I acquire, by observation and by reading about the observations of others, a knowledge of what physically happens in the universe ... and then ... try to fit my reasoning to those facts without predicting phenomena which are known not to occur. Too many or too few (or incorrect) postulates results in a poor fit to the observations.
I understand that if I am able to account accurately for the most general phenomena, all the rest should fall in line without the necessity of being "forced" into a misshapen straightjacket. And ... that if I am forced to make drastic revisions to my thesis ... it is undoubtedly wrong in some or all of its initial postulates. Hence, I am loathe to "add just one more rule" to the theory. Such additions weaken it in my own estimate ... so it must be all the more weakened in the eyes of others who have little interest in the outcome.
As an example
There was a time at the beginning of my inquiries when I had only spherical bodies floating in a three dimensional void ... but after some years of "squeezing" everything out of it that I could ... I realized that it was just too sparse to generate what is observed.
Since a Euclidean grid was "assumed" in that incarnation, I gave it explicit reality and obtained another piece by means of which more "stuff" could be generated both of itself and in conjunction with the spherical frames of reference. Now, by stretching or compressing this Euclidean piece, a possible foundation for the electromagnetic interaction is obtained.
I now believe that this is sufficient to generate all phenomena. It will be tight, but that's what one must expect. I shall add no more fundamental ingredients.
I am prepared to "whittle" on this for decades till it takes a proper, natural, unforced shape and can comfortably answer to all questions.