Really? Would you say this is true for Newton's work, or Einstein's?
I find that new ideas in the sense of new insights about how the world works are extremely rare, and that having them is much harder than testing them, because having them requires you to change the way you think.
Perhaps you are influenced here by the situation with startups, where ideas are cheap and execution tends to count for more. But this is a different kind of idea. The reason startup ideas are cheap is that they tend not to be discoveries, but merely hunches about where to look for discoveries. E.g. the startup idea equivalent of Newton's work would be "Explain mathematically why planetary orbits have the shape they do."
Testing ideas, to me, is the best way to change how we think, especially when we're proven wrong. If we know how the experiment will turn out, there's little point in actually running it. And there, I think the process you describe for startups isn't so different than the process for science. In each, the hunch points the direction but the apparatus must be built to verify or reject it. But it's hard for me to see one hunch, rather than a tested series, leading to a true discovery.
To put it another way, in merely writing an idea down it may exist all by itself in some Platonic ideal. Not surprisingly, the Greek philosophers often left it to others to validate (or invalidate) what they produced. Running tests wasn't truly valued. I'd argue those ideas that have stuck with us the longest are those that engendered the greatest utility.
Then there's Heraclitus.
Almost all of the most important intellectuals in the history of western civilization have had one thing in common: their ideas were wrong. Plato, Piaget, Copernicus, Einstein, Lamarck, Chomsky, Adam Smith, etc.
If testing ideas for truth value were really the hard part then scientific experiments wouldn't be replicable and science would have no more weight than astrology. Furthermore, scientific beliefs wouldn't run in S-Curves and progress would be linear and continuous.
Coming up with an entirely new paradigm is, if not exactly a "wicked" problem, some other type of complex problem so difficult and uncommon that we don't even have a vocabulary to describe it.
So when I say that coming up with ideas is more difficult than testing them, what I mean is that coming up with an idea belongs to a more difficult class of problem, even if it is occasionally easier to create a specific idea than to test that specific idea.
But, sure, working the lab equipment and running the tests is a waste if you're not testing good ideas. Problem is, I'm the worst judge of what's a good idea especially if it seems good to me.
I'll venture so far to say that good ideas are the ones that work. The difficulty is predicting ahead of time whether they will.