In “Fooled By Randomness”, author Nassim Taleb makes a very incisive point: what we percieve as survivorship bias is oftentimes just a randomized effect. One can illustrate this by the following process that Taleb also makes in the book. Suppose you have 100,000 people engaged in some activity with a large element of luck or randomness involved (i.e. stock trading/forecasting, or scientific research*). If at every stage you eliminate 90% of the participants, after 5 stages, you will be left with 10 people. Are these people necessarily “better” than the others who were not eliminated? At the outset, before the process began, we might not have said so. However, once everything is over, the mental process of “survivorship bias” kicks in to make an ex post facto rationalization as to why these remaining 10 people are superior.
*One might find the use of “scientific research” in this context odd; after all, it is supposed to be a methodical process, right? Well, as someone who completed a PhD in chemistry, as a practitioner of scientific research, and as a scientist, I can attest that the process of research is all luck. Think about it: research operates on the frontier of human knowledge, whereby the outcomes of what you plan to do are basically unknown. One can make an educated guess, and these educated guesses (or “hypotheses”) form the direction of research inquiries. However, at the moment of doing the experiments, you do not know which one will yield the results you want. This is how science (especially chemistry) operates – by running a lot of experiments, and seeing that one of them worked (due to chance), and then retroactively fitting a model or theory around that result.
This makes me uneasy, in much the same way that finance makes Taleb uneasy, because the conclusion is that scientists are no better than financiers at dealing with randomness! And these are the people we look to for improving our quality of life, solving some of the most pressing problems we as a species collectively face, and furthering the boundaries of human intelligence!
It is disturbing to me that so many processes are now due to luck, or can be described as a “numbers game”. Activities falling in this category include job applications and dating, among others. The very fact that dating has become a “numbers game” is exploited by the numerous dating apps that have proliferated recently, including Tinder, Hinge, and Coffee Meets Bagel. These apps attempt to make the process of meeting people easier, by removing the necessary barriers (i.e. proximity) necessary for communication; however these same barriers are also lowered for rejection! By dramatically lowering the barrier for rejection, people are more prone to wait for someone perfect to come by (even if that is an ideal that may never be materialized), rather than giving existing people of the opposite/same gender who may have expressed interest in them a chance. A similar situation exists in today’s job market; with the current saturation of job seekers in the market, employers can afford to wait for the perfect “purple squirrel” candidate, and if he/she doesn’t turn up, complain that there is a “shortage of STEM workers” and bring in more H1-B workers**.
In short, processes that used to be systematic have now become stochastic. I’m not entirely sure why that is, but my observations about the world around me seem to be consistent with this sentiment.
**On this note, I would like to point out Norman Matloff’s excellent blog. Matloff is a realtalker and makes a key distinction that people fail to observe with H1-B workers: while the program is supposed to bring in the “best and the brightest”, what it is actually accomplishing is bringing in mediocre immigrants en masse and displacing otherwise qualified Americans from their jobs. It’s ironic that I would be bashing the H1 program, as my family immigrated to the US as part of that; however, my father definitely belonged to the “best and brightest” category, having been given a job offer as CTO of a tech startup during the 90’s-00’s dot-com boom.