Only if he goes on to talk about the statistics of large numbers, just choosing some arbitrary numbers, dividing them and looking at the mantissa is not making a useful point at all.
It almost looks like 384000 might have been truncated as text to 3840, then divided by 3474 - so as to treat them both as 4 digit numbers. Doesn’t make any sense, but it fits for the other examples you gave - it’s as though the placement of the decimal point for both numbers is dictated by its placement in the divisor.
The OP has established that the guy spewing this nonsense is a mathematically illiterate numerologist.
Not much more to usefully say after that. The guy may as well assert that 5 = 15 for all the sense his blatherings make to actual non-wacky people.
I didn’t burden you all with his ‘probability’ calculations, because they weren’t even wrong in the Paulian sense.
I hoped that there might be a nugget of truth to the normalization scheme, or a kernel of Benford’s Law in the leading digits, but it seems less likely than not.