How much time has humanity, collectively, spent researching physics?
What about chemistry? Neuroscience? Algorithms? (etc…)
How has this rate of study changed over time? How much of it is duplicate work (e.g. two nations or organizations independently working toward the same milestones before sharing their findings)?
What fraction of this is done by professors, post docs, PhD students, etc?
Particularly in the last 100 years, what have these numbers been?
I’m interested in visualizing some of these metrics, because I think they would be useful for putting the work that I see happening around me in machine learning and in neuroscience into perspective. I also think this data would tell a fascinating story.
For example, one plot that I think would be compelling would be person-hours-per-year spent thinking about physics plotted as a function of time over the last 100 years, with important physics milestones shown on the same plot (e.g. as a time series plot with events.)
One worry with such a visualization is that the population is growing rapidly, and so the interesting story of how humanity’s attention has shifted between fields over time may be lost to a visualization that simply shows population growth. It may be necessary to normalize by population to tell such a story.
It may seem infeasible to precisely collect this data, as there is no public registry of who is thinking about what. However, I think the data is perhaps more readily available than appears at first blush.
Where would the data come from?
- Search trends
- University enrollments
- Publication counts
- Surveys
Search trends are of course a more recent source of data, only going back one or two decades. The pre-1990s version of search trends might be library records, or book sales.
University enrollments and degrees granted tells a significant story about how many people are studying different areas. You can roughly estimate how much time a student majoring in physics puts into their studies, and if course enrollments are available that estimate can be even more precise.
Scientific fields make their mark on the world by publishing their research. Counting publications can provide an estimate of one piece of humanity’s effort studying science.
Finally, for contemporary data, and maybe even a little historical data, one could uses surveys / polling techniques to estimate where scientific mindshare is going today.
Among this data I’d be curious to see when historically there have been large swings in mindshare toward a particular topic. Right now, anecdotally, machine learning research is engaging a quickly increasing number of people, including drawing people away from other fields. I’d love to see this anecdotal evidence backed up (or refuted) by data. And I expect there are plenty of similar historical precidents that I would be quite interested in learning more about as well.
I don’t foresee collecting this data or performing these visualizations myself in the near future. This post is just an opportunity to think aloud, not a plan for my own action. But if you have data like this, or know of someone who has done a related analysis, I’d be quite interested to hear about it. Don’t hesitate to get in touch.