I think it can have some merit in certain, very clear-cut seasons. Take for example 2006:
NORRIS: Nicklas Lidstrom, DET 1,152 (91-28-8-2-0)
2. Scott Niedermayer, ANA 817 (29-57-16-15-3)
3. Sergei Zubov, DAL 464 (0-21-42-31-14)
4. Zdeno Chara, OTT 430 (5-14-35-30-17)
5. Wade Redden, OTT 115 (0-1-12-12-12)
6. Mathieu Schneider, DET 111 (4-5-1-4-19)
7. Chris Pronger, EDM 97 (0-1-5-15-20)
8. Dion Phaneuf, CGY 61 (0-0-6-6-13)
9. Bryan McCabe, TOR 44 (0-1-2-6-9)
10. Lubomir Visnovsky, LOS 39 (0-0-2-6-11)
One could argue that if this cohort of defensemen existed in 1966, Niedermayer would have been the clear-cut Norris winner. It would also have been regarded as a very weak season for defensemen, like the real 1966, but a Norris is a Norris to some extent.
On the other hand, I don't see how this:
1.Chris Chelios 201 (33-10-6)
2. Ray Bourque 97 (6-19-10)
3. Larry Murphy 93 (9-11-15)
can be inferred as a Norris for Bourque in 1993 or how this:
1. Brian Leetch 494 (42-8-3-1-0)
2. Vladimir Konstantinov 178 (2-10-13-6-5)
3. Sandis Ozolinsh 176 (2-12-9-8-3)
4. Chris Chelios 172 (0-7-18-9-6)
5. Scott Stevens 171 (7-8-4-7-4)
6. Nicklas Lidstrom 60 (0-5-2-3-6)
7. Darryl Sydor 45 (1-1-1-5-8)
8. Ray Bourque 45 (0-2-2-4-9)
9. Eric Desjardins 21 (0-1-0-3-5)
gives Stevens a Norris in 1997. Voting patterns are much too fickle to be applied in that kind of linear fashion.