[naloth]

Originally Posted by Thamior

This is why we have critical failure. Or else I suggested a margin of failure before. Reread my earlier posts.

You've moved the goal posts a bit but you have exactly the same problem that I've mentioned before. You've either made the answers so reliable that you can count on them (90% in your example) or you leave them unreliable enough that you can only count on your normal percent chance of success.

If you have a 50% chance of success now, you'll be right 50% and you could be misled 50%. Yes, that's a cointoss. The first alternative you've suggested means that 50% you'd be right, 49.5% you'd know you failed, and .5% you'd get a misleading answer. That means 99.5% of the time you get a useful answer. Changing it such that you only fail on a 15+ (11-14 give "I don't know") changes it to 50% certain, 40% certain you failed, 10% mislead.

If you're getting an impression of a dozen potential murderers, you can almost certainly eliminate half with either ability. Two characters could eliminate most suspects instantly (without even questioning them!) and detect lies over 75% of the time regardless of how good the lie is. Three characters would be better than Wonder Woman's lasso.

1) 20% success 2) 70% you don't know for sure 3) 10% failure

Bellcurves don't make this a very likely progression, especially when +1 IQ either way will drastically improve or drastically reduce these numbers.

For example, if 8 or less is ~16%, 8-14 would be ~68%, 15+ 16%, then a simple +1 would change that to: 9- ~26%, 10-15 ~69%, 16+ ~5% which already reduces your misleading chances to very low. A +3 (basically where Sensitive -> full Empathy) gives you a 50% chance of success with critical failures less than half a percent of the time.

Ok, example with guns (attacks). Why it differs. I'll explain.
1.) You have a clear way of improving attack skill with 2 options: buying stat and buying skill levels, These options are pretty balanced. If you say something about reliant enhancement for Sensitive: 5% for +1 to IQ for this ability. You get +10 for only 50% the cost, or for just 3 points. Lol.
2.) Attack is inherently repeatable (B348).
3.) Attack has a clear indicator of success and failure (in almost all cases).
4.) Most of the time you get no penalty for miss except the missed opportunity to hit with an attack (yeah and you spend bullet, but for the love of god don't blow it out of proportion, it's very minor expenditure). It's almost the same as Do Nothing. You're not much worse than that for having missed. Compare to GM BSing you.
5.) Clear and more varied mechanisms for improving the odds of success.

1) Sure, reliable is pretty cheap. Perhaps it should be restricted to talent or further modified by other factors.

2) Most things are, not just attacks. That's why Campaigns pretty well outlines that unless you're destroying the subject, you can try again. If you build something wrong, you can build it again. If you draw something wrong, you can draw it again. If you do a bad job making supper, you can make something else. If you clean something poorly, you can keep cleaning.

3) I actually agree here -> there should not be a clear indicator of success. That's why I likened this to a sense roll. Failure shouldn't give you warning that you need to dig deeper for an accurate answer.

4) Again, sure. This is an information roll. If you know the information isn't reliable it gives you a cue to get it a different way, have someone else try, re-use the skill, etc. Making it this reliable effectively lets you eliminate a lot of stuff and zero in on what you need to be checking out. It's not supposed to work that way.

5) I don't see what's unclear about using difficulty modifiers if you choose to use them.