How can this be? Harley-Davidson at the time was a very prosperous manufacturer, so R&D expense was surely not a problem. So why zero wins?

There is a large difference between racing or military R&D, and commercial R&D practice. That difference is time. In racing, as in wartime, new capability is needed now, not in six months. Making things happen quickly is expensive.
During WW II a number of experienced auto manufacturers developed prototype aircraft engines of high performance, but were unable in the time available to make them combat-ready The plant and machine tools earmarked for such projects were reassigned to other activities.

This was explained to me by a retired P&W VP, who described that company’s method as “trampling problems to death.” It is common in commercial development to deal with problems serially, as that saves money and can be conducted with limited staff. In the P&W method, an initial meeting defines all possible solutions to a given problem, such as the crankcase cracking that the otherwise promising Chrysler IV-2220 inverted V-16 aircraft engine displayed. All solutions were then developed simultaneously. This was successful because it assigned highest value to time rather than to money.
Can a house, divided against itself, win Superbike races? Early development of the VR1000 used the talents of the Roush organization, and eventually a Roush employee, engineer Steve Scheibe, was hired to manage the VR program. Why not give the project to Harley’s engineering department? I don’t know, but Scheibe seemed chronically under pressure to yield control of the VR.

“The Detroit Way” had an influence at H-D. This was to break down a project into separate areas, hand those areas to separate teams, and then to hold meetings in which progress would be reported and forecast. Other meetings would devise the ‘metrics’ by which such progress could be objectively measured. Once the sub-projects had delivered, the process of ‘systems integration’ could begin, eventually resulting in a successful product.

Systems integration is no small problem. Have a look at the little book “Not Much of an Engineer,” the autobiography of Stanley Hooker. Hooker had been hired at Rolls-Royce straight out of university. His speculative math model of how a centrifugal supercharger should work turned into a 30% power increase for the company’s ‘Merlin’ V12 aircraft piston engine. Just in time for the Battle of Britain.

Later, a disagreement with R-R management caused him to leave the company. In the 1960s, R-R’s development of a large fan engine for commercial aircraft – the RB-211 – became stalled, with both thrust and fuel consumption far from their projected values. Confidence was disappearing. A compromise was struck by which Hooker returned to R-R to “have a look at RB-211.”
What he found was excellent in detail – even record-breaking - individual engine elements – fan, LP and HP compressors, &c. What else he saw was that the output of each stage was not that planned for the next stage downstream. When Hooker, looking at the engine as a complete system, adjusted these stage-to-stage incompatibilities, thrust rose nicely and fuel consumption fell. The separate teams had done excellent work, but there had been no one with the experience to make a complete system of their work. How can that person be found? Could this be the reason Honda and Yamaha have hired MotoGP- experienced Italian engineers?

A few years ago I was shown the elements of a MotoGP engine being constructed by a novice organization with some novel ideas. The team appeared to believe once all parts had arrived from suppliers, the next step was to assemble engines and start winning races.

No. A running engine is just the beginning of development, reliability first, performance second. Can the team shape torque delivery to lie within the capabilities of existing tires?

BMW, supremely capable and experienced in high-power F1 engine development, took years to make their S1000RR-derived Superbike compatible with available tires. At the start of a World Superbike race, Troy Corser would take the lead and the team would rejoice. After five or so laps, he would begin a fade to the rear caused by tire drop. This was no small thing – in one case I recall he fell back to 13th. Having a powerful engine and a workable chassis are important, but absolutely essential is matching engine torque delivery to what tires can tolerate.

Remember also that as an engine is developed to higher power, its torque curve becomes less and less usable by a human rider.

BMW achieved this, their developed machine taking World Superbike rider’s championships in 2024 and 2025.
This tells us that to win races, a constructor must develop a whole system that can produce the desired results. A racing motorcycle is much more than an object. It is the product of a complex process.

At the beginning of the new century I spoke with Claudio Domenicali at Ducati. He told me they expected the price of “getting in” to the new MotoGP series would be US$32 million, with operations thereafter costing ten million a year. Yes, there has been some cost inflation in the past 25 years. Much or most of that funding, he said, would have to come from outside sponsors. This, too, must be part of a “racing system” (‘No bucks? No Buck Rogers’ was a NASA saying of the late ‘60s).

The above is not a list of problems that prevented Harley’s VR Superbike from winning. It is rather examples of the kinds of complexity that can make rapid progress difficult or impossible.