I do not believe that the seating depth has anything to do with accuracy.
I will gently disagree (and will admit I am a little surprised).
We don't have to agree and nobody should take offense at the concept that primer seating to fine degree has affected group size or accuracy, or if they don't see it.
I will admit, not every system is capable of showing it when other dispersive parameters are drifting to a larger degree.
For my own outings, it just happened that the same levels that are spoken of by many here and a few of the ones on YouTube), have ended up being optimal at one or two thousandth plus or minus a thousandth of crush (assuming we all agree on the definition of crush).
While I was still working, I ran independent QC of primers of several kinds and batches. I can say that when you take a very close look at the issue in several fine dimensions, i.e., surfaces instead of single linear dimensions, the height of a primer anvil above the cup isn't without challenges.
They are often less than flat or parallel to the cup, and this is in addition to the dispersion on that average. Those errors in parallelism, flatness, imperfections, rough edges, etc., can account for a good chunk of that one or two thousandths, and that is before you can even say where the surfaces of the cup to anvil would average.
It should not be a surprise then, that it takes a little "crush" just to get the anvil down to first touching, then parallel, then preloaded, and then include a small fraction for deflection (Hertzian Contact Stress) in the tooling and primer components, before we see the optimal performance.
Since my own personal rigs have shown the performance changes with seating depths that either come short of this, or exceed it by very much as
@mikecr is pointing out,
my own opinion is what we are really doing is just trying not to screw it up. In different words, we are getting the anvil to flatten out and slightly preload to get them to their baseline design. There is no law or rule that says someone else may find a point short of this or beyond it that will give them an optimum.
Yes, they will ignite when short of this place, and also when they have been "crushed" much more, however the former invites timing delay and harmonic issues, and the second invites reliability and dispersion issues.
There are some specifications that require primer component testing. The tooling includes concepts that guarantee the activation energy of the strike tooling. The tooling is used to test exactly how much sets off a primer. When we seat those samples, we are very picky about their dimensional position to keep the testing fair.
As such, the opportunity to screw around with short seating and heavy crush, is something I couldn't resist.
You would be able to see the shortcomings of both conditions. It is just my own impression, that the baseline design is to flatten the anvil defects and just a little past that point, and stop... Not to come short, or to crush farther. I also think it isn't a coincidence that is where I get my own best performance.
Primer specs could be done better, as could cartridge brass specs, but committees are not a user to producer majority. When the manufacturers have too much control, sometimes things that should converge on an improvement will be traded off in a compromise to get anything done at all. Just one opinion and nothing to get worked up about... YMMV
Merry Christmas and Happy New Year!!!