Randomness does not have a memory, nor is it as uniform as you seem to think it should be.
But computers cannot generate truly random numbers, so there is a possibility that the draw algorithm is getting stuck in a rut.
I noticed this as well. In fact, it was getting so annoying that I started keeping track of how many pillars were coming up on my initial draws. After about 100 games, I realized I hate statistics and gave up on the project. Here's a bit of what I came up with (pics are thumbnails to keep load time down):
My deck:
(
http://www.screenshotdumpster.com/view/m7Gbd51292/deck2)
The deck I played varied a little from time to time, but nothing that should have affected too much (e.g, I switched out 2 Eternities for 1 Eternity and a Protect Artifact, I traded a Bone Wall for a Dissipation Shield, I upped some cards, etc).
The number of pillars/towers and deck size were constant throughout my trial.
With 17 pillars/towers, and a 47 card deck, I had:
~ 2% chance to draw 0 pillars
~ 11% chance to draw 1 pillar
~ 26% chance to draw 2 pillars
~ 31% chance to draw 3 pillars
~ 21% chance to draw 4 pillars
~ 8% chance to draw 5 pillars
~ 0.2% chance to draw 7 pillars
~ 0.008% chance to draw 8 pillars
(These add to slightly more than 100% due to rounding)
Like I said, I played about 100 games with this set up. I can tell you for sure that I drew 0 pillars in more than 2 of them. I didn't go much further than that, because, as I said, I hate statistics/probability and I didn't feel like doing distribution functions.
However, I did notice a few things worth mention. I tended to draw the same few cards over and over:
(
http://www.screenshotdumpster.com/view/m7Gbd51297/e1)
13/47 cards drawn (~28%)
1/17 pillars drawn (~6%)
2/2 Bonewalls drawn (when I had two bonewalls in the deck)
2/3 Quints drawn
(
http://www.screenshotdumpster.com/view/m7Gbd51298/e2)
15/47 cards drawn (~32%)
3/17 pillars drawn (~18%)
2/3 Dimensional Shields drawn
(
http://www.screenshotdumpster.com/view/m7Gbd51299/e3)
14/47 cards drawn (~30%)
2/17 pillars drawn (~12%)
2/3 Dimensional Shields drawn
(
http://www.screenshotdumpster.com/view/m7Gbd51300/e4)
8/47 cards drawn (~17%)
0/17 pillars drawn (0%)
2/3 Dimensional Shields drawn
2/3 Quints drawn
(
http://www.screenshotdumpster.com/view/m7Gbd51301/e5)
10/47 cards drawn (~21%)
2/17 pillars drawn (~12%)
2/3 Dimensional Shields drawn
(
http://www.screenshotdumpster.com/view/m7Gbd51302/e6)
9/47 cards drawn (~19%)
1/17 pillars drawn (~6%)
4/6 Sundials drawn
(
http://www.screenshotdumpster.com/view/m7Gbd51303/e7)
9/47 cards drawn (~19%)
0/17 pillars drawn (0%)
(
http://www.screenshotdumpster.com/view/m7Gbd51304/e8)
16/47 cards drawn (~34%)
3/17 pillars drawn (~18%)
2/3 Dimensional Shields drawn
3/3 Steals drawn (one was just used to avoid discarding)
(
http://www.screenshotdumpster.com/view/m7Gbd51305/e9)
15/47 cards drawn (~32%)
3/17 pillars drawn (~18%)
3/3 Quints drawn
I have more, but I'm sure we all get the point. Say what you want about my deck design/pillar-ratio-and-deck-size, I find it
extremely hard to say decks are "well shuffled" when I can consistently draw clumps of cards in the first third of my deck.