One of the purposes of the historical Point Allocation system is to enable players from different eras to be compared on approximately the same basis. For example, we adjust every season's numbers as if the season was 80 games in length, regardless of how many game a league actually played in that season. This is to put everyone on the same footing, and to render the numbers in a form familiar to modern eyes.
This presents an issue, however, where goaltenders are concerned. For the first 70-plus years (or so) of the history of organized hockey, a team's goaltender was expected to play every game. Now, of course, that is no longer the case, and sometimes (in the 1980s, for instance), a starting goaltender would be expected to play only about 50 of his team's 80 games. So when we're comparing the careers of a goalie from the 1920s and one from the 1980s, they're not really on the same footing, because the earlier netminder will be given credit for a significantly higher number of games per season.
To account for this, when dealing with modern NHL goaltenders we don't calculate their effective games played based on a season of 80 games. Instead, we find the goaltender with third-highest proportion of minutes played in the league, and use that as the maximum instead of 80 games. So if league-season features goaltenders who played 64, 62 and 58 effective games for their teams, all goaltenders in the league will have their games played adjusted to 58 games rather than 80 (anyone with more than 58 is capped at 80 games, of course). This means that at least three goalies will be credited with the maximum 80 games for any season. There's no particular reason that it's the third-most rather than the second or the fifth, it's just an attempt to be fair. Most starters after the third-highest will be in the 70-plus game range, which is basically what we're shooting for.
The question is, when do we need to start applying this method, rather than just using the 80-game basis? Well, some teams have experimented with two-goalie systems at times. In 1912/13 and 1913/14, for example, the NHA's Ottawa Senators had both veteran Percy LeSueur and young Clint Benedict (both Hall-of-Famers), and went into both seasons intending to rotate them in the nets, rather than just keeping one as an emergency backup. In 1912/13, LeSueur played 18 games (994 minutes) and Benedict 7 games (215 minutes). Since the team played 20 games, that means there were five games in which one relieved the other. In 1913/14, they had only two shared games, but they still split the team's netminding duties; Benedict played 9 games (474 minutes) and LeSueur 13 games (773 minutes). But this was a very unusual situation.
What we really want to know is when teams, as a whole, no longer expected their goaltenders to play every game. As late as 1957/58, three of six NHL teams had goalies who played all 70 games of the schedule. Goaltenders still normally played every game. In the next two seasons, Glenn Hall was the only goaltender to play every game, but there were also two or three other guardians who played almost every game. So it seems that they were still expected to play every game, it's just that over a long 70-game schedule, that's not always going to be possible.
When we get to 1960/61, we still have Hall playing every day. However, no other goaltender played more than 59 of his team's 70 games. And then two seasons later we have no goalies playing every game, though several missed only a few games. In 1964/65, Roger Crozier played 70 games, but the next-highest total was only 53.
It seems to me, then, that 1960/61 serves as a good demarcation point to begin using the "modern" system for calculating netminders' games played for Point Allocation purposes. That's the dividing line that will produce the fairest results.