When discussions about interviewing and hiring come up, I always like to remember that these things are strategies, and false negatives are simply a cost of doing business. While I'm sure we can find of some great businesspeople with good ideas who didn't teach themselves some simple programming, a strategy of only working with those who did is probably still an excellent filter that will get rid of far more false positives than trip over false engatives.
When I was in university, I was a TA/Tutor and found that even upper level CS students have trouble coding. There are many cases where the students understood the upper-level concepts behind say networking, or DFA minimization but couldn't produce simple Fizzbizz level code.
The problem wasn't apparantly with their intelligence (as they could annunciate and expand on the concepts they were taught in class), or their rationality (they made it through the Discrete Math weedout), or even their determination (they worked hours upon hours outside of class)....
So now I'm one of those who believes that sometimes no matter how hard you try, there's a switch that won't go off to make one a 'coder'.
That said, I think a business guy should at least try to learn coding if his idea demands doing a lot of it to start with.
In fact, if it turned out that someone wasn't able to code, I probably don't want them as a co-founder because I don't think they are smart/determined/focused enough to run a business.
I'll back this up with an anecdote. At a relatively small company I used to work for, the CEO was someone who used to code, some 17 years ago. However, he recently decided that he knows enough about coding to act as the lead developer for that company's flagship product. He then proceeded to drive development into the ground, arguing over the merits of using malloc instead of calloc, globals over locals, and the beauty of goto's.
He successfully runs several multi-million dollar companies, yet his smartest move was stepping back out of the development lead position, and back to that of managing the company.
Did knowing how to code get him where he is today? Ultimately, no. His business savy and contacts did that.
In an office environment, different roles are much more important.
It also means that he was able to learn to code, but not to the standard that your product required.
However, out of the spectrum of idea guys, one signal I look for is their desire to pursue their idea. One great way to do that -- teach themselves to code.
It doesn't mean that their ideas aren't any good. But at a glance, if someone has an idea that they believe in enough to teach themselves basic skils... well, then they become more than just an idea guy.
Someone who is unwilling to teach themselves basic programming has no business in a web startup.