It's a big part of it, but careful about what you mean by "domain knowledge." Look at Richard Feynman's stint as an analyst for Kurzwiel as parallel-programming

It's a big part of it, but careful about what you mean by "domain knowledge." Look at Richard Feynman's stint as an analyst for Kurzwiel as parallel-programming got underway; in theory Feynman had no significant domain knowledge but in fact the immense range of mathematical knowledge he brought with him turned out to be extremely valuable, perhaps irreplaceable for that particular project (given how creative he was.) He really was a (more than) ten times guy, there.