# Idea to replace halving mechanic

GbajiTheDeceiver said:
I'm a "3 times better" man, so any mechanism that involves halving doesn't bug me as much as it might bug someone who supports the "x number of points better" view.

I'm a "3 times better" man myself. There's a scale issue, though. 25 -> 75 might be three times better, but it's on the "human" scale. On the other hand, 200 -> 600 is also three times better, but it's on the "HOLY CRAP THAT'S OFF THE GRAPH!" scale.

First I want to say it's been fun watchin all the math-fu flying around on these skill topics.

In reading all the posts the feeling I get while the skill mechanic hurts the PC sometimes in tests it also can benefit the players. The heros will be fighting many villians who will have better skills than they have and so the players will be helped by the odds giving a little edge to the less skilled.

So we're really trying to represent skill as a two dimensional function then (one dimension being skill A related to skill B, the other being a relation to the human norm) with a one dimensional mechanic.

:shock: No wonder things are getting weird!

For action based skills (where you perform some sort of physical action), I think halving works fine, IF it is optional to use, and IF it grants you another action at 1/2 the score.

But given that it is possible in MRQ (albeit unlikely) to have 400% in Athletics, you could potentially get a person that jumps several hedges and runs quite a distance in just one round (4 actions @ 100% rolls).

While the -100% method works great in Genesis, I do not think it will work as well in MRQ.

Thats why suggest an alternative.

Tempt Fate
1. If you have 100+% in a skill. You may use the Tempt Fate rule, which may give you a boon or ruin your result because you overdo it.

2. When invokint the Tempt Fate rule, you first roll at your unmodified %. If this is a success, you halv the % until you end up with a score less than 100%.

3. You roll against this halved result. If you succeed you increase the level of success by one step (a success becomes a critical, a critical becomes an amazing critical, or is just plain doubled in effect, whatever works).

4. Should you fail this second roll, you have overdone it, resulting in a bad result. Fate does not smile upon you. The level of success is worsened a level (a success becomes a failure, a critical a normal success).

This would require making a success level that is better than a normal critical. And while I have no more example on this, beyond combat (a critical becomes a damage x 2 critical), it could potentially be used for Crafting skills to create some amazing mastery craftsmen level object (something of superiour quality), or for such things as Hunting/Fishing etc. where it grants you twice the normal "gains" from the use of the skill.

GbajiTheDeceiver said:
I think one problem here is that people can view skill scores in two different ways.

Take an example. Rurik (not our Rurik, but the one from RQ2 ) has a skill of 75 whereas Cormac has a skill of 25.

Is Rurik 3 times better than Cormac, or is he 50 points better?

To exacerbate it, let's give them both some experience. Now Rurik has 600 and Cormac has 200.

This time, is Rurik still 3 times better than Cormac or is he actually 400 points better?

Depending on how they view it, each person will obviously come up with a different solution. I'm a "3 times better" man, so any mechanism that involves halving doesn't bug me as much as someone who might support the "x number of points better" view. And unless we find one that supports both, this thing will just go round and round and round.

I describe the two as Linear or Relational solutions.

Reducing by 100, or keeping the spread the same, or Bumping are Linear. For example a 360 to 320 skill gets reduced to a 60 to 20 skill. They are still 40 points apart, but note the higher skill is three times higher than (or 300% of) the lower skill, where before it was onlt 112.5% of the loer skill. This method works great with numbers under 200 or so, bit doesn't look so good in the above example.

Halving is an example of a relational. 360 vs 320 becomes 90 vs. 80, which seem a fairer representation than 60 vs 20. But it sure sucks with say 101 vs 50 getting reduced to 50 vs 25.

A linear solution works best if you never expect skills to approach 200, and a number have good solutions have been mentioned.

I think Mongoose went with halving (and Matt has said it was pretty controversial in the Mongoose offices) because they system will support very high skills. The Rules cover skills up to 500, and an example uses a skill of 350. This IMHO is why they went with halving, those kinds of skills need a relational reduction, and halving is the easiest way to do this.

Finding a Relational system that is 1) Easy and 2) Mathematically Satisfying is the hard part.

Older RQ fans will remember years of promises about Heroquesting, and these never came to be. We got HeroWars/Quest years later, but it was not RuneQuest at all. With Mongoose we are getting our HeroQuesting rules within the first couple months (when Legendary Heroes comes out in September). Yay Mongoose!

In light of that there is some defense for halving. If you don't like it don't use it, and If you don't plan on using High Skills in your games there have been a lot of good solutions that are easy.

I guess I'm not really making a point, just summarising some observations after spending a long time going over solutions on the boards.

Uh, sorry folks but it isn't the 3x or +50 thing that is screwing it up, is is that there is a relationship between the skill score and the D100.

Consider: A guy with 101% vs oone with 94%. Acoording toboth the ratio and the difference between the scores, the relative advantage is 7% (x1.07 or +7).

But, once you halve to 50% vs. 47%, both character went from almost certain chance of rolling a success to 50% or less.

atgxtg said:
Uh, sorry folks but it isn't the 3x or +50 thing that is screwing it up, is is that there is a relationship between the skill score and the D100.

Consider: A guy with 101% vs oone with 94%. Acoording toboth the ratio and the difference between the scores, the relative advantage is 7% (x1.07 or +7).

But, once you halve to 50% vs. 47%, both character went from almost certain chance of rolling a success to 50% or less.

Really, halving works fine if the characters have the same skill. 150 vs 150 and 75 vs 75 have the same odds. How much it hurt the higher skilled character skilled character is directly proportional to how much better than the lowe skilled character he is. 101 vs 95 halved to 50 vs 47 is not so bad odds wise. 101 vs 40 halved to 50 vs 20 totally screws the high skilled guy.

Rurik said:
atgxtg said:
Uh, sorry folks but it isn't the 3x or +50 thing that is screwing it up, is is that there is a relationship between the skill score and the D100.

Consider: A guy with 101% vs oone with 94%. Acoording toboth the ratio and the difference between the scores, the relative advantage is 7% (x1.07 or +7).

But, once you halve to 50% vs. 47%, both character went from almost certain chance of rolling a success to 50% or less.

Really, halving works fine if the characters have the same skill. 150 vs 150 and 75 vs 75 have the same odds. How much it hurt the higher skilled character skilled character is directly proportional to how much better than the lowe skilled character he is. 101 vs 95 halved to 50 vs 47 is not so bad odds wise. 101 vs 40 halved to 50 vs 20 totally screws the high skilled guy.
Whoever said a game mechanic had to be statistically precise anyway? Fudge a few percentiles either way, and it won't make much difference.

GbajiTheDeceiver said:
Rurik said:
101 vs 40 halved to 50 vs 20 totally screws the high skilled guy.

Whoever said a game mechanic had to be statistically precise anyway? Fudge a few percentiles either way, and it won't make much difference.

While not looking for precise mathematical integrity, in the above example A character with a 100 skill vs. a 40 skill has his an 87% chance of winning. If he raises his skill 1 point, it goes to 61%, which is a pretty severe hit.

Rurik said:
GbajiTheDeceiver said:
Rurik said:
101 vs 40 halved to 50 vs 20 totally screws the high skilled guy.

Whoever said a game mechanic had to be statistically precise anyway? Fudge a few percentiles either way, and it won't make much difference.

While not looking for precise mathematical integrity, in the above example A character with a 100 skill vs. a 40 skill has his an 87% chance of winning. If he raises his skill 1 point, it goes to 61%, which is a pretty severe hit.
Ah sorry, I was referring to halving when the values are similar.

I think halve plus add might be a reasonable solution... but what to add...?

GbajiThe Deceiver said:
I think halve plus add might be a reasonable solution... but what to add...?
I'd hoped so too, but in briefly looking at the maths I think that gets difficult as the penalty changes for each set of figs. afaik, it seems that at something like 120 vs 40 (halved to 60 vs 20) the bonus would be about 30% to have the same chance of beating the little guy, but at 140 vs 40, only 15 and at 160 vs 40 the bonus would only need to be 7 or 8...

It may be that allowing multiple attempts at %ge chances totalling his skill (horror? or not?) if the player wishes
... or ...
(as already mentioned, the old way of) a special/critical beating a non-special critical may be a pair of the simpler ways as characters with higher skills will have higher chances of specials & criticals. Takes away some simplicity, though.

Reading all the discussion about skills over 100%, I've been running Gloranthan games using the Elric rules, with RQ3 magic, since 1995. Here are the rules conevrsions from Elric to Glorantha: Elric RQ.

I've run 2 campaigns and numerous one-off sessions with my Elric RQ rules, including a Star Trek game and Traveller game using the same core mechanics. They're not a million miles from MRQ which is one thing that excited me about MRQ, but the main problem areas in MRQ are areas that the Elric game solved a decade ago.

For all skills, combat or not, the way you handle skills over 100% is the same. If your skil is over 100%, your chance of scoring a critical is equal to the ammount your skill exceeds 100. Thus a character with 140% skill criticals on a roll of 01-40, succeeds on 41-95, etc. This is very simple, requires only one roll and no maths whatsoever. It does also scale to skills over 200% without any problems. The actual level of succes is then the difference between the levels of success of the characters (so e.g. a critical roll versus a sucess roll is reduced to a plain success result, similarly for 'super-crit' versus crit).

I've seen one credible objection - that it exagerates the difference between characters with seemingly similar skill levels, e.g. 340% skill versus 380% effectively means the skill rolls are reduced to 40% versus 80%. In this system the important factor is the difference in skill levels, not the ratio. Once you understand that, the system is pretty consistent. Yes there are slight statistical uneveness around the transition points at the round hundreds, but they are relatively minor and in fact completely negligible compared to the huge swings in probability due to the current MRQ halving system.

It's a simple system that is very well playtested by now, and I know from personal experience works just fine.

Simon Hibbs

Halfbat said:
GbajiThe Deceiver said:
I think halve plus add might be a reasonable solution... but what to add...?
I'd hoped so too, but in briefly looking at the maths I think that gets difficult as the penalty changes for each set of figs. afaik, it seems that at something like 120 vs 40 (halved to 60 vs 20) the bonus would be about 30% to have the same chance of beating the little guy, but at 140 vs 40, only 15 and at 160 vs 40 the bonus would only need to be 7 or 8...

It may be that allowing multiple attempts at %ge chances totalling his skill (horror? or not?) if the player wishes
... or ...
(as already mentioned, the old way of) a special/critical beating a non-special critical may be a pair of the simpler ways as characters with higher skills will have higher chances of specials & criticals. Takes away some simplicity, though.
That pretty much settles that it's not a linear relationship with halving, then.

New proposal:
me said:
1. If highest exceeds lowest by 100 or more, highest has a straight 95% chance of winning.
This replicates the old Resistance Table rule, and fits into bluejays calculator also (100 vs 1 = 94.95% highest wins).

2. Otherwise, subtract the required multiple of 100 from both to bring them back to a 1 to 100 scale, then resolve.
This retains equal chances for equal skills, and also retains the relationship in difference between the skills.

Edit: (1) would mean that there's no difference between 120 and 666 when resolving against a 20% person, but then you're so far off the scale so far as that person is concerned, that "no difference" is probably an accurate reflection of things.

Rurik said:
Really, halving works fine if the characters have the same skill. 150 vs 150 and 75 vs 75 have the same odds. How much it hurt the higher skilled character skilled character is directly proportional to how much better than the lowe skilled character he is. 101 vs 95 halved to 50 vs 47 is not so bad odds wise. 101 vs 40 halved to 50 vs 20 totally screws the high skilled guy.

Yes it does work fine if it the valeus are the same. But that doesn't happen too often.

My popint was that by halving the skills you take what was too very high chances for success and cut them in half, turning more of the resolution into a coin toss. THe slight discrpancy in the 100s to the 90s is the "best case". As the spread gets wider the distorion get greater. Eventually it turn into-If High Skilled Character makes roll =win, if not =fail.

THe ratio or relative differnce is only the "apparent" chances of success.

FOr instance a 90 vs 30 id a 3x or +60% situation. Yet the "one character succeds other fails" situation is only:

21% vs 3% .

If we just put in criticals and use some form of bump it would work.

[Fairly Long]

I didn't realise there were so many problems with the Halving Rule/Lowest Wins resolution until I did some calculations.

The following were calculated with the following rules:
1. Two loops 1 to 100 to capture all possible dice roll combinations.
2. Better Level Wins (Critical beats Success, Success beats Failure)
3. On a level tie, either highest or lowest roll wins, depending on rules
4. On two Failures, they may tie, or use the highest/lowest roll wins, depending on the rules
5. It can take into account criticals, depending on the rules

The results are in the following format:

Player/Skill/Modified Skill/Wins/Percentage Wins
Resolution Rules - Which rules I am using
Win Order - Levels on the left beat levels further to the right

Resolution Rules:
Lowest Wins - on the same success level, the lowest roll wins
Highest Wins - on the same success level, the lowest roll wins
Restricted Levels - Only take into account Success/Failure
All levels - Take into account Criticals as well
Tie on 2 Failures - if both fail then the result is a tie
No Tie on Failures - if both fail then use highest/lowest roll for winner

So, looking at them:
1. Having a tie on 2 failures cuts down victories for the lower skill.
2. Using Lowest Roll Wins favours the lower skill by a fair amount
3. Including criticals doesn't really alter the results, except by the odd percentile

So, for it to balance properly, you probably need Highest Wins and Two Failures Tie, the combination of which means that considerably higher skills win considerably more often.

I hadn't realised it was broken so badly. So much for my gut feeling that it would work well.

(4 skill combinations, 8 rule combinations per skill combo means 32 results to look at, sorry about that).

Player Base Modified Wins %age
1 : 300% 75% 6190 62%
2 : 80% 20% 1790 18%
Resolution Rules:Lowest Wins,Restricted Levels,Tie on 2 Failures
Win Order (Leftmost Wins):Success,Failure

Player Base Modified Wins %age
1 : 300% 75% 6490 65%
2 : 80% 20% 3465 35%
Resolution Rules:Lowest Wins,Restricted Levels,No tie on Failures
Win Order (Leftmost Wins):Success,Failure

Player Base Modified Wins %age
1 : 300% 75% 6205 62%
2 : 80% 20% 1780 18%
Resolution Rules:Lowest Wins,All Levels,Tie on 2 Failures
Win Order (Leftmost Wins):Critical,Success,Failure

Player Base Modified Wins %age
1 : 300% 75% 6505 65%
2 : 80% 20% 3455 35%
Resolution Rules:Lowest Wins,All Levels,No tie on Failures
Win Order (Leftmost Wins):Critical,Success,Failure

Player Base Modified Wins %age
1 : 300% 75% 7290 73%
2 : 80% 20% 690 7%
Resolution Rules:Highest Wins,Restricted Levels,Tie on 2 Failures
Win Order (Leftmost Wins):Success,Failure

Player Base Modified Wins %age
1 : 300% 75% 8965 90%
2 : 80% 20% 990 10%
Resolution Rules:Highest Wins,Restricted Levels,No tie on Failures
Win Order (Leftmost Wins):Success,Failure

Player Base Modified Wins %age
1 : 300% 75% 7270 73%
2 : 80% 20% 715 7%
Resolution Rules:Highest Wins,All Levels,Tie on 2 Failures
Win Order (Leftmost Wins):Critical,Success,Failure

Player Base Modified Wins %age
1 : 300% 75% 8945 89%
2 : 80% 20% 1015 10%
Resolution Rules:Highest Wins,All Levels,No tie on Failures
Win Order (Leftmost Wins):Critical,Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 5280 53%
2 : 80% 40% 3180 32%
Resolution Rules:Lowest Wins,Restricted Levels,Tie on 2 Failures
Win Order (Leftmost Wins):Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 5580 56%
2 : 80% 40% 4355 44%
Resolution Rules:Lowest Wins,Restricted Levels,No tie on Failures
Win Order (Leftmost Wins):Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 5286 53%
2 : 80% 40% 3177 32%
Resolution Rules:Lowest Wins,All Levels,Tie on 2 Failures
Win Order (Leftmost Wins):Critical,Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 5586 56%
2 : 80% 40% 4352 44%
Resolution Rules:Lowest Wins,All Levels,No tie on Failures
Win Order (Leftmost Wins):Critical,Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 6680 67%
2 : 80% 40% 1780 18%
Resolution Rules:Highest Wins,Restricted Levels,Tie on 2 Failures
Win Order (Leftmost Wins):Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 7855 79%
2 : 80% 40% 2080 21%
Resolution Rules:Highest Wins,Restricted Levels,No tie on Failures
Win Order (Leftmost Wins):Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 6657 67%
2 : 80% 40% 1806 18%
Resolution Rules:Highest Wins,All Levels,Tie on 2 Failures
Win Order (Leftmost Wins):Critical,Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 7832 78%
2 : 80% 40% 2106 21%
Resolution Rules:Highest Wins,All Levels,No tie on Failures
Win Order (Leftmost Wins):Critical,Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 7135 71%
2 : 10% 5% 485 5%
Resolution Rules:Lowest Wins,Restricted Levels,Tie on 2 Failures
Win Order (Leftmost Wins):Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 7435 74%
2 : 10% 5% 2535 25%
Resolution Rules:Lowest Wins,Restricted Levels,No tie on Failures
Win Order (Leftmost Wins):Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 7153 72%
2 : 10% 5% 471 5%
Resolution Rules:Lowest Wins,All Levels,Tie on 2 Failures
Win Order (Leftmost Wins):Critical,Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 7453 75%
2 : 10% 5% 2521 25%
Resolution Rules:Lowest Wins,All Levels,No tie on Failures
Win Order (Leftmost Wins):Critical,Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 7485 75%
2 : 10% 5% 135 1%
Resolution Rules:Highest Wins,Restricted Levels,Tie on 2 Failures
Win Order (Leftmost Wins):Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 9535 95%
2 : 10% 5% 435 4%
Resolution Rules:Highest Wins,Restricted Levels,No tie on Failures
Win Order (Leftmost Wins):Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 7431 74%
2 : 10% 5% 193 2%
Resolution Rules:Highest Wins,All Levels,Tie on 2 Failures
Win Order (Leftmost Wins):Critical,Success,Failure

Player Base Modified Wins %age
1 : 150% 75% 9481 95%
2 : 10% 5% 493 5%
Resolution Rules:Highest Wins,All Levels,No tie on Failures
Win Order (Leftmost Wins):Critical,Success,Failure

A very thoughtful post soltakss.

I noted the following.

soltakss said:
Resolution Rules:
Lowest Wins - on the same success level, the lowest roll wins
Highest Wins - on the same success level, the lowest roll wins

I was working on a similar results and would like to post the following formula for calculating chances of winning an opposed roll, in case anyone wants to use them)

A=Higher skill
B=Lower Skill
C=100-A (for the range above the attackers skill)

The formula for A's % to win according to the rules as is is:
A + ((C2 - C) - ((B2 + B))/200)

The Chance of a Tie is (B+C)/100

To simulate the effect of changing High Roll Wins on a Fail/Fail, add to A's chances the following:

((A - B) X C)/100.

Don't know if anyone wants them, but they are what I have been using.

EDIT:
These results jive with BlueJays calculator, but do NOT take a 95% cap into account, so if you enter a result above 95 it will not jive. Simply reduce A to 95 (or whatever the cap should be after halving for very high skills).

You can use the calculator to generate the chance of success and add to it the ((A - B) X C)/100 to simulate High Roll Wins).

In light of all the above math I think I am going to use the following:

1) Tied rolls go to the higher skilled character (this will give at most a 0.99% advantage, but since he is the guy getting screwed lets give him every bit). You will only need to re-roll ties if the skills are the same.

2) For normal opposed tests, use the rule as is - low roll wins on fail/fail results. (probably with no crits, no automatic failures).

3) For Halved tests switch to High Roll Wins on Fail Fail results. This gives a bonus to the higher skilled character who has been halved. In the extreme cases 101 vs 20, he still suffers, but not as bad.

So a 75 vs 40 normal test results in a 70.45% chance of success (I added the tie in as per point 1).

A 75 vs 40 that is the result of halving from 150 vs. 80 (or 300 vs. 160)results in a 79.2% chance of winning.

Not perfect, but it is in the spirit of the original rules, simple, and requires no extra rolling (less actually because of ties are no longer re-rolled).

EDIT: As soltkass points out using criticals doesn't really change the odds, and IMHO makes the mechanic much more awkward, as you have a roll under skill usually being desireable, and you have to use odds before halving, etc.

I don't use the automatic failure because it doesn't change the odds that much, and only hurts the higher skilled character. I'm not even positive it was intended to be used with opposed rolls in the first place.

Rurik said:
A 75 vs 40 that is the result of halving from 150 vs. 80 (or 300 vs. 160)results in a 79.2% chance of winning.

What are the probabilities for characters with 100% skill and 101% skill versus the same opponent (e.g. 40%)? That's the real test of the system.

Characters with 150% or more are the most likely to give sensible results after halving and so don't tell us much about the system.

Simon Hibbs

simonh said:
Rurik said:
A 75 vs 40 that is the result of halving from 150 vs. 80 (or 300 vs. 160)results in a 79.2% chance of winning.

What are the probabilities for characters with 100% skill and 101% skill versus the same opponent (e.g. 40%)? That's the real test of the system.

Characters with 150% or more are the most likely to give sensible results after halving and so don't tell us much about the system.

Simon Hibbs

100 vs 40 tie to defender:

A = 92.4 (not using the 96-100 failure, with it it becomes 87.35

101 vs. 40 becomes 50 vs. 20 (note BlueJays Calc reduces this to 51 vs 20 - but the book says round down in all cases).

By the book: 60.15
My Method: 76.15 Not perfect, but a lot better. This is the most extreme case (actually 101 vs 10 is more extreme).

Ah, I hadn't seen that we were supposed to always round down. I'll amend my calculator accordingly!

Done!

Replies
19
Views
558
Replies
22
Views
510
Replies
10
Views
604
Replies
10
Views
723
Replies
2
Views
664