To ensure that a desired result is achieved, engineers design systems that measure the system output and compare that output to a reference value. If the output does not match the reference — if there is an error — the system adjusts itself to minimize the error. This type of system, one that samples the output and feeds it back to the input for comparison and correction purposes, is called a feedback system.
As an example of a feedback system, consider the system that consists of you and your automobile, zipping along on the interstate with a police cruiser not far behind. To make it in time to a very important interview for a lucrative new job, without getting a ticket, you plan to accurately drive along right at the speed limit. Visual feedback from the speedometer tells you the system output (speed) and whether or not you are traveling above or below the limit. If an error occurs (whoops, going a bit too fast), your brain sends a signal down to your foot to ease off of the accelerator until the speed is reduced to the limit. This feedback loop, from output (as measured by the speedometer) to your brain to your foot to the gas pedal, maintains your car at the desired speed.
Now imagine that the feedback path is broken; e.g. your speedometer suddenly quits working and always reads zero. In this case you rely on a secondary feedback path, the passing landscape, and estimate your speed. Since it’s only a rough approximation, however, you’ll play it safe and drop your speed a few miles per hour to be sure that you are under the limit, although this might make you a bit late.
But what if it’s a dark night with no moon and your headlights fail? You now have no feedback, and are forced to stop, ruining your chance for the new job.
As feedback decreases,
Because of its extreme importance, feedback is used everywhere. In many cases, as in the example above, sensors provide feedback to human operators, who act on that feedback to achieve the desired results. In many other cases feedback is completely automated, without any human intervention at all.
Feedback is applicable to social organizations as well. Many commercial organizations exist for the purpose of providing goods or services to the population. These companies receive feedback from their output (how much they sell) and then adjust prices accordingly. A “price discount” in this case is analogous to the accelerator in the auto example, and sales is analogous to the speed. If sales are down, a company will step on the accelerator by offering higher discounts in order to increase sales.
Likewise, individuals who work for private companies experience feedback in the form of salaries and promotions based on their work efficiency. Slackers tend not to be paid as well, if they remain employed at all.
At the personal level, a spouse who provides quick feedback about their partner’s perceived inappropriate behavior will reap the rewards of a much more efficient and positive relationship, than one who remains mysteriously and sullenly silent.
Large bureaucratic systems, such as the federal government, are the least efficient organizations for providing services, because (like a sullen spouse) they do not employ effective feedback. In fact (unlike the sullen spouse), bureaucratic systems have no choice — their very nature precludes the existence of meaningful feedback.
In other words, there is no output/price feedback. Consumers cannot chose Federal Government A over Federal Government B because A charges less for services than B; we are all forced to “buy” government services from a single monopolistic federal entity, under penalty of fines or imprisonment if we don’t.
Although governments don’t have output/price feedback, one could argue that they do have feedback from elections (and sometimes in the interim from noisy constituents). However, the time lag between governmental actions and subsequent voter response on election day is so large as to render the normal benefits of feedback almost moot. Indeed, as we’ll show in an upcoming post, a time lag can create system instability, or even the opposite of the desired result.
If one accepts the fact that quick feedback is essential for efficient and accurate results, then an important Engineering Thinking conclusion is this:
Regardless of the Intentions or Talent or Compassion
or Political Beliefs of the Individuals Involved,
Excellent Results Are Much More Likely to be Achieved by
Individuals or Organizations That Employ Effective Feedback.
Poor Results Can Be Expected from Those That Don’t.