Noise and Bias

We all expect consistency from our friends, relatives, organisations, institutions etc. But does it happen? Are we ourselves consistent in our behaviour or decision making? The answer is in negative.

Organisations expect the same consistency from their employees that they should treat identical cases similarly, if not identically. But humans are unreliable decision makers, their judgements may be influenced even by irrelevant factors like weather or mood swings.

Decisions may vary from person to person and from time to time for the same person. That’s why contradictory views emerge. We come across even court decisions contradicting each other. One judgement is superseded by the higher court, and so on.

This chance variability of decisions is called Noise. It’s undesirable, and sometimes disastrous. It’s however different from Bias, which arises out of errors in judgement and decision making due to different perceptions or mindsets on the basis of gender, ethnicity or even certain irrelevant factors. However, many errors arise from a combination of bias and noise.

Looking to the importance involved for an organisation, some psychologists have suggested for noise audit. The degree to which the decisions of the employees of a unit vary is the measure of noise. But it’s difficult to measure bias.

The most acceptable solution to a noise and bias problem is to have unambiguous rules and procedures (SOPs) and in certain cases, human judgements should be substituted with a set of algorithms that are noise-free.

Unlike humans, a formula will always return the same output for a given input. These are the reasons why important iterative processes like Income Tax assessment, are being mechanised to make them faceless.

Can we then substitute men with machines and algorithms? The answer is no. While humans can provide useful input, algorithms do better in the role of decision makers. After all, humans create and run these algorithms.

The importance of human factors and feelings can’t be decimated, but at times, the algorithms and machines assume significance to bring in objectivity and impartiality.

–Kaushal Kishore