I think one of the irreconcilable differences when it comes to doing all-time rankings is an observer's personal perspective on the weighting of peak vs. career. I dislike the expression "agree to disagree", because I think it is usually a retreat tactic employed by someone who has run out of arguments, but I think it sometimes legitimately comes into play during all-time debates when an impasse has been reached because of a fundamental difference in evaluation criteria.
Conceptually, I think there is one group of people that approaches the issue as if they were a general manager looking to build a hockey team in an expansion draft. The draft eligible group consists of every 18 year old kid that has ever played hockey, and their goal is to pick the one that provides the highest contribution to that team over their entire career. In this scenario, things like durability, consistency, and longevity are important considerations.
The other group comes at it from a totally different angle. They tend to think of themselves as the coach of a one game playoff. They can pick anybody who has ever played to be on their team, and each player will participate during the prime of their careers, at the absolute height of their skills and abilities. In this scenario, durability and longevity are very much secondary considerations.
This is an oversimplification, because I don't think you will find many people that will take either case to their extremes (e.g. take Mark Recchi over Guy Lafleur, or Jose Theodore over Ed Belfour). Both peak and longevity should have some weighting, but at the end of the day everyone is going to favour one or the other.
Anyone who has read more than a few posts on this blog will be aware that I fall into the second group. For that matter, anyone who has just read this website's title could probably guess that correctly. I've said this many times before, but I am far more interested in how good somebody was than how valuable they were. When a player has retired, and a few decades have passed, and the memories have mostly faded away, the only thing that sticks in the collective recollection of hockey fans is the talent and skills that player displayed when they were at their best. Bobby Hull unleashing a wicked slapshot, Rocket Richard cutting towards the net, Bobby Orr wheeling up ice, and so on. For anyone who is over 25, picture Wayne Gretzky as a player. Is he wearing the jersey of either St. Louis or New York? I didn't think so. When your grandchild asks you, "How good was Dominik Hasek?", he is going to be a lot more interested in how Hasek played against the Canadians and Russians in Nagano than how he did as a 40 year old platooning with Chris Osgood.
If you had a time travelling machine and you went back in time, collected every goalie who has been a top starter in the NHL at their peak, gave them all access to the same nutrition, training, equipment, etc. (to minimize era effects, as otherwise you would just take whoever is the best in the league today), and then lined them up against a wall to pick sides, I might not even glance Brodeur's way until a dozen guys are already off the board. But if I am a GM looking over the 18 year old versions of every goalie who has ever played NHL hockey and trying to figure out who I want in net for my team for the next 2 decades, I would give Brodeur a much closer look.
In summary, when I rank hockey players, I tend to divide them into tiers. Within each tier, things like longevity and durability and so on become important, because if you had to choose between two guys of similar abilities then you would rather have the guy you can count on to deliver every single night. But a flaky superstar always beats a consistent very good player because he is on a different level, and all the consistent 30 goal or 80 point seasons in the world don't bridge the gap, in my estimation. Which is why I don't consider career records to be that important, whether they are career wins, shutouts, goals, assists, passing yards, strikeouts, whatever.
Here's a discussion question, for those who like the career value approach: Let's say Chris Osgood, through some experimental genetic engineering or possibly a fortunate archaeological find, is able to stay eternally youthful and play goal in the NHL for as long as he wants. Let's also say that Osgood stays on a perennially strong team in Detroit, continues to put up his customary slightly-above-average save percentages, and averages 30 wins and 5 shutouts per season and one Stanley Cup win every 10 seasons. How many more years would Ozzy have to play before you would rate him ahead of Dominik Hasek in your all-time goalie rankings?
(In the unlikely event that there is someone reading this that doesn't like goalies, I'd suggest substituting Rod Brind'Amour, 70 points, and Mario Lemieux into the previous paragraph to get a more-or-less equivalent scenario for skaters).