When I started with Motorola in the mid/late '80s, the RISC processor called the 88100 was in the process of being released. It was a physical monstrosity compared to the MC68030 and prior, and consumed an outrageous 25 watts. I made a comment to one of the architects that they should have put it in a circular package instead of square one, and he went off about how incredibly wasteful that would be, how hard it would be to escape the signals, etc... before finally asking me why would I even consider that. I said, at 25 watts, it would make a much better coffee warmer than the 68030 (the 68030 was on the order of a 2.5W CPU, typically passively cooled).
I learned that day that some architects have very little sense of humor.
Later, talking to some Intel reps about the BTX specification that was in the planning stages, I suggested that they put a metal plate standing up from the motherboard in their spec, between the graphics card slot and the CPU. When asked why, I suggested it was so that in a tower case, a person could place their cookies or brownies on the plate to cook, given that their CPU was 25 watts over the 100W bulb used in an Easy Bake Oven (TM somebody, I forget) and with all the fans, the convection cooking capability could be wonderful. This discussion was going on as I was trying to convince them to let us (my employer at the time, Dell) produce a demonstration desktop computer around Yonah (mobile CPU, prior to Intel giving up on the P4 architecture, which was competitive with the P4 on integer benchmarks, but had relatively poor fp performance). Funny thing about that, they seemed more offended about my suggestion that a 'mobile' cpu architecture could be competitive in a desktop environment than their current desktop CPU offering could make every office smell like chocolate chip cookies every day. My, how the worm turned.