P=NP is impossible on a turing machine

The reason?

Because you can not add in one motion.  It is impossible to be given numbers: 3,4,6 and add three to each in one motion.  If this was possible NP complete problems would be simple.  Without this criteria it is impossible to solve NP complete problems without exponential time.

The problem of NP complete problems are not suited to the computer.  Another means of calculation would be better suited.

>>1 was made hasty and with no thought behind it.

Basically P=NP is just a bitch because with traditional computer methods (numerical) it becomes exponential due to all the additions that need to be made and the act of the final "is this number equal to the sum" check which is also exponential.

Pictorial systems that model numbers closer to reality suffer from the obscene space requirements to store precise information.  For instance if you wanted to store the number 529520 in a pictorial system it requires 10^6 bits.  Of course you get the ability to store a lot more numbers in this space than just one (since overlap means jack shit in NP complete problems).  Although you get quick addition, final check, and copying (sort of if you take it independent of the huge storage needed). 

Of course the "reality computer" has these obscene space storage abilities although I doubt we could ever use such a thing due to the imprecision of physical measurements.