It’s Stereophille
https://www.stereophile.com/reference/arc_angles_optimizing_tonearm_geometry/index.html
and if to "cut" the core is in the very end so will just copy it:
In all that has gone before, we have assumed that the Löfgren-Baerwald approach of minimizing the maximum LTE distortion across the modulated extent of the groove is indeed the optimal one. But there is good reason to suppose that it isn’t, because lateral-tracking-error distortion is only one form of distortion afflicting vinyl-disc replay. Another source of distortion is tracing error, caused by the replay stylus being unable to follow the same path through the groove as the cutting stylus. In large part this is a function of the shape of the replay stylus, and depends critically on groove curvature. Because of this, tracing distortion worsens toward the end of a record side, as the waveform cut into the disc bunches up.
Although tracking-error and tracing-error distortions are different, there is clearly a case for supposing that, as tracing error increases toward the end of the side, tracking error should decrease. In this case, the optimal arm/cartridge alignment might be one that results in a distortion-vs-groove-radius curve like that in fig.10. But as tracing error differs according to stylus type, this approach also suggests that arm/cartridge alignment should vary accordingly, with a bigger disparity from the conventional alignment for a conical stylus than a line-contact stylus. The concept of a single "right" alignment would then be redundant.
