My understanding.
Chance of spinning a pillar/pendulum = ~A snowball's chance in hell.
Thus, a deck composed of 6 copies of one card with 24 of one kind of pillar = 9001% chance of winning 3 copies of the non-pillar card.
When a pillar/pendulum would be chosen for a slot, the game re-picks a limited number of times. The result is that the odds of a pillar filling a spot are proportional to the percentage of the deck that is made up of pillars and pendulums, taken to some exponent(>1). So if it retries each slot 50 times (a number once posted by Zanz, but it was so so long ago that I don't trust it), then a deck with 80% pillars (24 pillars, 6 rares) is over 1.75
million times as likely to fill a slot with a pillar than a deck with 60% pillars (18 pillars, 6 rare1s, 6 rare2s). This is why people make farms with two rares, because the odds seem to be around the same and variety is fun.
But yes, all availbable information suggest that splitting up your pillars will not change the chance of it choosing a pillar, it just replaces some possibility of using pillar a with the same likely-hood of using pillar b, and could occasional block a player from winning 5
+ a pillar.