I think a greater parts count would be better, but I don't think it's such a difficult task. Lots of things escape common understanding; a few even escape detailed analysis and simulation.
You can even have many of the same "abuse" categories as the IOCCC (it's three C's, right?). Logic chips selected for marginal thresholds; strange behavior at odd temps (including using the chip itself as an oven to reach that temp, or..). Building everything out of discrete transistors, using multilevel logic thresholds, would be similar to macro expansion abuses.
The real trouble I think is going to be availability: there's so damned many people using C right now that there's bound to be a lot of smart hackers working on its, shall we say, dark side. There just aren't so many hardcore analog engineers out there I feel.
I'd love to be corrected on that though. I'll try my best to come up with something for sure
Edit:
In response to the video, and in similar fashion to the IOCCC, all details required to implement the circuit -- the source code, as it were -- must be provided. The schematic at a minimum, but also layout details if necessary (because let's be honest, a lot of *normal* microwave circuits look pretty damn crazy to begin with), components selected for properties, and so on. It can't be a contest if the judges can't construct the circuit for themselves!
One interesting difference, programs can be self-modifying, or introspective (I recall one IOCCC entry which produced platform-dependent results, detecting the processor it was compiled and ran on). Whereas, the circuit is naturally static in configuration, and you have to go out of your way to change it (analog switches, muxes, etc.), in which case the purpose is apparent. Likewise, whereas debugging a program requires knowledge of the data structures in memory, all the signals are there for the probing -- the voltage signals, anyway. (One could attempt to obfuscate this by using current-mode signals with as little voltage swing as possible, or extremely high impedance signals which defy observation via conventional probes.)
Tim