But what if, after spending more time working on and thinking about it, someone does find a way to make predictions based on it or on an extension/modification of it?

If that's a possibility (I can't say, I'm not a physicist) then I'd be a mistake to discard it on the basis that it's just not predictive now.