This hurts my head enough to think:
Isn't Fixed TN easier just for the not having to game probabilities?
Full Version: Solving 6=7
Tried making a test with less than 1/3 chance of success by an average character recently? Before you answer "threshold", consider what that does to the probability of success of people with low skill, as well as if it's really any easier to calculate.
(The short answer is that the fixed TN system discards the utterly trivial part of probability calculation, the part that we're all doing in our heads for this discussion, but keeps the bit where it gets messy and which we're all putting off until the last possible moment. I guess it is a fix for the 6=7 bug, but, well, when you've got the ability to write the rulebooks and the ability to get paid for it most of the difficult parts of that go away.)
~J
(The short answer is that the fixed TN system discards the utterly trivial part of probability calculation, the part that we're all doing in our heads for this discussion, but keeps the bit where it gets messy and which we're all putting off until the last possible moment. I guess it is a fix for the 6=7 bug, but, well, when you've got the ability to write the rulebooks and the ability to get paid for it most of the difficult parts of that go away.)
~J
You need statistics for both systems. Basically any discussion on this depends on the implementation of the system in SR.
A fixed TN/threshold system has a binominal distribution with p=0.33. That can be calculated ingame, but I would use the no-longer-needed BBB as an impromtu club on the offender. What works here is that you instantly know that less dice is bad and compare average successes with threshold. Easy to grasp != there is no math.
If you intend to improve the variable TN system, you *need* to make the whole range of skill ratings work. Variance of actual results decreases with an increasing number of dice, as it does with the equality of hit/miss chances. Being able to calculate that does little to improve low skill ratings usability. Three dice can all come up as a success at TN 12, but that deviation from the average is compensated by getting a miss more often than the average would suggest. More dice = closer to the average = better. The less dice the system is build for, the higher the chance of an absolute beginner to hit a TN that is rated ultra-hard.
A fixed TN/threshold system has a binominal distribution with p=0.33. That can be calculated ingame, but I would use the no-longer-needed BBB as an impromtu club on the offender. What works here is that you instantly know that less dice is bad and compare average successes with threshold. Easy to grasp != there is no math.
If you intend to improve the variable TN system, you *need* to make the whole range of skill ratings work. Variance of actual results decreases with an increasing number of dice, as it does with the equality of hit/miss chances. Being able to calculate that does little to improve low skill ratings usability. Three dice can all come up as a success at TN 12, but that deviation from the average is compensated by getting a miss more often than the average would suggest. More dice = closer to the average = better. The less dice the system is build for, the higher the chance of an absolute beginner to hit a TN that is rated ultra-hard.
| QUOTE (Penta) |
| Isn't Fixed TN easier just for the not having to game probabilities? |
you don't have to game probabilities, but if you want to, you can with any game that uses random chance in its mechanics. but, yes, one of the reasons i dislike fixed TNs is that there's less to game with.
| QUOTE (mfb) |
| you don't have to game probabilities, but if you want to, you can with any game that uses random chance in its mechanics. |
Only if the possibilities and probabilities are disclosed, and can be reduced to a set that can be usefully analyzed. Granted, we're stretching the definition of "game" in all the examples I can think of that deny nontrivial gaming.
We should come up with a different term there, as we've got name collision something fierce.
~J
haha, well, you can game those too, just not with any objective guarantee of success. eg, betting red when the wheel has spun black the last few times.
I'd draw a line between "gaming the system" and "applying magical thinking", myself, but we've reached the hair-splitting phase of this discussion I think 
~J
~J
| QUOTE (ShadowDragon8685) |
| Is it really that big a problem, unless you're the kind of person who worships probabilities and orgasms over statistics? |
Please..leave Perkins on the Battletech forums! And..GAH! Why do people work statistics and pointless minutiae into all game boards! Just shut up and roll the dice already!
Why do people who design game systems not have the foggiest clue how the game system is going to work in actual play because they have have not the vagues idea about the probability of success or failure? I mean 1st edition was totally nuts, as the examples in the book would happen about once in 500 tries, but there the examples of typical pedestrians who have a negligible chance to notice a crowd of terrorized citizens charging at them accompanied with machine gun fire in the background.
Unless I'm misreading the post, I suspect it's because attitudes like Cardul's are not uncommon, including among designers.
~J
~J
Most people writing for our game systems happen to be writers, not game designers.
What we do here when we are not playing is not what we do in game time. Some oddities are no problem, some can be quite game-breaking. We needed to fix 6=7 (and more important 6=0,5*5) because it encourages gaming for TN mods.
I´m all for avoiding complicated systems, but here we have some very easy solutions at our disposal. The argument Cardul makes here is common, but to me sounds like "stop thinking about rules when you play". That is only possible if the thinking was done beforehand.
What we do here when we are not playing is not what we do in game time. Some oddities are no problem, some can be quite game-breaking. We needed to fix 6=7 (and more important 6=0,5*5) because it encourages gaming for TN mods.
I´m all for avoiding complicated systems, but here we have some very easy solutions at our disposal. The argument Cardul makes here is common, but to me sounds like "stop thinking about rules when you play". That is only possible if the thinking was done beforehand.
| QUOTE (Cardul) | ||
Please..leave Perkins on the Battletech forums! And..GAH! Why do people work statistics and pointless minutiae into all game boards! Just shut up and roll the dice already! |
Because the statistics ARE the game. They ARE the world. They're the unified theory of physics for the character your world lives in.
I just came up with this in another thread, after I finished making an ass of myself.
One way to fix the 6=7 problem is, when rolling for TN 6, and ONLY when rolling for TN6, using exploding d4's against a TN6.
Probability of TN5 on exploding d6: 33.33%
Probability of TN6 on exploding d4: 18.75%
Probability of TN7 on exploding d6: 16.67%
As TN increases, probability of success decreases monotonically (until you hit the TN12=13 problem, which is a LOT less common of a problem)
It's not pretty, but it's pretty simple.
One way to fix the 6=7 problem is, when rolling for TN 6, and ONLY when rolling for TN6, using exploding d4's against a TN6.
Probability of TN5 on exploding d6: 33.33%
Probability of TN6 on exploding d4: 18.75%
Probability of TN7 on exploding d6: 16.67%
As TN increases, probability of success decreases monotonically (until you hit the TN12=13 problem, which is a LOT less common of a problem)
It's not pretty, but it's pretty simple.
A little thing I put together here about 2 years ago directly addresses this issue.
http://forums.dumpshock.com/index.php?showtopic=12340
It may be a little cumbersome to some people as it requires charts on hand for reference (one chart for all target numbers 1-20 is what we use), but on the other had there's no more buckets of D6's to deal with. Every roll is a percentage roll with 2D10's (with the occasional reroll to determine the decimal of the percent rolled - rarely needed)
http://forums.dumpshock.com/index.php?showtopic=12340
It may be a little cumbersome to some people as it requires charts on hand for reference (one chart for all target numbers 1-20 is what we use), but on the other had there's no more buckets of D6's to deal with. Every roll is a percentage roll with 2D10's (with the occasional reroll to determine the decimal of the percent rolled - rarely needed)
| QUOTE (Kagetenshi) |
| So you do find it workable in day-to-day play? I've always sorta looked askance at it. I'll have to slap together a chart when I get the chance. |
Check the related SR3R thread Kagetenshi; the rerol 5,6 add 4 is in the chart I already calculated. It was number 3 on my list of possible changes, mostly because as noted above there is a significantly higher probability of success at higher TNs. You get a probability curve much closer to the traditional d6 method--though still smoother--if you go with either of these:
1 Reroll 5s and 6s. Add 3 to any rerolled 5s, and 4 to any rerolled 6s.
2 Reroll 5s and 6s. Add 2 to any rerolled 5s, and 5 to any rerolled 6s.
...but I still like "Reroll 5s and 6s; add 4 to the re-rolled result" as it's simpler to describe.
Narse's solution in this thread is somewhat like "Reroll 6, add 5 to the result", but with the math complications you mention. It makes adding and subtracting TNs non-trivial, which is kind of a deal breaker right there, plus it makes tests with TNs greater than 6 much more difficult than the traditional method.
If you want a laugh, read Stumps's "solution" in the other thread. It basically "solves" the 6==7 problem by making 7==8, plus adding in more math headaches than Narse's solution.
Yeah, I remember it from there—though I guess I'd forgotten the chart (with good reason, apparently—no offense, but I nearly went blind looking at it again 
).
~J
).~J
Yeah, sorry about that. One of these days I'll work out this whole newfangled "pictures" thing and post a "chart", for all you people who don't live and breathe numbers.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.