Naturally stuff that's not up for interpretation, like armor penetration, can come off the dice pool before the roll.
It's just useful for those up in the air situations, like social modifiers and whatnot.
On another note, for extended tests, do you want to do full dice pools or diminishing dice pools?
Diminishing is doable, but more of a pain. Makes the game considerably tougher when you realize that only the SR4a core book had them in mind when setting up thresholds. Good luck hitting an Unwired target number with diminishing dice
this is an uncharacteristically bad example of doing an 8x8d6 extended test. I wouldn't let it cloud your choice too much.
So, this is full pool.
8d6.hits(5) → [6,1,2,1,3,2,4,1] = (1)
8d6.hits(5) → [1,1,2,4,4,1,5,3] = (1)
8d6.hits(5) → [6,3,1,6,1,1,3,3] = (2)
8d6.hits(5) → [5,3,3,1,4,5,4,3] = (2)
8d6.hits(5) → [3,4,4,1,3,2,4,4] = (0)
8d6.hits(5) → [3,6,1,1,1,1,6,6] = (3) Glitch: roll 1d6 and subtract # from hits so far.
8d6.hits(5) → [6,5,4,3,5,5,1,6] = (5)
8d6.hits(5) → [4,1,4,5,4,6,2,1] = (2)
If this was a Data Search, the player with an 8 dice pool would have been able to find publicly available information about a common, popular topic with 4 minutes of searching, and probably had unpublicized information after 8 minutes. Hidden and actively erased info would have still eluded him. Again, these were some garbage rolls.
and this is diminishing
8d6.hits(5) → [6,1,2,1,3,2,4,1] = (1)
8d6.hits(5) → [1,1,2,4,4,1,5] = (1)
8d6.hits(5) → [6,3,1,6,1,1] = (2) Glitch: roll 1d6 and subtract # from hits so far.
8d6.hits(5) → [5,3,3,1,4] = (1)
8d6.hits(5) → [3,4,4,1] = (0)
8d6.hits(5) → [3,6,1] = (1)
8d6.hits(5) → [6,5] = (2)
8d6.hits(5) → [4] = (0)
If this was a Data Search, the player with an 8 dice pool would have been unable to find publicly available information about a common, popular topic with 8 minutes of searching. Zilch.