Good point. If only I could double my like. But others welcome.
You disagree with me, but I do actually fully agree with you! Maybe you didn't read carefully what I wrote, so I will not try to dispute. It is written on, maybe not clear enough, English is not my native language. I want to supplement your post with the following:
What really makes a vintage R2R DACs sound? R2R ladder chip or something else? I believe the later. A Philips TDA1540 came together with interface chips. One of a feature was 4x oversampling combined with noise shaping. It was intended to increase resolution little bit above 14-bits which was a maximum of the ladder. When a famous TDA1541 (16-bit) came out it was convenient to use the same technology to smooth the sound even further. There were already new SAAxxx chips, but none of them had offered non-processed and NOS like a modern DAC do. A number of vintage CD players came with this chip, I still have one. A short note of a history is here:
https://www.dutchaudioclassics.nl/history_of_the_Philips_tda_d_a_converter/
Regarding Burr-Brown PCM1704 (and Texas Instruments licenced version) I can't say much, I owned CD player with PCM63, it was also oversampled with HDCD chip. So I think it is similar situation. If you want to compare NOS PCM1704 sound to the current discrete ladder DACs, there is a group buy from Audio GD for a new version of DAC19.20 with this chip. It will not use noise shaping it is a pure NOS like other A-GD DACs:
https://www.head-fi.org/threads/new...hip-resistor-ladder-dacs.853902/post-16009443
With increased number of resistors in the ladder there is no reason for noise shaping. Unless... As Soekris use opamps, it is important to reduce transients of the ladder by deploying noise shaping. It is not clearly visible on your graph, as it ends on 20kHz, check a wide frequency FFT, it will show as increased high frequency noise.