I just don't understand the, Why? Did I miss it in there?

Does HFCC have enough randomness in it that redundancy checks are not as valuable as in HCC?

Ideally all projects should be run without redundancy because it is obviously a waste of computing power to compute the same job twice (and in the past it was three times!).
However WCG and the scientists need to be sure that the results are reliable, i.e. that they have been computed correctly, and for some projects like HCC and TCEP there is currently no other way to be sure than to use redundancy. The reason may be the nature of the work done (HCC) or the software used (TCEP).

DDDT, FAAH and HFCC are all using the same software, Autodock, for which a way has been found to check the reliability of a result without having to duplicate the whole computation. To sketch it simply each WU contains a mini-WU whose result is known, and if this result is OK and the device is known as reliable it is assumed that the rest of the work has been done seriously (see paragraph 4 of the post I have referenced in my previous post). This way only the computation of the mini WU is "wasted", not the whole runtime.
In addition devices are checked randomly or when a result seems unusual (e.g. much shorter or longer than the average for thiis project and this device) to make sure that they are still reliable. This is done by using redundancy for a number of WUs sent to this device.

NRW and HPF2 use another method for reducing duplicate work. Have a look at their respective posts in the thread I have indicated earlier to see how the techs have implemented the distribution for these projects.

I hope I have not finished to confuse you. Jean.

That seems reasonable.

I would expect even the reliable computers have the occasional problem.

I just worry about a single mistake causing them to miss something important.
The 2 quorum system pretty much guarantees that not happening, but if they are computing multiple variations of very similar sets of data maybe it's not such an issue.