Research Ideas: Think Solutions, Not Problems
Scientists apply a familiar scientific method for posing and addressing research problems. But successful R2O processes not only must support rigorous scientific work. R2O processes must also position the research results in the form of a deliverable research byproduct that practitioners can use to solve an operational challenge. This real-time deliverable built from raw or “first level” data constitutes the complete solution under a R2O framework.
It is possible to distinguish the complete solution from the scientific component. More importantly, the science is not the entire solution. Transitioning research to operations is as much about improving the characteristics of the deliverable as it is the quality of the science. While a solution without solid science is unlikely to yield a meaningful research byproduct, science without a tailored solution is not going to add much, if any, value to practitioners whatsoever.
The R2O solution is the method of communicating and visualizing information from research byproducts that leverage a scientific result. Scientists should aspire to envision the nature of the solution first, if at all possible, even before research begins. Assistance from the operational side is most valuable here because knowledge of information gaps and previous visualization techniques can provide ideas for a solution that fits the practitioners’ mindset. The solution does not necessarily need to be perfect, as the cyclic R2O processes will improve it with time, but it should serve as an empty vessel with which to fill the new information for conveyance.
Consider an example where scientists are proposing a new technique for identifying turbulence from thunderstorms using satellite imagery. For the sake of this example, we will assume that the science behind it has completed peer review. For the practitioners, there is great demand for this kind of information, particularly due to the many transoceanic air routes between continents where turbulence is, at least, an unpleasant interruption for passengers. But the oceans are vast and in-situ observational data is sparse. It may be possible to train practitioners (weather forecasters) to identify the spatial patterns and understand the presentation of turbulence indicators on satellite imagery, but this could require a lot of time to perform operationally due to the frequency of imagery and size of the ocean basins. Ideally, a real-time deliverable would present possible areas of turbulence, with additional information about the nature of the automated detection, such as the strength of the feature and the confidence of the signal, but is not bimodal.
In most cases, quality communication and visualization solutions for applying research results to operational problems in real-time share certain characteristics:
- Information is synthesized to what is necessary, with hints to the practitioner as applicable. For example, conditional (sufficient, necessary), confidence, intensity (low, moderate, high), or strength (weak, strong) adjectives should accompany unfamiliar indices or quantities so that it is easy to ascertain the importance of the values and their quality. Having this information in separate reference materials requires additional time and effort on behalf of the practitioner. Given that, there should not be any extraneous information that detracts from the intended message or is otherwise secondary.
- The visualization approach should decrease the amount of time that the practitioner requires to review the byproduct to gather the necessary information for the operational decision. Colors and shapes should be optimized to draw attention to features of importance.
- New imagery and information should overlay or coexist with familiar imagery and information for the practitioner. Practitioners will need to be able to extend what they know into what they want to know in order to build confidence in the science and solution. New solutions should allow them the opportunity to reconcile the new information into their existing “mental model”, especially during the initial stages of familiarization.
These characteristics can be taken to the extreme, reaching the point at which it is a “black box” for the practitioner. This is not desirable. It is possible to strike a balance between presenting a solution where the practitioner consumes the information and, with it, adds value to their user, compared to one that generally does not require a practitioner to interpret at all.
To that end, if the intent is to eventually transition research results to operations, scientists should initially ask, “What new information might help practitioners in this situation?” and think backwards to arrive at the scientific challenge(s), instead of starting with, “What is the research problem in need of refinement here?” and subsequently trying to fit the result into a deliverable solution.