in reply to Re^2: Stop with the interview questions already in thread Stop with the interview questions already
Stepping aside from this heavy discussion to a lighter moment ;) ... I guess an interview question like "How do you divide by 2 without using the divide function?" is probably not intended to test your knowledge but to test other desirable skills/qualities, and the way you answer it would tell something about your personality. Perhaps there is a wittier answer to it than the obvious "multiply by half" or the dull "not possible".
Reminds me of this incident ... my brother was once asked this question towards the end of an interview for a programmer job,
Q.: What is the probability of picking a red ball from a bag containing equal number of red and blue balls?
A.: 0.5
Q: Ok. How would you increase the probability of picking a red ball from the bag?"
Without going into the theory of probability to explain that there wasn't a fair way to do so, he just replied jokingly, "Paint one of the blue balls red".
Re^4: Stop with the interview questions already by Jenda (Abbot) on Sep 09, 2009 at 10:26 UTC 
I think the "divide by 2 without divide function" was meant to test whether the person knows about bit shifts. Bit shifting and bit masks are often overlooked and often very useful. And I am not speaking about microoptimizations by replacing division operators by bit shifts.
Jenda
Enoch was right!
Enjoy the last years of Rome.
 [reply] 

Not to nitpick too much, but...
Perhaps the only goal is to find out whether or not they know about shitbifting, particularly if they'll be expected to optimize code on any sort of particular hardware.
There's another potential motive, however. And that is whether or not candidates understand how it's possible to model real systems using discreet mathematics  particularly that subset of discreet mathematics bounded by such things as bitwidth.
That Von Neumann was a real stinker.
Cheers,
Matt
P.S. Speaking of discreet systems. Ever tried to figure out whether or not two integral functions intersect or not (or how often)? Euclid had it figured out. It's useful for figuring out potential solutions for diophantine systems of functions.
 [reply] 

 [reply] 
