Arnold Kling  

Complication vs. Complexity

PRINT
The Labor Market for Philosoph... Forecasting for a Swing...

In an interview, Ben Ramalin says,


We treat complex things as if they were merely complicated... distinguished between complicated systems, which can be modeled mathematically, and complex systems, for which there is no mathematical model which can say, if X is the situation then do Y. Sustainability, healthy communities, raising families have all been given as examples of such complex systems and processes. Peacebuilding would be another, women's empowerment, natural resource management, capacity building initiatives, innovation systems, the list goes on and on. Complexity science pulls back the curtain on these processes and it can force you to think about the world you live in in a different way.

People tell me that I would really like complexity theory. But if all it tells me is that, say, macroeconomics is too nonlinear and dynamic for us to be able to build mathematical models to manage the economy, then I know that already. Does it tell us anything more?


Comments and Sharing





COMMENTS (6 to date)
Fabio Rojas writes:

At its best, complexity theory is a set of models that explain how piles of small things assemble themselves into bigger things. There is no strict definition, it's more of a philosophy about modeling. I like to think about it in terms of contrasts:

- individual choices vs. aggregate patterns
- closed form solutions vs. simulations
- exogenous variables vs. feedback loops
- continuous variables vs. combinations of discrete things
- homogenous agents vs. heterogeneous agents
- one "level" (e.g., micro) vs. multiple interacting levels (e,.g., micro-macro interactions)
- simple stable agents vs. complex, possibly, evolving agents

So "complexity" isn't really a single theory that will immediately tell you something new about macro. Instead, it's a framework for developing different kinds of models, which *might* be useful. Complexity doesn't bother with equilibria or closed form solutions, but they're nice if you actually have them. Instead, it's about computationally estimating qualitative properties of systems. It's then up to people to figure out if such models are accurate or helpful.

Dave writes:

I have been asked Russ Roberts a few times over the last few years to interview someone from George Mason's Center For Social Comlexity or the Sante Fe Institute on Econ Talk.

I find it odd that some Masonomists call for reform of mainstream macroeconomics, speak of the economy as if it were a complex adaptive system resulting from the actions of individuals, but still seem to dismiss complexity science and agent based modeling out of hand. Why not invite them for a discussion and challenge them on their methods directly? You and Russ can tag team one of them in an interview.

fundamentalist writes:

The Complexity Economics article in Wikipedia makes it sound promising.

Dave writes:

I need to hire a proof reader. My comment should start:

"I have asked Russ Roberts..."

Troy Camplin writes:

It explains how self-organization occurs (in economic terms, how microeconomics gives rise to macroeconomics), and it explains how and why such systems are in fact orderly and are creative and give rise to greater complexity.

Really, many more economists need to be using this stuff than are currently using it.

jc writes:

When I think of complexity and/or chaos theory, I immediately think of Hayek, spontaneous order, and man's fatal conceit.

Regarding useful insight and/or prescription, "it's too complex to predict" doesn't necessarily mean we can't gleam insight and/or prescribe.

This is probably more strategic management than econ, but I've seen studies, for example, that suggest probing in high velocity industries. Rather than predicting what people will want and what you'll be great at and then putting most of your eggs in that basket, firms should allocate more modest resource levels to a bunch of small baskets - probes - and then see what happens.

After seeing what happened, *then* allocate higher resource levels towards the promising probes, honing them and either building them up on the fly until you reach a critical mass in terms of economies of scale, or simply making innovation via probing a core competency.

Consider this Meg Whitman (CEO, Ebay) quote: "This is a completely new business, so there's only so much analysis you can do... It's better to put something out there and see the reaction and fix it on the fly. You could spend six months getting it perfect in the lab... But we're better off spending sixe days putting it out there, getting feedback, and then evolving it." (Pfeffer & Sutton, 2006, HBR)

Fwiw, Intel's Andy Grove is also a big proponent.

Another thing that comes to mind, in policy venues, is how the U.S. used to be a big laboratory. Rather than creating a massive federal program that may or may not work as intended, states may experiment (i.e., they may do the probing). If something works as intended or if (ala Asimov's assertion that the most oft uttered thing after scientific discovery is not "Eureka!" but "Huh, that's funny...") it works differently than intended but in a beneficial way, *then* we consider making the entire country adopt it.

Comments for this entry have been closed
Return to top