Friday, September 06, 2013

Allowing Little Traumas Prevents Major Collapses

Image Courtesy of the USDA
When the stock market collapses, a country erupts in rebellion, an electric grid blacks out or a child throws a tantrum in a crowded supermarket, the people nominally in charge of the systems often try harder to rein things in. But a new paper in Physical Review Letters shows that too much control is often at the heart of a major breakdown.

It's not an entirely new idea. Many people have noted that major wildfires are more devastating in areas where small fires are vigorously suppressed. But it wasn't necessarily clear just how much you should let small fires burn to ensure that the total damage caused by lots of little fires is less than that caused by one huge, occasional firestorm. Similar questions exist for just about any system that can experience avalanche-like breakdowns.

Now, Pierre-Andre Noel, Charles D. Brummitt and Raissa M. D’Souza of the University of California, Davis have developed a model that could clarify just how much control you should relinquish over small traumas in order to avoid a major collapse.

The researchers developed their model by looking at sand piles. Slowly adding grains to a pile can lead to avalanches of various sizes. Typically, a sand pile will experience very many small avalanches and only occasional major avalanches. If you were to find a way to entirely suppress small avalanches, you would eventually end up with a pile on the verge of a massive collapse. This is the sort of problem that occurs if firefighting crews put out every fire in a forest: eventually there's so much tinder that the whole thing goes up at once.

On the other hand, allowing every fire to burn could lead to more damage and expense than putting out some and letting others burn. This is comparable to stopping as many small avalanches as possible without creating a huge unstable pile.

The model should work for all sorts of complex systems, provided you can figure out a way to measure the cost of various size collapses. Consider the case of a toddler. Allowing every tantrum to run its course would likely drive a mortal parent to despair. But throwing a kid into timeout for every gripe will lead to a child full to bursting with of pent up frustration. I wouldn't want to be the parent that has to deal with it when that dam breaks.

While it's clear that there's an optimal amount of control a parent should exert, it's hard to know just what amount that may be, at least until you can find a way to quantify the suffering involved in each traumatic event. Perhaps continuously monitoring stress hormones in kids, and trying to find a way to minimize them in the long run, would lead to ideal parenting and optimal kids.

For things like electrical grids, on the other hand, the model should be a lot easier to apply. A grid that's 100% reliable most of the time will be entirely out some of the time, but a grid that's mostly reliable most of the time, may be designed to never fail entirely.  

Stock market collapses as well could be the result of regulations and policies that strive too zealously to staunch minor busts. No doubt some economist will find the right measure to plug into the new model to optimize financial regulation.

The UC Davis researchers focus on the physics of sand piles, and don't speculate much beyond them except to note that the work could apply to systems ranging from "finance and brains (frequent cascades) to power grids and forest fires (infrequent cascades)." I tend to think it might even apply to tyrants like Bashar al-Assad. If he could have managed to loosen his grip a bit (see number one in this CNN story Syria's rebels: 20 things you need to know) perhaps he wouldn't be facing the serious possibility of being toppled in a bloody revolution. Then again, it's the need for total control that defines a tyrant, an overbearing parent, tinderbox forests, and at least some failed financial systems. And it's precisely the attempt at total control that makes inevitable the ultimate revolution, dysfunctional child, wildfire and market crash, according to the physics of self-organizing systems anyway.

No comments:

Post a Comment