So, Whats the conclusion here.
Firstly, its not possible that the safe Tjunct temperature is the same as the Max TCase temp. They are two very different things. So if a guy from Intel claimed that Tcase and "core temp" is the same thing.. well...........
Secondly, its very true that the internal probes that software use to estimate the core temperatures are only really calibrated for the throttle point (max TJunct). Stuck sensors were a common issue on the Q6xxx range, but virtually every processor would come "unstuck" as you approached (or reached) the throttle point.
Thirdly wasnt the entire point of the thermal throttle to prevent damage to the cpu caused by heat. Unless the CPU is clocked so heavily that the CPU is already throttled as low as it can go (can intel CPU's throttle lower than the speedstep 6x multi for example?) then the thermal management should be fine.
Totally agree with Justintime here. Keep temps 15-20 degrees from thermal throttle on a normal day. That gives you an extra margin if there is a really hot day and your computer runs hotter as a consequence. If its within 10 degrees but your home/room is hotter than normal, it wont really do any harm.
However, Rather than the focus on speed, the real underlying issue is far more likely to be people exceeding a safe voltage for the processor. How many motherboard do you know that include a thermal probe which can touch the top of the heatspreader (which is buried under the heatsink!!!) Its not reasonable to expect anyone to actually know their computers TCase. Thermal probes under the socket are notoriously unreliable, and the Tjunct sensors as already mentined are allowed to run hotter, and on the whole their calibrated values are only officially known by intel, except for a few leaked documents.