buyout button

[Pete Mack]

Don't cast the spell, if the failure rate is 63%. ID is easy to find in stores, so stock up on it. (20 or more scrolls of ID is not an unreasonable burden, especially for rangers...)

[Derakon]

JonnyOThan was more complaining that 63% failure rate spells seem to fail a lot more often than 63% of the time, and suggesting that there should be a cap on the maximal number of failures in a row before getting a success (though it seems to me like that would need some rules to prevent it being abusable, i.e. "cast" the failures, then save the success for when you need it).

[Pete Mack]

I don't think this is a good idea-- imagine applying it to a 5% fail spell...

[zaimoni]

JonnyOThan was more complaining that 63% failure rate spells seem to fail a lot more often than 63% of the time, and suggesting that there should be a cap on the maximal number of failures in a row before getting a success

While there are some subtle issues with the current RNG exposable by careful savescumming, this is not one I can empirically document. The problems have more to do with excessively long sequences of highly probable events. (Generally, anything less probable ~1/27 with expected probability p, has little trouble problem attaining a streak of 3/p failures far more than 5% of the time.)

(though it seems to me like that would need some rules to prevent it being abusable, i.e. "cast" the failures, then save the success for when you need it).

True. I've considered something like this in reverse. (Infer savescumming by severe deviations between expected probability and measured incidence, then "dissipate" the accumulated karma.)

[Pete Mack]

Finally installed this after failing to get !rLife twice. Minor code fix:

	menu.cmd_keys = "[COLOR="Red"]K[/COLOR]\n\x04\x10\r?={}~CEIPTdegilpswx\x8B\x8C"; /* \x10 = ^p , \x04 = ^D */

Please use ascii order for these -- it's insanely hard to read otherwise.