Articles

"Pushbutton physics" is a term for the use of prepackaged experiments using "canned" control programs, automated calibrations, and integrated data analysis software. Such artifices, which are prevalent on top-of-the-line equipment today, have the effect of decoupling the researcher from his/her data acquisition activities, making it harder and harder to understand exactly what happens when data is collected. Take, for instance, the modern versions of AFMs. They typically offer, CFM, MFM, "tapping mode", "non-contact mode", line by line lift, point by point lift, etc. The trouble is when these instruments are used in unexpected or unanticipated ways, it can be nearly impossible to understand exactly how the software is responding to those unanticipated conditions. Anecdotal evidence is that this is becoming an increasingly serious issue for laboratories that go beyond simple materials characterization work. But there has been very little work done to study the impacts of this automation in the science outcomes of American physics labs.

​

Moreover, we can already see a corollary to this trend in the pages of our most top flight journals. For example in the pages of journals like Nature and Science we note the increasing use of figures that are composites of data sets - including Raman, XPS, TEM, EELS, etc., and usually so small that one can not even read the axes or scale bars. But in many of the papers that do this, there is relatively less detailed analysis of those data sets with only a summary statement and reference being offered. Again, this is anecdotal, there doesn't seem to be any serious study into the relationship between presentation styles, rigorous analysis of each curve presented, and data collection methods.

​

So, why is this commentary on a "rankings" page?  The fact is the way we use publications and their presentation of data and data analysis, is evolving. However, when we seek to understand a program's impact we typically (as we do here) include citation factor. It is in fact very difficult to go any deeper than citations, reads, downloads, searches for impact. But consider the scenario of a graduate student that goes to a national lab to carry out an experiment. The student likely doesn't even touch the instrument, and uses programs that have already been tested and proofed for a given application. They will also be able to have that sample looked at by the many numerous labs at the facility with little extra work. This all leads to a much higher probability that the paper generated by this exercise will show up in one of the top flight publications. The program has high impact. And there is no doubt that there is significant impact through the work that has been done, but one wonders if the student produced using such methods is truly skilled in the work at a level superior to students that worked on campus? That is, as a student are you really better trained like this? Certainly there can be arguments either way, but how does one rank this? In such cases a deeper dive into the nature of the publications and citations must be used. But it is never clear as to how to measure such impacts.