Delays

# 9. The lengths of delays relative to the rate of system changes Remember that bathtub on the fourth floor I mentioned, with the water heater in the basement? I actually experienced one of those once, in an old hotel in London. It wasn't even a **bathtub**, it was a **shower**--no buffering capacity. The water temperature took at least a minute to respond to my faucet twists. Guess what my shower was like.

Right, oscillations from hot to cold and back to hot, punctuated with expletives.

**Delays in feedback loops are common causes of oscillations.** If you're trying to adjust a system state to your goal, but you only receive **delayed information** about what the system state is, you will overshoot and undershoot. Same if your information is timely, but your response isn't. For example, it takes several years to build an electric power plant, and then that plant lasts, say, thirty years. Those delays make it impossible to build exactly the right number of plants to supply a rapidly changing demand. Even with immense effort at forecasting, **almost every centralized electricity industry in the world experiences long oscillations between overcapacity and under capacity**. A system just can't respond to short-term changes when it has long-term delays. That's why a massive **central-planning** system, such as the Soviet Union or General Motors, **necessarily functions poorly**.

Because we know they are important, we systems folks see delays wherever we look. The delay between the time when a pollutant is dumped on the land and when it trickles down to the groundwater. The delay between the birth of a child and the time when that child is ready to have a child. The delay between the first successful test of a new technology and the time when that technology is installed throughout the economy. The time it takes for a price to adjust to a supply-demand imbalance.

A delay in a feedback process is critical _**relative to rates of change in the system state** that the feedback loop is trying to control._ Delays that are too short cause overreaction, "chasing your tail," oscillations amplified by the jumpiness of the response. Delays that are too long cause damped, sustained, or exploding oscillations, depending on how much too long. At the extreme, they cause chaos. Overlong delays in a system with a threshold, a danger point, a range past which irreversible damage can occur, cause Overshoot and Collapse.

I would list delay length as a high leverage point, except for the fact that **delays are not often easily changeable**. Things take as long as they take. You can't do a lot about the construction time of a major piece of capital, or the maturation time of a child, or the growth rate of a forest. It's usually **easier to slow down the change rate**, so that inevitable feedback delays won't cause so much trouble. That's why growth rates are higher up on the leverage-point list than delay times.

And that's why slowing economic growth is a greater leverage point in Forrester's World Model than faster technological development or freer market prices. Those are attempts to speed up the rate of adjustment. But the world's physical capital plant, its factories and boilers, the concrete manifestations of its working technologies, **can only change so fast**, even in the face of new prices or new ideas--and prices and ideas don't change instantaneously either, not through a whole global culture. **There's more leverage in slowing down the growth of the system so technologies and prices can keep up with it. than there is in wishing the delays away.**

But if there is a delay in your system that can be changed, changing it can have big effects. Watch out! Be sure you **change it in the right direction!** For example, the great push to reduce information and money transfer delays in financial markets is just asking for wild gyrations.)