That being said, the fact that the forms are now HTML-based, it would be fairly simple to write a custom validation script to check if any superfluous button—I'm calling these the "not selected" items—within a group that has been marked as being "required" is checked. In this case, the assumption would be something like, "What the...? This group is marked as required, but the 'not selected' item is the only one that is checked—this means that none of the remaining items has been checked. TILT!"
Being a rather linear retired civil engineer, I, for one am really glad that w3c coordinates standards. We'd ultimately be sunk without them. As Scott says, the standard—in its current form—dictates that the browser agent must assign a "checked" state to the first item of any group that has no selected items:
I'll definitely admit that it would be nice to NOT have to write the custom script. Thus, it would seem that the w3c standard needs to be changed.
I would not like for CC to have to only "kind of" adhere to the standards. What if some browsers observe the w3c radio button standard, but others didn't (Internet Explorer, anyone?). Then, we're still in the business of browser detection and applying hacks. Isn't the goal to be able to avoid that?
Guess I'll bow out, now—my head is beginning to hurt, and it's time for an adult beverage!