User blog comment:BunsenH/Ghazt Stats Phase 3: Wishing Torches - Report 2/@comment-26245311-20150326004904/@comment-24577221-20150326031202

Yes, trials #2003-4004 had only 30 successes, giving a success rate of (1.50 ± 0.27)%.

I wasn't planning to post another update until we had around 8000 trials (doubling each time), but since you've asked... as of 6025 trials, we had 113 Ghazts, giving a success rate of (1.88 ± 0.18)%. So that third "group" had 2021 trials and 36 Ghazts, with a success rate of (1.78 ± 0.29)%.

There's a lot of variation between the sets, as you see. That's typical when the chance of success on each trial is so low.

The uncertainty figures are standard deviations &mdash; as in, 68% of the time the experimental value is expected to be within one sigma of the true value, 95% of the time within two sigmas, and so forth. So it's not too surprising that the groups' ranges don't appear to match.

My guess is that each torch gives a 1% bonus, at least for this combination. I don't have any real data about other combinations, just vague suspicions. I tried using wishing torches a couple of times when I was going for Entbrats, and still got failures, so I'm pretty sure that the effect on higher-probability breeding combos isn't anything like adding the initial probability again or exponential increases. My guess is that it's going to be 1% per torch across the board, a linear effect. My plan is that after we've got better stats with 1 torch, probably at around 8000 or 10000 trials, we'll try with 4 torches (12, 22, then perhaps 92). That should show relatively quickly if we're dealing with an exponential change.

The only real way of speeding things up would be to increase the number of games being run, e.g. by acquiring more "minions" or getting them to run more games each. Now, I've already got them each running another game, to look at the chances of getting Entbrats. That's just getting under way, and it's too early for me to report results, other than: yeah, 3-element + 1-element is much more likely to give success than 2-element + 2-element (at least for the combos we've tried). We're kind of running up against limitations on the demands that I can put on them, and on my time and energy to coordinate.