But if there's no standard, then how do you know which behavior to expect in the "global" case and which ones should be vendor-specific? And if only one browser experimentally implements a new feature, should you use the vendor-specific namespace for it, or hope that all future implementations in other browsers are the same?
Just to clarify I'm not suggesting there should be no standard, there definitely should be.
However for the initial implementation what's wrong with survival of the fittest? The most popular way will win through, which would most likely become standard. For those browsers that don't implement it in the same way you use their vendor specific namespace instead.
Yes it's likely to be hairy when it's first being put to use if browsers have slightly different implementations but no more so than using an experimental tag in the first place.
The difference appears to me to be that for the next five+ years I could be writing 8 lines of basically the same css for every instance of linear-gradient (and that's assuming we don't get any new browser vendors in the coming years) vs writing 2-4 lines with backwards compatibility or shock maybe even 1 line if I didn't care about older browsers (e.g. http://www.colorzilla.com/gradient-editor/).
