there
may be some group decision-making among team supervisors and an overall manager.
The simplest model is one team and problems arriving directly to that team's
leader; future work can expand in modular fashion.  The version of the model in
which problems arrive and drift through the organization's space randomly until
they encounter a team that can solve them appears to approximate - and explain -
the behavior of the Cohen, March and Olsen Garbage Can Model.  Other, more
hierarchical versions are likely to deadlock, overwhelming the managers and
unnecessarily idling many of the workers, in a manner that fit intuition for
certain large, tightly controlled bureaucracies.  Explicitly modeling the
attention required by managers and supervisors to assign problems and monitor
progress would add another level of complexity and realism.  This approach
appears to promise a rich variety of interesting results.


Presenter:

Douglas A. Samuelson is a senior analyst for the Homeland Security Institute,
Arlington, Virginia, and President of WINFORMS, the Washington, DC chapter of
the Institute for Operations Research and the Management Sciences (INFORMS.)
He has also been a Federal policy analyst, inventor, high-tech entrepreneur and
executive, and university faculty member.  He is perhaps best known for his
popular and long-running "The ORacle" column in OR/MS Today.   He has a D.Sc. in
operations research from George Washington University.

[Phil Henshaw-2]

I can't believe anyone talks this way.   With all due respect for the
offices and life and death issues sometimes involved, you need
responsive systems, not micro-managed decision makers.   What you want
is for the people with the ultimate responsibility to be free to poke
their noses into the process, any time, any where, at their leisure.  



Phil Henshaw                       ????.?? ? `?.????
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
680 Ft. Washington Ave
NY NY 10040                      
tel: [phone]                
e-mail: pfh at synapse9.com          
explorations: www.synapse9.com    

> -----Original Message-----
> From: friam-bounces at redfish.com
> [mailto:friam-bounces at redfish.com] On Behalf Of Stephen Guerin
> Sent: Tuesday, August 08, 2006 4:17 PM
> To: friam at redfish.com
> Subject: [FRIAM] **tomorrow** Lecture: Wed Aug 9 2p,Douglas
> Samuelson: Attention Allocation in OrganizationalDecision-Making
>
>
> ** tomorrow**
>
> TITLE: Attention Allocation in Organizational Decision-Making
>
> SPEAKER: Douglas A. Samuelson
>
> AFFILIATION: Homeland Security Institute
>
> LOCATION: 624 Agua Fria Conference Room
> TIME: Wed August 9, ** 2:00p ** (non-standard time)
>
> ABSTRACT: Consider how to improve organizational
> decision-making by streamlining the process of seeking and
> allocating the attention of top decision-makers. These
> decision-makers try to optimize the value they receive by
> allocating their attention, taking uncertainty into account.  
> In fact, optimizing the benefits of attention results, for
> the organization's original problem, in the well-known
> "satisficing" behavior described by Herbert Simon.  In
> practice, the behavior is often similar to the greedy
> heuristic for the knapsack problem: a few of the largest
> topics and many small topics get addressed, while most
> middle-sized topics are neglected until they become major
> problems.  As in the knapsack problem, more clearly
> identifying sizes (time and attention required) and values,
> and considering better ways to allocate space (attention
> available), produces better results.  By encouraging persons
> familiar with particular issues to "bid" for decision-makers'
> attention, giving short, clear estimates of importance and
> complexity of the issue, and by then rewarding helpful
> initiative while penalizing overbids, senior decision-makers
> can substantially decrease the likelihood of overlooking
> major problems until they become crises.
>
> Now consider agent-based models of teams of workers, each
> with a supervisor, with problems arriving at random by a
> Poisson process.  A problem requires certain skills and a
> certain number of units of effort for each needed skill.
> Workers have skills and various numbers of units of work they
> can accomplish, per skill area, per time period.  In
> alternative versions of the model, problems may arrive at a
> central point where they are sent to team supervisors, or
> they may drift through the organization's space until they
> encounter a team, or there may be some group decision-making
> among team supervisors and an overall manager. The simplest
> model is one team and problems arriving directly to that
> team's leader; future work can expand in modular fashion.  
> The version of the model in which problems arrive and drift
> through the organization's space randomly until they
> encounter a team that can solve them appears to approximate -
> and explain - the behavior of the Cohen, March and Olsen
> Garbage Can Model.  Other, more hierarchical versions are
> likely to deadlock, overwhelming the managers and
> unnecessarily idling many of the workers, in a manner that
> fit intuition for certain large, tightly controlled
> bureaucracies.  Explicitly modeling the attention required by
> managers and supervisors to assign problems and monitor
> progress would add another level of complexity and realism.  
> This approach appears to promise a rich variety of
> interesting results.
>
>
> Presenter:
>
> Douglas A. Samuelson is a senior analyst for the Homeland
> Security Institute, Arlington, Virginia, and President of
> WINFORMS, the Washington, DC chapter of the Institute for
> Operations Research and the Management Sciences (INFORMS.) He
> has also been a Federal policy analyst, inventor, high-tech
> entrepreneur and executive, and university faculty member.  
> He is perhaps best known for his
> popular and long-running "The ORacle" column in OR/MS Today.  
>  He has a D.Sc. in
> operations research from George Washington University.
>
>
> ============================================================
> FRIAM Applied Complexity Group listserv
> Meets Fridays 9a-11:30 at cafe at St. John's College
> lectures, archives, unsubscribe, maps at http://www.friam.org
>
>

[Stephen Guerin]

** today / non-standard time of 2p **

TITLE: Attention Allocation in Organizational Decision-Making

SPEAKER: Douglas A. Samuelson

AFFILIATION: Homeland Security Institute

LOCATION: 624 Agua Fria Conference Room
TIME: Wed August 9, ** 2:00p ** (non-standard time)

ABSTRACT: Consider how to improve organizational decision-making by streamlining
the process of seeking and allocating the attention of top decision-makers.
These decision-makers try to optimize the value they receive by allocating their
attention, taking uncertainty into account.  In fact, optimizing the benefits of
attention results, for the organization's original problem, in the well-known
"satisficing" behavior described by Herbert Simon.  In practice, the behavior is
often similar to the greedy heuristic for the knapsack problem: a few of the
largest topics and many small topics get addressed, while most middle-sized
topics are neglected until they become major problems.  As in the knapsack
problem, more clearly identifying sizes (time and attention required) and
values, and considering better ways to allocate space (attention available),
produces better results.  By encouraging persons familiar with particular issues
to "bid" for decision-makers' attention, giving short, clear estimates of
importance and complexity of the issue, and by then rewarding helpful initiative
while penalizing overbids, senior decision-makers can substantially decrease the
likelihood of overlooking major problems until they become crises.

Now consider agent-based models of teams of workers, each with a supervisor,
with problems arriving at random by a Poisson process.  A problem requires
certain skills and a certain number of units of effort for each needed skill.
Workers have skills and various numbers of units of work they can accomplish,
per skill area, per time period.  In alternative versions of the model, problems
may arrive at a central point where they are sent to team supervisors, or they
may drift through the organization's space until they encounter a team, or there
may be some group decision-making among team supervisors and an overall manager.
The simplest model is one team and problems arriving directly to that team's
leader; future work can expand in modular fashion.  The version of the model in
which problems arrive and drift through the organization's space randomly until
they encounter a team that can solve them appears to approximate - and explain -
the behavior of the Cohen, March and Olsen Garbage Can Model.  Other, more
hierarchical versions are likely to deadlock, overwhelming the managers and
unnecessarily idling many of the workers, in a manner that fit intuition for
certain large, tightly controlled bureaucracies.  Explicitly modeling the
attention required by managers and supervisors to assign problems and monitor
progress would add another level of complexity and realism.  This approach
appears to promise a rich variety of interesting results.


Presenter:

Douglas A. Samuelson is a senior analyst for the Homeland Security Institute,
Arlington, Virginia, and President of WINFORMS, the Washington, DC chapter of
the Institute for Operations Research and the Management Sciences (INFORMS.)
He has also been a Federal policy analyst, inventor, high-tech entrepreneur and
executive, and university faculty member.  He is perhaps best known for his
popular and long-running "The ORacle" column in OR/MS Today.   He has a D.Sc. in
operations research from George Washington University.