The percentage matters, or would in many things. How can you say that .001 second matters and .002 doesn't? In math and in many real world things it does double the effect. In pool time of contact is just one of a handful of variables. We don't have hard numbers for all of these variables so we guess or totally ignore them.
How efficient is the interface between tip and cue ball? If it is typically 50% efficient we could see huge gains or losses with changes and that extra .001 might be huge. However, if it is highly efficient, over 90%, nothing we can do will show a huge gain and we may not see a significant loss with a change.
I ran the test lab among other duties at a little R&D company. Every time the results on paper were different than those in the test lab the error was on paper. In fact, the error was the same thing we struggle with in these discussions, undefined variables. Sometimes there were enough undefined variables it was simply more practical to test than to try to run down all the variables. I genuinely had a blast running the test lab!
Just because of the empirical results from our limited testing and the results of play on the table I suspect the contact interface between tip and cue ball to be surprisingly good until we get out to the outer edges before a miscue. If a soft tip gives 90% and a hard tip gives 92% or vice-versa it just isn't going to matter for any practical purpose.
I like soft tips. I like a soft contact and over fifty years ago I formed the opinion that hard tips were more likely to miscue. I find miscues to be a nonissue with any tip now but I am happy with Elkmasters and after a brief flirtation with layered tips have found no reason to change.
Hu