Software and Computer Engineering

Opinion is Good, Data is Much Better

We are inundated by opinions -- often stated as facts -- on a daily basis through the Internet and social media.

Public opinion is often shaped by what feels or sounds right, or what -- openly or not -- reinforces already held beliefs and biases. There is very little usefulness to this kind of data, other than being a representation of “what public opinion is”.

Much more interesting to me is “what does reality say”.

For example, there's a well-known result in economics and psychology that stated preferences often are different than revealed preferences.

Put simply, stated preferences are defined as what people say they'd do in a given situation, and revealed preferences are what people actually do when put in the given situation.

You may have encountered this yourself: If you ask around if people would hypothetically purchase particular small thing, like a chance for some prize, you get some set of answers. If then by the next week you go around offering this same purchase, but now in actuality, you can expected the realized purchases to be much less than the hypothetical purchases.

For a divergence between stated and revealed preference to hold true, we need not ascribe any malice or fault of character to people: what happens is that it's human nature to fool ourselves when we are dealing with hypotheticals versus when we are dealing with the concrete.

(As a side note and example, there's a whole economy of gyms which ride on people paying for membership and then not using the facilities at all. These customers are very unlikely to be trying to fool the gym, they're just fooling themselves by believing they will go. Paying for the membership is the stated preference.)

I found the above very interesting because it very vividly and measurably illustrates how our intuition and planning mind can fall short of reality. As a takeaway, I find that when people -- including myself -- start to get in love with an idea, sometimes they can fall prey to liking the idea so much to the point of ignoring -- or downplaying -- some aspects of reality.

In such cases, I believe that for any intervention idea to have merit, its implementation must also include some measurement or test that can either falsify it or dimension its effectiveness. Unfalsifiable and immeasurable ideas are the domain of charlatans and cults.