What really concerns me about Y2K? Emergence

greenspun.com : LUSENET : TimeBomb 2000 (Y2000) : One Thread

What really concerns me about Y2K? Emergence.


We have reached the point of polarization. Most people talking about Y2K are in one of two camps. The doomers or the pollys. A doomer says that really bad things are going to happen and a polly says that, at most, it will be a bump in the road. In fact, they may both be right.

Modeling a thunderstorm...a first step at understanding Emergence.

How does one go about simulating a thunderstorm in a computer model? Are there variables for the intensity and frequency of lightning? A variable for thunder volume? Or wind speed? How about rain fall amounts?


The "rules" used in a computer model to simulate a thunderstorm are surprisingly simple. Rather than defining the dramatic components mentioned above, the rules apply to each individual puff of air in the virtual atmosphere. A rule, for example, might determine how one puff of air reacts when the one on its left changes pressure. A consistent simple set of rules is applied to each puff of air (call them agents) and the model is let loose.

Over time, the interactions (call them connections) based on the simple rules between the puffs of air will simulate a sunny day or a damaging wind storm; which one occurs depends on the initial conditions defined in the simple rules. Change the initial conditions even slightly in the simplest rule and the outcome can be dramatically different.

Why is this so? The answer is "Emergence".


What does this have to do with Y2K? First, a few definitions:

Agents - These are the "individuals" in any system. In the example above, the agents are the puffs of air. Any number of things could be defined as agents; individuals in social structures, businesses in an economy, creatures in an ecosystem, workers in an office, computers in a network.

Connections - These are the relationships that connect one agent to another. In the case of the thunderstorm model, the simple rules on how to respond to a change in a neighboring air puff is a connection. In an economy, distribution systems are just one example of the many connections present.

Complex System - Simply put, a complex system is one where there are more connections than agents. That is, each agent has connections to more than one other agent. In other words, the system is not linear.

In our thunderstorm example, each puff of air could be expected to interact with at least the one on the left and the one of the right. If each single puff of air interacts with more than one neighbor, there must be more connections than agents. Therefore, the system is, by definition, Complex.

Complex Adaptive System - This is a complex system where each agent can dynamically modify its behavior over time. A great example of a complex adaptive system is the Internet.

Emergence - Emergence is the phenomenon within a complex system where a behavior develops at a particular scale that is not present at other scales in the system. Going back to the thunderstorm example will make this clear. Let us say that the model is run and a thunderstorm is simulated. At an aggregate level, the system is behaving like a thunderstorm. The thunderstorm behavior is said to be emergent because each individual puff of air does not contain the behavior. Recall that the behavior of each agent is defined by the set of simple rules. (And these rules don't define a thunderstorm).

Scale - I used the word 'scale' in the definition of emergence. Think of scale as the level of magnification used to look at something. For example, let's say you were focused on the entire virtual atmosphere in the thunderstorm model. That is, you were standing back and observing the behavior of the all the agents at once. Now lets say you changed your focus and zoomed in on an individual puff of air. You just changed the scale at which you are observing the system.

Ok, ok. What does this have to do with Y2K already?

One interesting thing about Emergence is that it is extremely difficult to predict even in computer models, let alone the real world. Often, a modeler will change a simple rule or initial parameter in their model of a complex system and end up being very surprised at the resulting emergent behavior.

To further complicate things, emergent behavior is often not visible at a scale we are used to looking at. That is to say, you cannot look at a single agent (or even all the agents at a single point in time) and see even a hint of the latent emergent behavior. It simply doesn't exist at that scale. One more time to our example: If you looked at a single puff of air in the model and the rules it follows, you would not see anything even closely resembling a thunderstorm.

So here it is. All of us live in one great big complex adaptive system. To be more accurate, we all live in many complex adaptive systems that are interconnected to different degrees and at different scales.

Y2K can be expected to affect everybody to some degree or another. Some countries (states, companies, individuals, etc.) are further along than others in remediation and testing. However, no testing can be expected to be perfect. Gartner Group estimates that 5 percent to 9 per cent of defective lines of code will remain after remediation and testing have been completed.

For the first time in our history, we are experiencing a phenomenon that has the potential to affect all the agents in all the systems at one time. Without getting into it, I believe that a month or even a year window could be considered "at one time". It doesn't all have to happen in one second; 12:00:01 AM 1/1/2000.

In order for our world to end up looking very different than it does now, the power grid in the United States doesn't have to crash. Forty percent of small businesses don't need to fail. The international currency market doesn't need to halt. The sky doesn't need to fall.

The initial conditions in the entire system are about to be reset all at once. The "model" will inevitably march on. Even small changes in the "simple rules" that apply to us all can produce dramatic changes.

No one could possibly have any idea what will Emerge.





-- Pete (pberry1_98@yahoo.com), November 28, 1999


Perry; This is very well stated. Well done, well said. Thank you. You are a systems geek of the highest order...

Takes one to know one, eh?

-- Osiris (...@.......), November 28, 1999.

Excuse me, my stupidity seems to have emerged... I should have said, "Pete".

-- (...@.......), November 28, 1999.

Pete: I am familiar with such models. Simplified, it says [to me], when you take a big computer and no data [we have very little on Y2K] and attempt to model a complex system, you will get "interesting" results. "May you live in interesting times".

Best wishes,,,

-- Z1X4Y7 (Z1X4Y7@aol.com), November 28, 1999.

That was brilliant; thanks.

Layman translation for scale: "You can't see the forest for the trees"

-- Hokie (nn@va.not), November 28, 1999.

Thank you for this well put overview. Systems, systems, systems! We have no choice but to wait and see, (which does not preclude preparation).

The map is not the territory, is it?

-- Donna (moment@pacbell.net), November 28, 1999.

Excellent Read!!!

To further complicate things, emergent behavior is often not visible at a scale we are used to looking at. That is to say, you cannot look at a single agent (or even all the agents at a single point in time) and see even a hint of the latent emergent behavior. It simply doesn't exist at that scale.

Emergent Behavior i.e. In Retrospect

-- Emergence (karlacalif@aol.com), November 28, 1999.


"No one could possibly have any idea what will Emerge."

How about:

Something other than what we had prior to "the change".

-- GoldReal (GoldReal@aol.com), November 28, 1999.

Excellent analogy Pete!

This gives us all a new perspective on the problem. Resetting all the conditions on the same system eh?

I'll have to think some more about the implications of such...

-- (Kurt.Borzel@gems8.gov.bc.ca), November 29, 1999.

Moderation questions? read the FAQ