I'll include this again, as I was adding/editing it to a previous post as you typed your reply.
Andy, I think you've really hijacked this thread to shout down what you feel is bad and dangerous behavior in a diver, rather than answer the basic question.
Did the diver ignore his computer? Yes. Is this bad? Maybe, but with the OPs level of understanding of deco theory and physiology, probably yes.
Did the diver know from past experience that his computer will tell him he has a shorter NDL than his diver partners? Yes. Is this different than the many thousands of technical divers out there who "negotiate" the ascent curve and length of their deco profiles, and will lengthen or shorten those stops, and thereby deviate from the algorithms? Not so sure.
When I dive my gradient factor (GF) 70/70 profile and my buddies dive their equivalent of a 10/85, and we get out of the water at different times, running different ascent curves, and are all unbent, is that wrong? No.
This thread is riddled with unnecessary histrionics.
If the diver can run through Subsurface with his entire dive trip to see the impact of repetitive dives and what the expected decompression obligations should be for those dives, and then he finds that based on the chosen GF he should not have been in deco, my question is why does he have a question? Clearly, his computer is running an algorithm that he seems to feel is excessively conservative.
So my question to the OP is: Now that you have validated to yourself that your computer has a questionable implementation of a decompression algorithm, what are you going to do about it?
My recommendation is that if you want to continue to use a computer you should use one that implements a standard algorithm. An algorithm proven to introduce less decompression stress and sickness in a number of recent studies performed by leading decompression scientists and researchers and national navies. One that also allows you to change the "conservatism" based on what you're comfortable with. The commercial algorithm that does that is Buhlmann (ZH-L16C) and the best line of computers on the market with that algorithm is Shearwater (Petrel, Perdix, etc.). Case closed.