I would have agreed with you in the past, when I used Audyssey XT32 - now with Dirac Live, I find that not to be the case - the setup sounds better with Dirac, than in Pure Direct.
I did (during the change over) compare the old AVR to the new one - and in pure direct, they sounded identical (to my ears)... but the old one on 2 channel stereo music, was consistently better in pure direct than in Audyssey XT32, the new AVR is consistently better on 2ch music with Dirac active than in Pure direct.
There are obviously numerous potential caveats - but for myself, I have found that Dirac has put forward a very very convincing case for its benefits, regardless of the material being listened to. - The key Caveat, is that I was using Audyssey on a 2013 vintage AVR - there was no option to turn off the MRC default - and that 2kHz BBC "dip" that is built in (supposed to compensate for common issues in crossovers between midrange and tweeter) - really impacts on the midrange - BADLY. - Current D&M AVR's with the associated apps, give you the option to disable MRC... and my experience MIGHT have been different.
In your case - assuming a D&M AVR, have you turned off MRC before comparing the EQ vs the nonEQ for 2 ch music?
Also - have you tried to use the EQ, without any target curve... (that requires matching the target curve to the measured curve as closely as possible, so processing is then limited to impulse/phase and frequency EQ is minimised) - that would allow a comparison of the processing without altering the voicing of the system... this is essential, as both Audyssey and Dirac do both things, processing and voicing - and if you want to know the efficacy of the processing, you need to take the voicing part of the equation off the table!