Friday, November 02, 2012

Publishing Industrial Research

The topic came up in another forum yesterday about how research articles published by industrial people either "reflect[s] work that is no longer mission critical to the company" or is"well timed [and] heavily scrutinized for IP purposes".

While that is certainly true in some cases (very few I suspect), there are plenty of exceptions. Before I address those criticisms, let me summarize what I have observed about industrial research and its suitability for publication.

First off, even if IP protection and trade secrets and keeping a competitive advantage were overlooked, most of industrial research is completely unsuitable for publication. Some of it is only relevant to the company doing the research, such as "What are the best conditions to run that modified reflux unit off the SW side of the plant when we are making product X-2134?" Does anybody in the world care at all? A competitor would hardly even care since they have an entirely different process or at least a different refluxer for making their version of X-2134.

In other cases, the research is very much oriented towards an end use and not towards advancing science. "What combination of comonomers will give our tape the best adhesion to 316 SSL?" A competitor would certainly want to know as they could then quickly duplicate it, but would publishing this suddenly make Dr. Rhe Surch at Stanvard University change direction in his research? And could you even get this published in a well-respected journal?

A tremendous amount of industrial research is done using designed experiments, a statistical procedure that is great when you have a large number of variables to explore. You are able to change multiple variables at once and still "understand" the results. I say "understand" as the output is a polynomial equation, the form of which is chosen a priori instead of being theoretically derived. As a result, the output does little to advance science since it is only useful in the operating window that was explored. I don't recall ever seeing a DOE published in JACS and doubt a reviewer would ever let it pass. (Well, then again...)

So given this, is it worth the effort to publish? You would either have to add additional experiments to attempt to make the research "science-worthy" or you could publish in a lesser journal. And what does it gain you? Companies reward their workers for developing new products that save money, generate sales,...Publications do none of this, so those industrial researchers who do publish usually do it because it is of personal interest, not of corporate interests.

There are exceptions, and my employer is one of them. While we do very little in actual publications, we do give a number of talks at technical conferences about what research we have done for clients, and we use that as a means to promote sales. Of course, the talks are heavily whitewashed so that we don't reveal any confidential information. We can get away with the lack of details in a talk, but I can't imagine any journals that would accept the equivalent lack of details in a publication.

So now to the criticisms from yesterday. "No longer mission critical"? It depends on how you define mission critical. If you are on mission critical research, you are far more concerned about getting it done than taking the time to write an article. Once the research is done, then there might be time to sit around, reflect and write something up, but in the heat of the battle when big bucks are at stake (and yes, mission critical means mission critical), there is no time for publications.

"Well-timed"? I'm not even sure what that means in this case.

"Heavily scrutinized for IP purposes"? Of course, but even academic research has these concerns given that the Bayh-Dole Act allows Universities to patent the results of their research.

I agree with what I think is the general feeling about published industrial research - it just isn't as exciting as academic research. But this cuts both ways, and the best example is to read the patent literature. Patents by industrial firms can be extremely exciting and revealing, but patents written by academics are some of the worst ones ever published. They are heavily oriented along the lines of a research article (I've read some that were almost verbatim their published research) and they show very little effort or creativity in broadening the claims.

Scientific publications are still devoted to their original mission - to publish scientific results so that other researchers can use them to advance their research and then publish their results so that other researchers...Industrial research is far more applied and so even if it wanted to be part of this cycle, it really wouldn't be a great fit. To complain about whatever research is made available is really missing the point and showing a lack of understanding about industrial research.

No comments: