Wednesday, November 01, 2006


Software makers began claiming that their products would make you save money, machine time and who knows what else. They didn't that just for the money: even non-commercial software has been "marketed" in this way. Into a world where the computer was a relatively rare thing, the adoption of a software normally went along with the purchase of a new computer, so the total investment needed an adequate justification. People began assimilating the misleading idea, and their expectations began raising. They said: "Using some statistic or DSP trick, a software can improve the resolution, remove the noise, assign my peaks. It doesn't matter if the sample is shimmed properly or not, and I don't need to acquire more than 1 scan, I have seen it published".
Certainly a well-written software can make good use of a computer. For example, with a well-written software you don't need a Gb of RAM or a CPU running at 2 GHz. Has this anything to do with NMR? Statistic and Digital Signal Processing are about the treatment of information, so if the information is missing or poor, what can a software do?
Fortunately, competition produced better spectrometers, sustained by an adequate marketing, whose mission was to stress the opposite message: "invest in higher magnetic fields, advanced amplifiers, etc...".
Eventually everybody had a computer, and everybody began living with a computer. At this point it was possible to change the marketing strategy: "My new software does no miracle, yet it is ten times better than the one you are using and cursing daily". And the chemist began saying: "If I am to spend hours at the computer, I want to spend those hours comfortably". He also felt the pressure of his colleagues who were creating better-looking presentations, pdf documents and so forth. (The good thing of the type-writer age was that the appearance had less relevance, in science at least). The sooth-after miracle was no more the extraction of a readble spectrum from a sea of noise, but the automatic creation of an article from a collection of automatically acquired spectra. Until the scientist discovers that science can go on without human beings and finds himself working at McDonald's.
Instead of looking at the past or at the future, we can look at the present and adopt a minimalistic attitude. What about a bug-free program? Am I asking too much? What about leveraging the power of present-day computers and the estate offered by today's screens? We were forced to buy new computers not just because we needed faster machines, but, sadly, only because Windows XP and Mac OS X came with huge demands of RAM and CPU speed. Enviromentally, it was better to stick to our old 40 MHz machines, but now that the damage has been done, can we exploit the power we have paid for? Or shall we stick to our consolidated work-flows?
Most of all, the main purpose of an NMR software is not to process the FID, that's only a mean to an end. The goal is the discovery, through observation. The two stupid things that I am asking to an NMR program are:

  1. let me see the spectrum
  2. don't distract me


Post a Comment

<< Home