Doing The Do-able
I've just been reading a biography of Edgar Schmued, the German-born engineer who came to the US in 1930 and later designed the North American P-51 Mustang. He comments at one point that good designs don't merely serve adequately at the moment of creation, but must include room in which to grow or develop. Althugh he doesn't say so, I would add that there is usually a novel combination of ideas that, in its first expression, fails to reach perfection. Successful later development consists in better exploiting that combination.
Dr. Carcano's 350 Moto Guzzi single is worth examining in this connection. The obvious way to win races was certainly to add power, but in that period, with the tires and suspension available, more power meant handling troubles too great to solve quickly. Carcano went the other way; instead of adding machinery, he took it away. Instead of plunging into a morass of crank torsional problems, bearing failures, overheating, and what-have-you, he stayed with the simplicity of a single cylinder. Instead of focusing on speed down the straights, he concentrated on achieving rapid direction-changing and powerful low-speed acceleration. It all went together neatly into a single package, one whose development was able to outpace that of constructors who had chosen complexity as their weapon.
Complexity was in time mastered -- first by Gilera and MV in the 500 class, later by MV in 350, and ultimately by Honda in that class and smaller ones. In 1953, however, Carcano clearly chose the problems he could expect to solve with the resources he had.
One book I have on auto racing engines rates the BRM H-16 very highly despite the fact that it was a dud, praising its valve area, RPM potential, and so on. However, that organisation picked a set of problems it could not solve -- again and again. In the 1940s they elected to build a V16 with centrifugal supercharging (how could they go wrong? Wasn't Rolls-Royce -- who were assisting with the project -- the world's authority on this technology? Never mind the fact that aircraft engines don't have multi-speed gearboxes, &c) and it was a bomb. After that catasterstroke they chose simplicity, but this time gave their 2.5 liter four such a whacking great bore/stroke ratio that existing valve materials wouldn't keep stem and head together. These people LIKED dealing with problems they couldn't solve.
Or there was the case of the Space Shuttle Main Engines. Let's design this thing in the most fabulous blue-skies manner we can think of. Let's integrate the turbopumps into the engines themselves so any developmental glitches in one single item will hold up the whole works. That way we will at all times have maximal problems of a crisis nature. They made it work, but it wasn't easy.
Edgar Schmued chose novelties to put into his design, too, but they were manageable ones - the laminar-flow airfoil and conic-section fuselage contours. His creation was ready for its engines in 102 days from program start, and its development isn't really over even today - those crazy fellows out at Chino Airport are still getting more from Mustangs every year.