Why the need for the change:
The point-buy system in the Wilderness of Mirrors rulebook looks neat on the surface, but (as I'd mentioned in my initial pre-game post) it boils down to only 6 different arrays of stats.
- 5, 5, 5, 1, 1
- 5, 4, 3, 2, 1
- 4, 4, 2, 2, 2
- 4, 3, 3, 3, 1
- 5, 5, 2, 2, 1
- 3, 3, 3, 2, 2
To save the group a lot of time learning crunching numbers, I wrote down those 6 permutations and brought it to the game I GM'd yesterday.
All four players in my group looked at that an decided 5, 5, 5, 1, 1, was superior to the other five possible arrangements. And honestly, they're right. It gives you a total of 17 dice, whereas the other versions give between 13 and 15 dice total. Analyzing those trade-offs, The more you think about it, the worse the other set-ups look, because you'll be rolling one extra die in your weak area at the cost of losing a total of three (or more) dice from two of your best areas. And it's not like that second die in your weakspot is going to really help much - you'd need to roll an 11 or 12 on 2d6 to get any measure of narrative control.
My group is not typically a munchkiny bunch - we run weird one-shots and run lots of goofy narrative systems. Realizing how heavily the system rewards that choice, the decision was easy. When one option is so much better than the alternatives, no one wanted to be the only person left in the dust.
This quickly developed into several different (but related) problems.
All or nothing: With the stats being just 5's and 1's, that means that for any given stat you either always have narrative control, or never have narrative control (since you're looking to roll a 11+ on either 5d6 or 1d6). If it's just a binary function, why spend time rolling dice and adding numbers? Okay, I'm over simplifying it - there's actually plenty of times you want to roll 16+, which means the options aren't "yes" and "no", they're more like "50/50" and "no". That's really not an improvement.
Character redundancy: With all characters having the 5, 5, 5, 1, 1 stat arrangement, you can be certain that any two characters will have one stat that's rated the same - they'll overlap one of the 5s. This is detrimental to accepted notions such as individuality, character concept, and character niche. Instead of "I'm the assassin, and he's the techie" you end up with "We're both assassins and techies, but I'm also the team leader and he's also very stealthy".
Saturn is a dump stat: The character overlap problem gets even worse because you almost never roll Saturn. The team leader needs it at level 5, everyone else is going to want to rate it at 1. Which means the above problem of character redundancy just got worse, since every character except the team leader has a Five-Star rating in 3 out of 4 remaining stats. This also has the unfortunate effect of making the team leader the weakest character in the group.
Special powers: The highest-ranked PC in a Stat gets a corresponding one-time-use power. If there's a tie, they both get that power, but whoever uses it first uses it up for the whole group. For our group of 4 players, this meant that the special power for Saturn went to the team leader. The special power for Mercury was shared between two players. The special power for Mars was shared between 3 players, as were the special powers for Vulcan and Pluto.
It was ugly.
We all looked at it and said "this isn't gonna work". After a few seconds of pondering aloud, I instructed everyone to switch to the 5, 4, 3, 2, 1 spread instead. That change resulted in characters that overlapped less, but it still resulted in 4 out of 5 characters having a "1" in Saturn. It was better, though. With this spread, everyone felt like they had a focus, and the characters weren't just carbon copies. We played it out, and were happy with the results (though Saturn still seems like a dump stat).