On Social Science as a Laboratory Science
To
@ajbouh
@anelsona
@antlerboy
@corneil
@fluffbuster
@jessitron
@mamund
@Merrion
@PezeshkiCharles
@startuployalist
@swardley
@yrashk
and others who are helping me put these thoughts into words. Thank you, and let’s keep it going.
1/27
To
@ajbouh
@anelsona
@antlerboy
@corneil
@fluffbuster
@jessitron
@mamund
@Merrion
@PezeshkiCharles
@startuployalist
@swardley
@yrashk
and others who are helping me put these thoughts into words. Thank you, and let’s keep it going.
1/27
I have been using Twitter to capture fragments of a larger research concept as these fragments show up, and I am slowly collecting and combining them into a coherent thesis. This essay is the next step in that process.
2/27
2/27
I hope that it reveals my destination: a larger fulfillment of this mission statement in my Twitter profile: “Systems thinking applied to human activity”.
3/27
3/27
This work started taking its present shape with the revelations of @carolecadwalla and @profcarroll, with her TED talk
ted.com/talks/carole_c…
and @thegreathackdoc, augmented by the results of the investigation led by Robert Mueller.
4/27
ted.com/talks/carole_c…
and @thegreathackdoc, augmented by the results of the investigation led by Robert Mueller.
4/27
I realized that the 2016 Presidential election in the US and the Brexit referendum were related instances of creating a significant social effect with relatively small, strategically placed, inputs. The bad guys understand the pressure points, and we can't defend ourselves.
5/27
5/27
This imbalance persists to this day, and we are reduced to arguing about securing election sites. To put it in pseudo-scientific jargon, the other side understands how to change the state of a nonlinear system with very little input energy, ...
6/27
6/27
... and we are stuck in a linear-system mindset, trying vainly to push back on the system’s outputs. I wrote about this here, for example:
.
7/27
.
7/27
Strangely enough, the turn in my thinking was pushed by two independent lines of research on the origins of life.
In one case I attended a talk by Jeremy England at MIT.
8/27
In one case I attended a talk by Jeremy England at MIT.
8/27
To oversimplify, England is running “in silico experiments” (computer simulations) to reproduce the effect that was shown in the famous Miller-Urey experiment
en.wikipedia.org/wiki/Miller%E2…
that created amino acids out of a simple primordial soup plus inputs of energy.
9/27
en.wikipedia.org/wiki/Miller%E2…
that created amino acids out of a simple primordial soup plus inputs of energy.
9/27
England is a physicist, and his talk at the Karolinska Institute
is framed in terms of thermodynamics. But I saw in some of the results he showed at MIT that there is an information domain that parallels the energy domain.
10/27
is framed in terms of thermodynamics. But I saw in some of the results he showed at MIT that there is an information domain that parallels the energy domain.
10/27
I signed up for the @ComplexExplorer course “Origins of LIfe”
complexityexplorer.org/courses/95-ori…
...
11/27
complexityexplorer.org/courses/95-ori…
...
11/27
... and was struck by physicist Eric Smith’s lecture on Nonequilibrium Physics
complexityexplorer.org/courses/95-ori…
in which he explained in physics terms phase change as a general phenomenon, and introduced to me the concept of fracture propagation.
12/27
complexityexplorer.org/courses/95-ori…
in which he explained in physics terms phase change as a general phenomenon, and introduced to me the concept of fracture propagation.
12/27
I reframed this into the social domain and tweeted about it:
.
13/27
.
13/27
While all this is going on I’m translating this new information into the domain of interconnected state machines because of the earlier influence on my thinking of @stephen_wolfram’s book “A New Kind of Science”
wolframscience.com/nks/,
which impressed me greatly.
14/27
wolframscience.com/nks/,
which impressed me greatly.
14/27
The underlying idea of Wolfram’s book (as I took it) is that you can build just about anything out of some combination of interconnected state machines. You can play with this idea: check out the Game of LIfe
en.wikipedia.org/wiki/Conway%27…;
also search “game of life software”.
15/27
en.wikipedia.org/wiki/Conway%27…;
also search “game of life software”.
15/27
This is the research question I’m addressing:
Can we learn how systems built out of interconnected state machines generate emergent behavior? If so, perhaps we can understand our own society better as a system that we can study with simulation, just as England is doing.
16/27
Can we learn how systems built out of interconnected state machines generate emergent behavior? If so, perhaps we can understand our own society better as a system that we can study with simulation, just as England is doing.
16/27
I believe that this research question is a broad one, much broader than politics; in fact it might be near-universal, as I have suggested in this thread about education and systems thinking.
.
17/27
.
17/27
I would like to see built a class of simulators that could be used on a variety of problems, based on this template:
.
An important aspect of this template is that it ties together *two* levels of phenomena, Things and Emergents (aggregates).
18/27
.
An important aspect of this template is that it ties together *two* levels of phenomena, Things and Emergents (aggregates).
18/27
Note that Systems Dynamics simulations
systemdynamics.org/origin-of-syst…
are one-level simulations at the aggregate level, in which behaviors such as flows are controlled by functions whose variables are properties of the aggregates, not of the elements of the aggregates.
19/27
systemdynamics.org/origin-of-syst…
are one-level simulations at the aggregate level, in which behaviors such as flows are controlled by functions whose variables are properties of the aggregates, not of the elements of the aggregates.
19/27
In the two-level simulations I envision, behaviors are controlled by messaging among Things, and Emergents are more like statistics, computed functions on groups of Things.
20/27
20/27
A key idea buried in this template is the concept of “lateral signaling”, expressed within a simulator in the messages that the Things send to each other. I see lateral signaling as a key mechanism in the creation of emergent behavior.
21/27
21/27
Here is a simple example
whose essence is captured in this reply
.
22/27
whose essence is captured in this reply
.
22/27
I first mentioned the relationship between lateral signaling and emergent phenomena in connection with terror attacks in this tweet:
.
23/27
.
23/27
Lateral signaling is the mechanism that has been employed to influence group behavior in social networks. It should be possible to study it through simulation.
This hints at a larger ambition I see for this work.
24/27
This hints at a larger ambition I see for this work.
24/27
I see simulation, using two-level models such as I have suggested, turning the social sciences into laboratory sciences, similarly to what Jeremy England is doing with biology and Jay Forrester did with management consulting.
nytimes.com/2016/11/18/tec…
25/27
nytimes.com/2016/11/18/tec…
25/27
This graphic suggests that simulation and physical experiment have alternative, even interchangeable, roles in improving our models.
It’s based on a contribution of Karl Popper: testing of falsifiable hypotheses is what drives scientific knowledge forward.
26/27
It’s based on a contribution of Karl Popper: testing of falsifiable hypotheses is what drives scientific knowledge forward.
26/27
en.wikipedia.org/wiki/Falsifiab…
I see model-based simulation as part of the future of social science, uniting in a common experimental methodology what have previously been traditionally distinct disciplines.
27/27
I see model-based simulation as part of the future of social science, uniting in a common experimental methodology what have previously been traditionally distinct disciplines.
27/27
@pdfmakerapp grab this
