However, keep in mind that someone on these forums actually tested for very weird things regarding CPU temps, and mostly debunked this line of reasoning.
For example, he put fingerprints on the CPU, and installed the thermal paste/heatsink and there was no change in temperature compared to having no fingerprints.
I think he also tested residue, without perfectly cleaning the CPU.
Pretty much, just make sure you get a good fit, and things like residue and fingerprints won't affect the temperature in any noticeable way. If you want to pretend you are a CSI agent and doing forensic prep on your CPU, that can be a fun hobby aspect to computing, but it won't help lower temps than if you just wipe it off.
I'm sure I wasn't the only one to do those kinds of tests, and you may well be thinking of another forum member, but I did happen to run some of those tests too. I posted them up in the Case and Cooling subforum in my
H100 vs. NH-D14 thread.
There was a bunch of "myth busting" stuff we did towards the end of that thread, lots of good test suggestions from the community and so on.
The tests I think you are thinking of were the ones that involved this:
I unmounted a previously mounted HSF, did absolutely nothing to the existing TIM (left it where it was, all messy on both the IHS and the HSF), I then intentionally stuck my thumb into the TIM that was on the IHS leaving nice oils and skin cells, and to put a finishing test on the affair I pulled a hair out of my head and placed it on top of the IHS TIM as well D:
Then I remounted the H100, didn't add any fresh or new TIM, and fired up LinX to see just how bad the temperature degradation would be. (internet lore at the time would have had us believe this setup would catch fire, spawn a tsunami, and simultaneously give me and my unborn grandchildren liver cancer)
But...alas, surprisingly the results were hardly as catastrophically bad as expected. (see original posts
here and
here for the full story)
^ What we observed was that after having mounted/unmounted the same TIM no less than three times (without ever replacing the existing TIM), adding skin oils and thumbprints, and adding hair, the impact to the peak operating temperature for a 2600k at 5GHz (the biggest delta we could ever hope to observe would rear its ugly head here vs. at a lower clockspeed) was a paltry 5.5°C.
(that final temperature looks suspiciously close to the thermal throttling temp for the 2600k but I verified no throttling occurred during the entire test.)
So, it still goes without saying, keep your bodily excretions (oil, skin flakes, hair, etc) away from the TIM and the surface of the IHS...but at the same time it is not going to be the end of the world if you get a little sloppy and end up with some in there. In fact you probably won't even notice if you did.