Which I thought was neat, and made me think about extended rolls in Exalted.

Extended rolls (3e Core, p.189) are typically used to determine if a player will succeed in time. They're good for something like a footrace (first to 10 successes wins) where each roll represents an independent unit of effort that sums linearly to the final outcome.

What I like about the decay idea is that it combines the dramatic benefits of sequential, cumulative rolls with something like "what is the ceiling of the character's capabilities". If you keep banging away at a big problem (say, writing a novel, building a house, cracking a cipher) or anything that benefits from advanced planning, taking your time to polish and refine and redo should be better than your first attempt, but you would eventually expect to see diminishing returns as you run out of new things to try, improvements to make, etc., the character achieves the best version they can of whatever it is, given the specific circumstances.

It also has an in-built push-your-luck mechanic, because as the pool gets smaller, the chance of botching one of the rolls goes way up. (not sure

*exactly*what to do with this; using the normal extended action "any botch means you start over" might be too draconian for some applications).

So I wrote a quick simulator to see what sorts of numbers this would generate. I immediately ran into a problem in that, if you only remove 1s, even with a modest (for an Exalt) dice pool like 10, at you end up making an average of 28 rolls (90th percentile of 44) for an average total successes of about 18 (90th percentile of 29). And that's a frankly unreasonable number of rolls to expect someone to make at the table. I didn't account for botches but it does also mean that you are

*guaranteed*to botch eventually because you keep rolling until you have 1 die left and it shows a "1".

But since I already had the simulator written I decided to see what would happen if your removed any dice that did not show

*a success*instead. This felt much more interesting (and much more reasonable to actually do at the table).

10d@diff1: Average rolls 3.7 (90th percentile 5), Average successes 8.3 (90th percentile 11)

10d@diff3: Average rolls 3.7 (90th percentile 5), Average successes 3.7 (90th percentile 7)

10d@diff5: Average rolls 3.7 (90th percentile 5), Average successes 1.5 (90th percentile 4)

25d@diff1: Average rolls 4.7 (90th percentile 6), Average successes 20.8 (90th percentile 30)

25d@diff3: Average rolls 4.7 (90th percentile 6), Average successes 14.3 (90th percentile 22)

25d@diff5: Average rolls 4.7 (90th percentile 6), Average successes 10.1 (90th percentile 16)

Anyway, does anyone else see potential for this type of "extended roll with decay" or "exploding successes" (depending on your point of view) for various subsystems? Just curious.

## Leave a comment: