Readers of this column may be all too familiar with one particular aspect of the scientific research process: planning. From student research projects all the way to grant applications it seems almost that we are expected to know what our results will be before we even do the experiment. This is, of course, an absurd idea, yet it pervades research programs everywhere. This ‘outcome bias’ probably has its origins in today’s obsession with a business model approach to nearly everything in life.
The problem this ‘research program’ approach has is that it makes stationary aspects of scientific thinking that need to flow. Not only are we asked to investigate something interesting (and thereby reveal more about the world – the central ambit of science), we are also asked to explain, sometimes quite specifically, why and how the research might be significant. This kind of reasoning makes us vulnerable to ‘Black Swan Events’ because it limits the expected results and predetermines our response to them. We are locked in to the program and extreme results, outliers, can result in complete failure.
As Nassim Nicholas Taleb has explained in his essays and books, research needs to have options. It may sound heretical to traditional science, but to really advance science, the researcher needs to conduct multiple strands of inquiry that allow choice between outcomes. That is, if one experiment fails, go with the one that worked! This is akin to financial options in the sense that the researcher has the option, but not the obligation, to keep the result.
In actual fact, this approach is already built in to scientific work. It is, essentially, trial and error. The key to success though is that the losses from experimental failure are smaller than the gains from success. This asymmetry is known as ‘convexity’. It can explain why we advance our knowledge and technology. Obviously progress could not come from random chance (by definition – it would no longer be chance, and certainly not random). Convexity also requires a willingness to follow results where they lead, even if this is off-track from the original program, something that pits this approach at odds somewhat with the business model approach.
The driving force behind this convexity in science is that much science stems from technology, not the other way around as is commonly held. We start by asking what we can do with what we have, and realise that the benefits could be enormous (whereas failure simply leaves you where you started, minus a bit of money and time). Perhaps the best example of technology driven discovery is the discovery of the Cosmic Background Radiation (evidence for the Big Bang) by Bell Labs scientists Arno Penzias and Robert Wilson whilst testing some radio equipment for satellite communications. Serendipity? Yes. Predictable? No. Can you prepare yourself for such discoveries? Yes, you need openness to new threads and connections (much like the connection between artistic and scientific creativity.)
We need to retain a certain sense of playfulness in science. Results can be unexpected and wildly different to expectations. To make the big leaps we need to be ready to change gears, switch direction, dump the bad, embrace the good, and follow the road less travelled.