1) Consistent Exploit Code Likely 2) Inconsistent Exploit Code Likely, and 3) Functioning Exploit Code Unlikely
The first one means a software flaw could be attacked with highly predictable results, and would probably be very easy to exploit. This would be very bad, as exploits would surface, and would be turned into weapons for mass use. This would be a critical vulnerability, and would need to be patched. Designation two could be bad, or it could be not-so-bad. Maybe an attacker could create an exploit, maybe not. And how the at-risk system reacts to the attack may not be very predictable. The third designation, Functioning Exploit Code Unlikely, is obvious: Microsoft has determined that developing a useful, functional attack tool would not be likely.
My opinion from back then hasn't changed from my original post, which is that this won't be of much value to operations teams trying to assess their risk. Tomorrow, and in the months ahead, we'll see if it works as intended.