My point is clear, simple, rock solid, quite explicit, and very well supported: Again, yet again, this time just for you, my point from DoD research is that DoD research shows that research can find powerful solutions to important practical problems. Examples include a long list of astounding military technology from the atomic bomb, the hydrogen bomb, sonar in all its forms, radar, now a deep and astounding field, laser guided bombs, GPS, stealth, high bypass turbo-fan engines, e.g., first for the C5-A, carbon fiber materials, CAD, originally heavily for aerospace, etc.

What is "practical" is the unique power of research to find powerful solutions to important real problems.

I omitted all my peer-reviewed, published research results. But if you want some examples of research, try the shelves of any research library.

My post, if actually read, directly addressed the main problem in PG's essay, how to evaluate projects. My solution was my three steps. That is, start with an important problem and do some research to get a powerful solution. That solution addressed PG's issue.

For how to evaluate some routine application of software for a simple case of some social, sharing, mobile app, my solution is don't try and, instead, go with projects that use original research to get powerful solutions for important problems.

If you want to know what research is, get a Ph.D. in a technical field from a good research university. If you want to know how DoD does research and applies it, then get a job in a DoD laboratory that does such work -- the DC area is surrounded by such labs from NRL, NSRDC, JHU/APL, and many more, and there are many more such labs all across the US.