...because dealing with the V E R Y S L O W
signals from mice and keyboards is not a bottleneck for today's processors, so "making it more efficient" gains pretty much nothing in real terms. You have on the one hand a common, widespread, standard that works, and on the other hand mostly mythical improvements from developing a new interface and hoping that anyone will use it.
Consider how much time it takes to poll 100 times a second on a system running 2 GHz. 100 Hz (the polling rate) divided by 2,000,000,000 Hz (the processor clock rate.) A miniscule proportion of the processor time - 1 in twenty million cycles. Not going to speed things up much at all if you stop doing that, but it sure is going to cost a lot to change from USB to something new, and PS/2 ports are pretty obsolete in the other direction.
Many things that mattered when trying to bang out a computer on a 1970's/80's microprocessor don't make a heck of a lot of sense to perpetuate in 2016.
There is a great answer here for a related (more general) question on SO: http://stackoverflow.com/a/3072959/6207268
– Argonauts – 2016-12-28T04:23:29.423So many downvotes. Wow. – yoyo_fun – 2016-12-28T14:35:56.500
@Argonauts I knew all the information in that answer. My question was completely different. I know what is the difference very well. i am asking why nobody considered to make a better interrupt-based port. – yoyo_fun – 2016-12-28T16:55:41.310
1You may have known the material in that post but you either disagree or don't understand as it does tell you 'why' – Argonauts – 2016-12-28T17:37:43.433
@Argonauts I think the reason we are disagreeing here is that you don't actually understand that post not vice-versa. – yoyo_fun – 2016-12-28T18:37:29.240