The Nalimov EGTBs in any case are overkill. You have completely solved (Basic Rules) chess if you can produce a routine that guarantees a win from any winning position. It doesn't have to be in the minimum number of moves and it doesn't need to produce more than one move for a given position.
A chess beginner with an aptitude for programming could probably produce a routine to replace the 14 KB KRK Nalimov EGTB in less than a couple of hundred bytes if you drop that overkill. I could also replace the 1 MB + KBNK EGTB in around 1 KB with the same proviso.
Such EGTB replacement routines are no doubt possible for all generic endgames (including all 32 pieces). Moreover I think the ratio of the size of such routines as a fraction of the corresponding EGTB size would probably exponentially decrease with EGTB size.
The problem is, of course, that the larger the EGTB size, the harder is the analysis for such replacement routines, but this is not the same as saying that such routines do not exist.
If storage capacities continue to grow at their historical rate, I wouldn't be surprised if in a decade or two, a complete solution to chess would fit on the average person's mobile. I would be surprised if it actually did.
You're posting a paradox 
. There is no "solving" for "basic chess rules" that way. If you want to evaluate "winning positions", then you have to prove what is actually, provably "winning" first. Doing that requires traversing the 10^46.7 positions. It's not enough to say "this result draws/wins every game ever played" when every game ever played by man or engines is still a tiny fraction of total games.
An extrapolated proof is only good if you can be certain your extrapolations are valid, and in this case, you can't prove your extrapolations are valid without looking at all the positions and backing it out.
If you want to solve chess, as in the definition of "solved games" ala checkers, there isn't a shortcut. Even if you were certain you had the solution, how would you even prove it was correct? You'd still have to traverse the 10^46.7 positions...
P.S. If a chess beginner could just write a 100 byte routine and solve the game, it would have been done by now, because there has been a good amount of interest. Any engine developer ala Hans Berliner or whomever could have taken it on long ago. Don't try to posit an easy solution that you just don't have the time or will to come up yourself. You sound like Geodexic 
.
You are looking at a problem with 10^46.7 as the result set and then saying you can increase the calculating efficiency from 14K to less than 1K. Great. Only 45 orders of magnitude to go. This argument has been put forth dozens of times in the "will computers ever solve chess?" thread, but reducing the problem set by even 99.999999% still does not make a realistic dent in the calculation and storage required. It's still galaxy sized supercomputers trying to store a solution the universe isn't big enough to hold.
PATRIOT We can only go by what you actually post--not something that you might have intended to post?
The funny part is posting a "less obvious position" example that still has an obvious resolution.