Statistical Anomolies with Timing/Effect

[SableWyvern]

tbeard1999 has raised some interesting and potentially very important points about the statistical realities of the Timing/Effect system over on the CoTI forums.

http://www.travellerrpg.com/CotI/Discuss/showthread.php?t=14478&page=9

I haven't done any close analysis of the information he presents myself, but I think it's definitely something that deserves a closer look.

 

[AKAramis]

Ty's analysis ignores skill. Making it pretty much worthless.

Further, Ty's openly hostile to Mongoose's Traveller, and seems to be looking for reasons to hate it. Read page 1 of that thread. His open hostility to unified mechanics is fairly strong. (and shows in other threads as well.)

Be wary of the COTI participants; we each have long standing threads of likes and dislikes, and many are openly hostile to anything more than a clean-up of their favorite system.

I suggested last night a modification to results, tho', having noticed a strong tendency to have effect in the range of 5-7 pretty consistently.

Res  Time  Effect      Cont/Coop Failure      Cont/Coop  
 1   x10   Marginal       +1     Superlative     -5
 2    x9   Marginal       +1     Superlative     -5
 3    x8   Poor           +2     Exceptional     -4
 4    x7   Poor           +2     Exceptional     -4
 5    x6   Average        +3     Average         -3
 6    x5   Average        +3     Average         -3
 7    x4   Exceptional    +4     Poor            -2
 8    x3   Exceptional    +4     Poor            -2
 9    x2   Superlative    +5     Marginal        -1
10+   x1   Superlative    +5     Marginal        -1

He's dead wrong about never having a 1 show up, due to skill and/or difficulty effects. However, since the mods add, he's right that a 1 is quite unlikely under draft 2.

Now, if stats are returned to the draft 1 "don't add to effect die", then we again get 1's showing up.


That being said, he's right in that the success levels are too easily hitting superlative. The fix for this is what I've suggested: raise the bar.

 

[captainjack23]

...please, guys. Keep it at home, okay ? You guys squabbele enough over there. :!:

That said, good statistical points.

cap

 

[Halfbat]

Though the statistics are probably correct, within the limited analysis given, we tried it without mods and, yes, players regularly chose the higher effect in combat. This was due to the "all END reduced in one go = Unconsciousness" rule.

However, this was most prevalent amongst those characters with -2, +1 and +2 (all having +1DEX) total to their combat. One character, with +1DEX +4SKILL, had a regular thoughtfull call as to whether or not to using any "6" he rolled as a timing dice to reduce the effects of recoil, as his weapon was causing enough damage to level average mooks and he was hitting reasonably frequently...

Though I'm generally supportive of statistical analyses, I think we have to be careful of limited analyses and jumping to conclusions, and make a full, broader-spectrum analysis of effects in play.

 

[tbeard1999]

Ty's analysis ignores skill. Making it pretty much worthless.

In what way?

Further, Ty's openly hostile to Mongoose's Traveller, and seems to be looking for reasons to hate it. Read page 1 of that thread. His open hostility to unified mechanics is fairly strong. (and shows in other threads as well.)

What does dislike have to do with the accuracy of a factual assertion?

I suggested last night a modification to results, tho', having noticed a strong tendency to have effect in the range of 5-7 pretty consistently.

So...my analysis is "worthless", but you noticed similar problems to what I noticed.

He's dead wrong about never having a 1 show up, due to skill and/or difficulty effects.

Unfortunately, I said nothing of the sort. I did state that it is impossible for a 1 to show up in any 2d6 roll of 8+. Feel free to show us how that statement is incorrect.

That being said, he's right in that the success levels are too easily hitting superlative. The fix for this is what I've suggested: raise the bar.

Or ditch the system.

 

[captainjack23]

Ty's analysis ignores skill. Making it pretty much worthless.

In what way?

Further, Ty's openly hostile to Mongoose's Traveller, and seems to be looking for reasons to hate it. Read page 1 of that thread. His open hostility to unified mechanics is fairly strong. (and shows in other threads as well.)

What does dislike have to do with the accuracy of a factual assertion?

I suggested last night a modification to results, tho', having noticed a strong tendency to have effect in the range of 5-7 pretty consistently.

So...my analysis is "worthless", but you noticed similar problems to what I noticed.

He's dead wrong about never having a 1 show up, due to skill and/or difficulty effects.

Unfortunately, I said nothing of the sort. I did state that it is impossible for a 1 to show up in any 2d6 roll of 8+. Feel free to show us how that statement is incorrect.

That being said, he's right in that the success levels are too easily hitting superlative. The fix for this is what I've suggested: raise the bar.

The situation that Tbeard analyzed does occur in play, if not frequently. A skill 0 attempt, with no mods cannot succeed when one of the dice shows a 1.
However, it may be worth noting that version 3.1 reels in the play effects of a six or a one - and in fact makes them equiv. to a five or a two respectively.

The numbers that show are a bit wonky, but now that the effect values are no longer really special (good or bad) it doesn't matter as much -IMHO. A level 0 attempt can succeed with poor effect - although not as likely as with good effect; which is probably an arguably realistic approach to modeling task completion rates. (if your results are crappy as often as it is good, you probably don't qualify for the skill level) plus, from a game perspective it's important generally that players succeed.

As to Tbeards like or dislike of the system, well, if his gripes are quantifiable, who cares ? Sure, he finds what he looks for, but this is a playtest.

Cap

 

[tbeard1999]

As to Tbeards like or dislike of the system, well, if his gripes are quantifiable, who cares ? Sure, he finds what he looks for, but this is a playtest.

Cap

Agreed. I would add that I want Mongoose Traveller to be good. That said, I have been open about the fact that I think the effect/timing mechanic is broken and unfixable. I base this on my own inability to create a fix (yes, I tried). Of course, I may have overlooked something. But I doubt it.

And if I am right, then the game has a serious flaw in its fundamental mechanic. The result in that case is obvious -- yet another crappy Traveller redo. And don't we have plenty of those already?

And as you note, the flaws are either there or they aren't. My dislike of the system is irrelevant.

Nor is my dislike of the initiative die system relevant to this discussion. I detest the initiative die system on its own merits (or lack thereof) and I've been very specific in my reasons. It's really kinda lame for Aramis to dismiss these very specific objections on the grounds that I "just don't like the game".

 

[tbeard1999]

Here's my current take on the problems with the timing/effect system. If I am wrong about my assumptions, I'll gladly revise my assessment.

1. The Problem

Here are the possible 2d6 combinations for rolls of 8+:

2 6
3 5
4 4
5 3
6 2
3 6
4 5
5 4
6 3
4 6
5 5
6 4
5 6
6 5
6 6

Total number of rolls: 15

For reference, a single die roll of 5+ is an exceptional success and a 2- is a marginal success.

From the list above, here’s the distribution of numbers that will appear on each die:

1 -- None
2 -- 1 (7%)
3 -- 2 (13%)
4 -- 3 (20%)
5 -- 4 (26%)
6 -- 5 (33%)

So, each die has a 33% of coming up 6, a 26% chance of coming up 5 and only a 7% chance of com-ing up 2. No chance of a 1 coming up.

So...there’s a 7% chance of a marginal success (2 or less) but there is a 59% chance of an exceptional success (5 or 6).

What this means is that on a normal, unmodified roll, an Exceptional Success will occur 8 times as often as a marginal success.

Indeed, there’s a greater chance of it succeeding Exceptionally (59%) than of it succeeding in an Average fashion (33%).

EDIT: See the next post for the details.

 

[tbeard1999]

I figured out a better way to churn the numbers, and here are the results:

Here are the odds of rolling a success and rolling the indicated result on each die (percentages rounded to nearest 1%):

No Modifier
2-: 7%
3-4: 33%
5+: 60%

+1 Modifier
2-: 5%
3-4: 24%
5+: 71%

+2 Modifier
2-: 0%
3-4: 20%
5+: 81%

-1 Modifier
2-: 10%
3-4: 50%
5+: 40%

-2 Modifier
2-: 17%
3-4: 83%
5+: 0%

So...if I make an unmodified roll, and it succeeds, each die has a 60% chance of being a 5 or higher. The chance of both dice being 5+ is 36%. The chance of neither dice being 5+ is 16%.

Note the tremendous difference between a -1 and a -2. A roll at -1 has a 40% chance of being an exceptional success. A roll at -2 has NO chance of being an exceptional success.

In fact, the exceptional success is the most likely possibility for any roll with a modifier of 0 or more and a 40% at -1. But at -2, an exceptional success is impossible.

And here are the chances of rolling at least one 5+:

No modifier: 84%
+1: 91%
+2: 96%
-1: 64%
-2: 0%

Does this look like a good distribution of probabilities for an RPG?

 

[captainjack23]

Its a problem, granted. I'll need to think a bit if its a serious problem given that its an optional rule in all but combat. It seems to work in play, despite the above effects, but I'm not sure why, which is a concern for me at least.

Part of the problem may be that the above focuses on success - if one includes success and fail results with quantification (ie a 1 in a fail, a 1 in a success) it may look better ?

Functionally, I think that it works in combat given that there is a very real reason to allocate a good effect to timing, even with a fail overall.

Cap

 

[SableWyvern]

I started a new thread on a proposed fix for tbeard's problems, which has the added benefit of streamlining the system somewhat.

http://www.mongoosepublishing.com/phpBB2/viewtopic.php?p=469205#469205

 

[tbeard1999]

I started a new thread on a proposed fix for tbeard's problems, which has the added benefit of streamlining the system somewhat.

http://www.mongoosepublishing.com/phpBB2/viewtopic.php?p=469205#469205

But as noted in that thread, the fix doesn't really solve the problems. It does mitigate them somewhat.

The real problem with the timing/effect system -- besides my subjective judgment that it is too fussy considering the minimal benefits it would yield if it actually worked -- is that the designer is simply asking one 2d6 roll to do too much. There's a reason most games use several discrete rolls to accomplish things like hitting and wounding a target -- the probability curves of the discrete rolls are regular and mechanics to suit them can be straightforward and uncomplicated.

With systems like the timing/effect system, the fixes necessary to force the system to yield reasonable results (if that's even possible) tend to be so fussy that they waste more energy than a simpler 2 roll system (or a simple "the amount yuou exceed the task roll determines your effect" system).

 

[EDG]

The real problem with the timing/effect system is that the designer is simply asking one 2d6 roll to do too much.

Sounds like you've never tried the ORE system (Reign, Godlike, Wild Talents) . In that, a single d10 dicepool roll determines order of initiative, hit roll, damage done, and hit location all at once... and it works pretty well.

But that's not this system . Just pointing out that it can be done though.

 

[captainjack23]

Tough issue, (I actually woke up in the night thinking about it) but I am finally leaning towards the system as it is, statistics notwithstanding (and believe me, it pains me to say that) . Here is why:

First, and foremost, I'd say that the optional and situation limited nature of the rule AND the fact that most skill rolls are (or should be) binary anyway (quantification is irrelevant) means that for non combat resolution , it doesn't effect play or degree of success or failure much at all, ever.


As regard skill resolution:

1. The T/E choices will only be needed where time is crucial, or quality iscrucial: when one dominates the other, it makes sense that the character will either fail, or succeed accordingly - ie generally with better time OR better quality of effort. While you will NEVER succeed with a 1, you will still succeed or fail: a low/high or high/high roll will be the players goal; if the latter, all is good; if the former one is sacrificed for the other.

2. The fact that that sacrifice is less likely to be a massive failure is noted....but.....the game is about doing things so I don't have a big problem with that bias....; and the fail results (less than eight ) will also reflect the choice of time or effect, which seems to be being overlooked here.


3. It strikes me that the analysis needs to consider the distribution of failures, , the situation when a T/E read will be called for AND the nature of most resolutions.a 5/2 will still have the effect of the 5 being the desired effect or time, if only to allow a reroll. The fact that the fail results mirror the success rolls, (ie the 1's pile up) is simply what you get when performing under stress in cinematic conditions...great success or great failure. (although, I'd be prepared to argue that the actual observation of motivational effects do show this pattern -).

4. Really, part of the problem seems to be the words used in the quantification, and the desire for symmetry in the results. If they were simply"adequate, acceptable, good", I'd have a lot less trouble arguing that at most tasks one tries (at a reasonable level of challenge - ie appropriate to your training, a skill level ~= the difficulty) the high end for MOST results is reasonable. 50% near failures is not a reasonable description of most competently attempted tasks, ask your supervisor/advisor/boss/spouse

5. Finally, really, the key is to not assume that this is necessary OR needed for many rolls at all - only those where there is a significant trade off issue - ie great stress.

I have no problem with the skewed results in combat.

1. Both sides get it, so there the skew is equal (player and GM characters get it, NOT just the players);

2.it can be posited that the low effect/timing results represent non-crucial or irrelevant successes ("Ow ! creased my helmet !"). Ignoring 1HP scratches is a great feature of the system, esp if you've ever played one of the classic eighties multi table shooter RPGs.

3. The initiative/choice effect in combat is very important, and will similarly balance the results, as a failure still has consequences that must be chosen from. Even a miss need not be a disaster if ones timing is not badly reset.

4. Higher effect and init results make combat nastier (My preference) and keep it moving along.

Cap

 

[tbeard1999]

The real problem with the timing/effect system is that the designer is simply asking one 2d6 roll to do too much.

Sounds like you've never tried the ORE system (Reign, Godlike, Wild Talents) . In that, a single d10 dicepool roll determines order of initiative, hit roll, damage done, and hit location all at once... and it works pretty well.

But that's not this system . Just pointing out that it can be done though.

Well, I never said it couldn't be done with a d10 dice pool system. What I am saying is that it can't be done (well) with a 2d6 additive system the way Mongoose is trying to do it.

 

[EDG]

Well, I never said it couldn't be done with a d10 dice pool system. What I am saying is that it can't be done (well) with a 2d6 additive system the way Mongoose is trying to do it.

I think it's interesting to try to add extra dimensions to a simple d6 roll though (some systems do it with different coloured dice, e.g. IIRC WEG Star Wars).

I'm not sure this timing/effect thing works here myself though, from what I understand of it the timing/effect part seems too random when you consider that you're also trying to beat a target number with the summed total of the dice.

 

[tbeard1999]

Well, I never said it couldn't be done with a d10 dice pool system. What I am saying is that it can't be done (well) with a 2d6 additive system the way Mongoose is trying to do it.

I think it's interesting to try to add extra dimensions to a simple d6 roll though (some systems do it with different coloured dice, e.g. IIRC WEG Star Wars).

I'm not sure this timing/effect thing works here myself though, from what I understand of it the timing/effect part seems too random when you consider that you're also trying to beat a target number with the summed total of the dice.

If that is a problem, then it can be dealt with by moderating the actual effect of the timing/effect result. Of course, at some point, it becomes not worth the effort expended.

But I think that the real problem lies in trying to use a 2d6 target number roll, then trying to use the individual dice that comprise a successful roll. In doing so, you effectively lose most of the lower rolls (no ones can appear on a 2d6 roll of 8+; a two can appear in only 2 out of 15 combinations) and keep most of the higher rolls (a six can appear in 8 of 15 combinations). The result is a mechanic that seriously skews high and that yields dubious results when modifiers are applied.

It appears that the designer made the assumption that a one had the same chance of occuring as a six on the timing/effect mechanic. 'Taint so, I'm afraid.

I don't mind "adding extra dimensions" to any roll, so long as the mechanic to do so does not take more effort than it's worth.