Today I stumbled upon these pearls of wisdom by Paydarfar & Schwartz, thanks to Thierry Huck. Thought I’d share. It is a five-step process, one might say algorithm, to scientific discovery.
1. Slow down to explore. Discovery is facilitated by an unhurried attitude. We favor a relaxed yet attentive and prepared state of mind that is free of the checklists, deadlines, and other exigencies of the workday schedule. Resist the temptation to settle for quick closure and instead actively search for deviations, inconsistencies, and peculiarities that don’t quite fit. Often hidden among these anomalies are the clues that might challenge prevailing thinking and conventional explanations.
2. Read, but not too much. It is important to master what others have already written. Published works are the forum for scientific discourse and embody the accumulated experience of the research community. But the influence of experts can be powerful and might quash a nascent idea before it can take root. Fledgling ideas need nurturing until their viability can be tested without bias. So think again before abandoning an investigation merely because someone else says it can’t be done or is unimportant.
3. Pursue quality for its own sake. Time spent refining methods and design is almost always rewarded. Rigorous attention to such details helps to avert the premature rejection or acceptance of hypotheses. Sometimes, in the process of perfecting one’s approach, unexpected discoveries can be made. An example of this is the background radiation attributed to the Big Bang, which was identified by Penzias and Wilson while they were pursuing the source of a noisy signal from a radio telescope. Meticulous testing is a key to generating the kind of reliable information that can lead to new breakthroughs.
4. Look at the raw data. There is no substitute for viewing the data at first hand. Take a seat at the bedside and interview the patient yourself; watch the oscilloscope trace; inspect the gel while still wet. Of course, there is no question that further processing of data is essential for their management, analysis, and presentation. The problem is that most of us don’t really understand how automated packaging tools work. Looking at the raw data provides a check against the automated averaging of unusual, subtle, or contradictory phenomena.
5. Cultivate smart friends. Sharing with a buddy can sharpen critical thinking and spark new insights. Finding the right colleague is in itself a process of discovery and requires some luck. Sheer intelligence is not enough; seek a pal whose attributes are also complementary to your own, and you may be rewarded with a new perspective on your work. Being this kind of friend to another is the secret to winning this kind of friendship in return.
All in all my own process does follow those rules, although I’m still waiting on the major discovery. Sometimes I commiserate about spending too much time in step 1, only to be reminded that all I was doing was step 3. I have been working on some methodological improvements to climate reconstruction techniques for a while, and they are finally panning out… Expect some fireworks soon. Step 2 Idefinitely follow in bursts – feast or famine. Step 4 I learned doing my postdoc, which is when I actually starting looking at data and not just model output. Step 5 is an ongoing process. Going for a drink tonight …