Indeed. It's obviously difficult to prove, but it is generally thought that the great majority of chromosomal 'mistakes' which arise during the reproductive process are incompatible with life, and therefore do not result in any live offspring. Of the remainder, most do not confer any particular 'survival advantage', and therefore do not preferentially proliferate - and that just leaves a tiny number which do offer a 'survival advantage and therefore do preferentially proliferate ('survival of the fittest') - hence 'evolution'.
So I wonder whether those facts transfer to 'evolution of language'? Just as with chromosomal mutations, I presume that millions of people make 'random mistakes'in the use of language every day, but the vast majority of those mistakes are one-offs and/or are person-specific, and hence do not proliferate. However, just a tiny proportion of those mistakes do 'catch on', and proliferate, so one has to wonder whether they afford some sort of 'advantage'. If not, why do only a tiny proportion of linguistic mistakes come into such widespread use that they come to be accepted as 'evolved language'?
You and EFLI are not alone! There are a good few aspects of current use of English which annoy and irritate me intensely, sometimes 'sending shivers down my spine'. However, I'm not sure that I classify them as 'mistakes', given that they are pretty ubiquitous in my offspring's generation - but I certainly would not waste my time 'correcting' use of language which was regarded as totally normal by a whole generation!
Kind Regards, John
