Probability and Statistics, Are the developers checking them? Options

[mfb]

that doesn't fix the rules for regular parachuting. i mean, for chrissake, why do you default to body!?

[blakkie]

[Nikoli]

Jumping out of a perfectly good airplane doesn't seem like a good idea either.

[mfb]

depending on what you think skill 1 represents, people do it all the time. moreover, the skill of the guy pulling the ripcord doesn't have a lot of bearing on whether or not his parachute deploys, as the rules suggest. the rigger that packed the 'chute ought to have some dice in there somewhere.

[Req]

[mfb]

you must have used hand of god.

[Kagetenshi]

You find it amazing he survived. I say: Maybe he didn't!

~J

[Req]

[Jrayjoker]

Isn't it a little early for Shedim?

[Nythrun]

The examples of skill use listed in pretty much all of SR3 wend from statistically implausible to hellbound snowmen: flipping open Rigger 3 Revised randomly to page 83 yields Josie throwing eight dice at TN5 (six successes), TN7 (four successes), and TN8 (four successes) to reduce footprint. Now this isn't as egregious as the samples in, say, Magic in the Shadows, base TNs can start at 12 and where five dice routinely provide a handful of successes at TNs of 9 or more (and maybe "a few" task pool dice in the R3R example actually means "all three that you can get" rather than "two") - but this passage isn't convincingly indicative of statistical analysis. And I do realize that skill tests in examples need to be resolved exempli gratia, which is fine - but it's also reasonable to wonder why example tests demand fudging so very often, and surmise that maybe someone wasn't doing their homework.

Of course, the results of a rigger's Reducing Footprint test apparently have infinite duration and the test only has to be made once at chargen anyway, rendering the rules for it irrelevant and dysfunctional. That, however, is not a probablistic objection.

None of which, by the way, is intended to single out SR3 for mathematical scolding; D20 and White Wolf are no better and often worse. If, however, SR4 intends to sanity check the requirements for each category of things you can do, then I have to ask:

Really?
By Gencon?

Because it's a claim made in advertising often, and in practice never.

Let X equal the number of dice rolled
Let N equal the desired number of successes
Let C equal the number of possible combinations of N items in a set of X items

C{[4^(X-N)]/[6^X]} isn't a lot to solve.

[mfb]

yeah, i agree. the autofire example with Wedge was another questionable example, as i recall. i hope they use more realistic examples in SR4.

[blakkie]

Hehe, i always envisioned the die roller in those examples as squirming in pain as the horseshoe pinched his lower colon. :) Or they had a pile of Karma Pool available so they could reroll a lot, and that the rerolls simply weren't mentioned.

Note: In fairness for some of those examples the extra successes were good to have to show different aspects of the rules that one or two successes would not. They could have increased the number of dice rolled, but then we'd likely be going "WTF are they getting that many dice from?"

[mfb]

fudging the numbers a little bit is okay. but check out the ECM example in SR3: 10 dice, TN 8: 4 succs. 8 dice, TN 3: 2 succs. what the hell?

[blakkie]

Maybe the 10 dice were rolled behind the screen? They seem to mysteriously roll better back there sometimes. :vegm:

I hear and agree with what you are laying down though. It certainly would be a nice polish to have the examples constructed so that the majority aren't at the extremes of chance. Sorry i didn't make that clear.

[mfb]

personally, i wouldn't be adverse to showing examples that only use strictly average rolls. if you can't get your example to work at some minimal level with average rolls, well, maybe you should take a good, hard look at the mechanic you're trying to introduce.

[blakkie]

The problem with strictly average rolls is showing off some of the extremes of the system, such as when extra rules kick in past a certain number of successes. The Rule of 1's, a relatively rare condition, is also such a place to use rare rolls.

EDIT: But that's a good suggestion you make, that if a rare roll is needed to activate it you should reevaluate the rule and the underly mechanism that requires the rule.

Generally you also want to have as few matching numbers in the example as possible, unless there is something specific that needs the numbers the same.

An example:
For an opposed PC #1 rolling A dice gets B successes; PC #2 rolling X dice gets Y successes.

The numbers assigned to A, B, X, and Y should all be different numbers unless you want to show what happens when PC#1 and PC#2 roll the same number of successes (B and Y). This usually makes it easier to understand the example as matching numbers can become red herrings.

EDIT: Notice that the example you mention made that mistake. The TN of one character is the number of dice of the other character, but there is no meaning behind that match.

If you allow a bit of leeway it becomes easier to find a nice mix of numbers to fit that condition. A guideline could be "if a roll is lucky enough that it would enter your personal gaming story lore, good or bad, and the example can be clear without that particular roll, then change it".

[mfb]

right, exactly. average rolls should be used unless you're specifically trying to show what happens with extremely high or extremely low rolls. the non-matching numbers is a nice way to reduce initial confusion as well, simply because of the way people's memory works--if you see two 8s in an example, you're likely to later group them together and get things mixed up.

[Ellery]

I am tempted to mention that having a little variability around the average produces better examples yet. But that's the second step.

The first step should be to make examples with very close to average rolls. If they work, great! If they don't, fix the rule!

Then in some cases you might want to introduce a little more variability to better reflect that dice are variable.

[Nythrun]

"...characters should assense a spell and try to learn its force before attempting to dispel it. (SR3, 184)"

That's sound advice, no one wants to kill their PC trying to dispel. And in the best case scenario of a PC with Aura Reading 6 as well as Intelligence 6, the "prerequisite" assensing might garner necessary five successes as much as half of the time. For a more typical PC, however, those five successes will occur only one time in sixty-seven - which is not quite "seven impossible things before breakfast" but does mean that most characters are recommended to never attempt any dispelling.

One possibility is that the Third Edition intended dispelling to be a marginalized technique completely ignored in most games, which is the situation that obtains. I will take a leap of faith here and assume that there's consensus as to why dispelling is generally a poor stratagem; arguments thereto may be left undredged for concision's sake.

But another possibility is that assensing a spell's force is a little more difficult than the writer of that first quoted passage remembered, and perhaps four successes would have been better. Without a typical case example, it's not easy to say which possibility is more correct.

[Jrayjoker]

[Ellery]

[Charon]

[Ellery]

The extremes can matter, though. If will 6 is only 20% better than will 5, instead of twice as good, it kind of cuts down on the possibility of having someone in a group of mundanes who is really magic resistant. You'd have to implement that other ways.

It can be a big change to the way a game is run, especially if the PC has made his career around being the one with Will 6 who takes down the spellcasters.

[Charon]

[Ellery]