Epistemic status: Heard it at my local pub, but I'm not sure if I believe it.

A story is told of a shopping mall investor, who estimated the size of the middle class in Nairobi by hiring a plane to count the number of satellite dishes on rooftops. Putting aside the merits of using such data, to make billion shillings investment decisions, people have used unlikely ways in the past, to gain valuable information about many things in life. The number of pizzas ordered has been used as a sign of a crisis in the US government. Statistical formulas have been used to estimate the production capacity of an enemy nation's tank making factory. We have developed many ingenious ways of getting valuable data. For example, while I might be interested in finding out how many people enjoy salt in their watermelon or how many like salt in their avocado, it would take a lot of time and money to ask every Kenyan this question. Consequently, we end up doing surveys where we only interview a few Nairobians and ask them these important questions. The small number we talk to, and who are willing to share their opinion on this matter, represent the whole country; they are a sample of the group. In this way we get to know more about the whole population without doing a costly census. The statisticians can then interpret the results and we get a good estimate of the percentage of people in Kenya who prefer salt in their avocado. This works well for questions where we can easily quantify responses. There are other topics where we should not and are not allowed to do polls of this kind. This can be either due to ethical concerns, privacy considerations or if it is prohibited by the nation's laws or taboos or when it is just impractical. In such cases we rely on proxy data, anecdotes and heuristics.

The investor’s story is an example of using proxies to acquire insight, in an area of interest, where it might be difficult to do so through a questionnaire. We are often warned about inferring causation from correlation, however, it seems conclusions derived through induction are preferred over making choices blindly, no matter the dubiousness of how it is done. One might counter and ask, are we simply being lazy and making decisions based on incomplete pictures of the world? At what point does the price of conducting proper research outweigh the benefits? There are also times where the potential reward is extremely high compared to the cost; there is a reason insider trading is illegal. 

The source and quality should still not be the only things we consider. The main problem is that census and opinion polls are unclear reflections of the real world, it would be like someone who saw a screenshot of a two hour movie and used that to draw conclusions on the plot. We need multiple sources. Secondly, of importance here, is not in how we find out but in how we know something is true. Having various ways of gaining knowledge gives us some sort of checks and balances on the accuracy of the facts presented. Simply put, when you have more than one source, with one you will make a claim and with another you will prove its veracity.

“In your own Law it is written that the testimony of two witnesses is true.” John 8:17