Contradiction of ARRL's RMDR

[JA2GXU]

My friend of Ham radio operator said receiving desired weak signal readability ranking are
DS_SDR > TS-590 > FTdx5000 at 20m band,
however FTdx5000 can't detect and read to the same desired readable weak signal (DS_SDR is direct sampling SDR). Those are DS_SDR heard 100% desired readable weak signal, TS-590 detects but heard not complete reading about the same desired readable weak signal and FTdx5000 heard nothing about the same desired readable weak signal.
Then I checked ARRL RMDR (Reciprocal Mixing Dynamic Range) and result are
-65dBc >- 94dBc >-104dBc (under -100dBc is better),
thus almost reverse results and contradict the fact of above readability ranking.
On the other hand, ARRL wrote "In many cases, reciprocal mixing dynamic range (RMDR) is the primary limiting factor of a receiver's performance", am I hearing this right?
Unfortunately I can't understand what RMDR, thus can't related to clock oscillator (CO) phase noise and local oscillator (LO) phase noise, and also actual reciprocal mixing test.
Reciprocal Mixing is absolutely related to CO and LO phase noise, unfortunately RMDR can't have those phase noise and can't discus about those phase noise, thus RMDR can't related to reciprocal mixing and it effects.
As above problems then I am thinking about reciprocal mixing and effects under the condition of typical DX pileup situation and typical low band situation.

[w-u-2-o]

Your friend is wrong in his rankings, and there is no contradiction.

https://www.remeeus.eu/hamradio/pa1hr/productreview.pdf

http://www.sherweng.com/table.html

[JA2GXU]

Are you read my contribution in this forums "Reciprocal Mixing and useless any RMDR".
This my contribution includes "Contraction of ARRL's RMDR".

73
JA2GXU