OK, I thought about cost and results are below.
Lets think that you use 3 deflagrations on random permanents. It'd cost 6
.
Remember it's random, so let's substitute 1 quanta.
You use deflagrations (at most) to destroy shields, weapons or permanents that aren't pillars. This can destroy pillars, so let's substitute another 2 quanta.
As you see, it decrease costs by half, what do you think now?
Lets use this line of thought
3 deflagrations = 3 cards and 6

This is one card so it is easier to draw and can destroy more than an equal number of deflagrations. To underestimate lets call this 4

(Aether for card generation [Mindgate])
Now it is indeed random and has to occur all at once. This means that it will destroy a less valuable 3 than 3 deflagrations. This could be worth -1 quanta.
Changing a duo cost into a mono cost is worth +1 quanta.
Total for Deflag comparision 10

.
Deflag is used for non pillars, this can also act as Earthquake instead
1 earthquake = 1 card and 3

Total for Earthquake comparision 3

.
Now the question is how is it random? Does it count each pillar separately? or just the slots?
Assuming slots there is a 2:1 ratio or greater of non pillars to pillars (i think)
So final underestimation would be (2x10+3)/3=~8 quanta.