Originally posted by James Winward-Stuart:

This actually doesn't sound too bad - it looks scary slow-play-wise when written down, but when you think about actually doing it - assuming the opponent hasn't gained life up to a huge number, and doesn't have a vast token army (and even if they do, it'll soon be whittled down) - it can actually be done fairly quickly* (even if the opponent doesn't scoop as soon as they see it start…). Statistically, it could in theory never win, but in practice it will, and it won't take too long.

It's not hugely relevant to the discussion at hand, but this doesn't seem to be the case. The ideal case (AP, NAP, GTP, Copy1) presents a 1/4 chance of success, and each failure decreases this chance since you add another copy. Running a Monte Carlo simulation for 10 successes (i.e. 20 life), the expected number of loops exceeds 250,000. Even for 5 successes, it exceeds 4000 iterations.

Originally posted by James Winward-Stuart:

I'm inclined to the view that doing it manually is advancing the game state (due to the growing pile of creatures etc. - if this isn't advancing the game state, then neither are most Modern Twin decks…).

Since the actions are advancing the game state, and it won't be as slow to get anywhere as things like Eggs, this seems fine to me.

So I was actually curious as to what the math on this would look like. I'll report back when ive come up with something.
Each activation increases the number of targets linearlly. Assuming that the only targets are you, the first test pilot, the first activation has a .25 chance of a desired outcome, the second a .2, the third a 1/6, the fourth a 1/7, the fifth a 1/8 and so on. So the probability of killing your opponent, even within 10,000 activations, is very very small.

It's not hugely relevant to the discussion at hand, but this doesn't seem to be the case. The ideal case (AP, NAP, GTP, Copy1) presents a 1/4 chance of success, and each failure decreases this chance since you add another copy. Running a Monte Carlo simulation for 10 successes (i.e. 20 life), the expected number of loops exceeds 250,000. Even for 5 successes, it exceeds 4000 iterations.

Damn, beat me to it. But yes, not only do failures add to the number of targets, but success do as well.

Yeah, my bad, I completely forgot to account for the increasing number of Test Pilots when I sketched out the maths. Oops.

We've established the basics of the scenario, so endlessly poking at the
numbers won't get us anywhere. This isn't the place to discuss optimization
of the combo, no matter how fascinating it may be. Unless there's something
actually new and useful to add, I feel like this discussion is done.

On Fri, Jul 17, 2015, 22:51 James Winward-Stuart <

But the number of test pilots increases way slower when you change the shortcut to “When the Goblin ability hits me or my primary Goblin hold priority, otherwise let it resolve” since he wouldn't care for his copies being killed since they already did their job. That means the number of goblins increases way slower. And to be hosest, like 15 activations should be enough to have enough hasty attackers to kill your opponent anyways…

As has been noted, this conversation is digressing away from the original intent (which I believe I already answered), and spiraling into loop theory. There are other places for that (and most of them serve ale).

d:^D