One of the biggest challenges for design engineers is to know how hard they should strive to make the product “even more better” versus when it's reasonably matched to the application and expectations. These two forces play against each other in the engineering mindset and produce an internal, personal conflict, as well as visible disputes between the engineering team members and with the marketing folks.
But the real challenge is to face the questions of who are we doing this for, and what do they expect of us? If you are making a mass-market product, accuracy to 0.01% may not be appropriate, while a physics experiment probing for neutrinos may need precision and purity that goes beyond even today's smallest-feature IC fab processes.
Of course, different audiences and markets have different perspectives, and that affects your definition of “perfect.” I recently attended an antique auto show (cars from before about 1980, with the bulk of them from the 1965-1975 era). While some of the cars were there just because their owners wanted to show them, many were in a formal judging process.
The cars that were entered were beyond pristine. They were chromed, polished, waxed, and dust- and fingerprint-free. They looked better than brand-new. They also looked, at least to me, somewhat artificial, since they were “perfect.” It's all in the eye of the beholder, I suppose.
In contrast, I attended a model railroad event several months ago, and the award-winning models had the completely opposite tack. There was a heavy emphasis, for both rolling stock and fixed structures, on making them look weathered, beaten, oil-stained, derelict, and as un-pristine as possible; in short: used and realistic, indistinguishable from the real ones except for actual size. (There's a basic example here.) There were also many well-attended classes on the subject–broadly called “weathering”–and how to use an array of powdered chalks, pastels, India-ink washes, the X-acto knife, sandpaper, gouges, and other techniques to achieve it. In short, the perfect model was one which is very imperfect.
Understanding who you are serving, and your personal motives (i.e. advancing the research envelope versus developing a lower-cost, mass-market product) are critical, and you can't let your personal ego get too much in the way of what your target audience wants. If it does, you may have to find another audience!
-x x x-