d20 vs. 3d6 in HeroQuest?

[buserian]

In the recent Suggest a game system for me PLEASE! topic on the RPG Theory forum, Nuadha said the following:

The problem with using Heroquest, is that it only uses one die. It sounds like you want a system with more of a dice curve, where skill ratings are more important than lucky dice rolls.

I would suggest trying GURPS, but I personally don't like combat in GURPS. However, GURPS combat can be pretty simple. It uses 3d6, so it has a strong dice curve.

Hero System also uses 3d6 but it tends to be more "crunchy."

I second the suggestion that you try creating your own system or adapting a system you like. For example, if you like everything else about Heroquest, I'm sure you can adapt it to use 2d6 or some other dice pool.

This made me think. Has anyone tried using a modified HQ system that uses something like 3d6 or 2d10? You'd have to make adjustments, of course (3d6 -- 3 = critical, 18 = fumble; 2d10 -- 2 = critical, 20 = fumble), and it could screw up some of the mastery counting. (I suppose [1d6+2d8-2] would be somewhat ideal, since it would generate a range of 1-20 and thus would not involve any adjustments to the math or lookups.)

But it might generate an interesting dynamic of play. Criticals and fumbles would be much less dependent on luck, and much more related to mastery advantage and spending of hero points.

It seems like it could be an interesting experiment.

buserian

[Mike Holmes]

I'm pretty sure that I'd miss the criticals and fumbles a lot. Even on the 2D10, they're five times more rare. Part of the charm of the system is how often these results occur, IMO.

Note, too, that the effect would be more pronounced than you think. Rolling 2d10 against 2d10 would result in differences of far less than a whole mastery being very pronounced. 5 points would be about as good as a mastery is now in most cases.

Much worse with 3d6 where the range is even narrower, too. And only one in 54 rolls will produce any sort of crit or fumble. Complete Success for mathed opponents would only occur naturally once in 46,000 attempts (same for Complete Failure).

Also, as I noted in that thread, since you roll against each other in HQ, the result is already a curve (OK, a pyramid, but who cares). So you already do have this effect to an extent. Going with more dice would just push this curve to a very unfun place. Basically, HP would become a currency for buying success, instead of adjusting the level of success.

I don't usually worry about realism, but in this case you're throwing out both realism and drama, I think.

Mike

[Mac Logo]

I'm with Mike on this. Everything about the HQ system screams out loud "LINEAR". Probability curves favour the average, since when did the concept of a Hero ever do that?

Question: Are there any systems that favour extreme dice rolls. Obviously it would have to be very "Nar" driven to be any sort of fun, but do such RPGs exist?

Cheers

Graeme

[RaconteurX]

I have heard HeroQuest variants using 1d12 and 1d10 suggested (each of which leads progressively to greater cinematic play, and even faster hero development), but none that use multiple dice of any sort. I can understand why one might introduce such a change. If one wanted a grittier game, it would be an easy way to do it... though you would need to alter Hero Points even more, as they would have much greater impact on die rolls. Perhaps a rolled critical should merit awarding a Hero Point, much as in James Bond 007. Advancement would have to be made separate from Hero Points, else the players will tend to hoard them as a means to achieve those increasingly rare critical rolls at vital junctures in the story. Alas, I think switching to 3d6 or 2d10 would require more than a little retooling to the entire currency mechanism... establishing a proper balance would no doubt be one of the first orders of business if one had such a change in mind.

[Nicolas Crost]

buserian,

I understand you problem. I myself like curved distributons much better than liner ones (might be my GURPS background coming thorugh).

What I am going to try in our Planescape campaign is this:
Just let the player roll 2d10 and the GM nothing at all. Since most people suggest rolling the GM die openly (no fudging), you might as well let the player roll it.
The dice are added with the effective skill of the character. The skill of the npc / obstacle ist subtracted. The you look up the result in a table that looks somewhat like this:
smaller than 0 : total failure
smaller than 10: marginal failure
10 :tie
bigger than 10: marginal success
bigger than 20: total success

This changes the probabilities of the game, I know, but it might be acceptable. It will take a bit of playtesting to figure that out.

Nicolas

[simon_hibbs]

This seems realy screwy to me.

Suppose I've got an ability rating of 10 and we're rolling 2D10s. Each HP I spend increasing my ability from here generates a reduced payback untill I hit an ability rating of 20. After that, each HP I spend generates an increased payback untill I reach a rating of 10w.

I don't see how the payback on ability ratings see-sawing back and forth as you go up the ability scale is so desirable.


Simon Hibbs

[Valamir]

The linear nature of HQ threw me at first too.  In general, I hate linear curves, at least ones where the random range is sizeable compared to the fixed modifier.  Two things really mitigate this for me in HQ.

1) the fixed score of the character's best abilities are quite substantial relative to the random outcome.  A 2 mastery character is still going to give you a 2 mastery result.  The randomness only effects whether you get an additional 3rd mastery result.

This isn't always obvious because the system has you cancel masteries first and then just use them to "bump" success levels.  But go ahead and roll a few times without cancelling masteries.  2W3 may FEEL like your only rolling against a 3 on 1d20, but in reality, you're rolling d20+43, with every increment of 20 being a success level.

Since the random range is so much less than the fixed modifier, the whiff factor inherent to linear curves like d20 or BRP is almost non existant (for your best skills anyway).


2) HQ is really a die pool system in disguise.  D20 combat is also really a die pool system in disguise.  You roll so many to-hits on single d20s that over the course of a battle you get a reasonable normal distribution of expected results.  Where d20 breaks down is when it shifts to non combat where the entire task is handled with a single roll, and you don't get this faux pool effect.  This is why Take 10 and Take 20 were invented.  Basically to patch the weakness of using single rolls for everything other than combat as a way of increasing the likely hood of getting the expected result.

HQ solves this problem by using the same degree of resolution for everything.  What you have to keep in mind is Augment Augment Augment.  When you make 5 or 6 or 12 Augment rolls leading into your main roll you're accomplishing 2 things.  First the number of rolls is basically equivelent to rolling a die pool (even more so if all of your abilities that you're augmenting from are the same level)...you're just doing it 1 die at a time.  And second you're boosting the fixed modifier on your final roll by a substantial amount, which enhances the effect of #1 above.

So ultimately.  Hero Quest really doesn't suffer from the usual foibles of a linear curve resolution.  It probably has the best distributed results curve of any single die system I've seen.

[Nicolas Crost]

This seems realy screwy to me.

Suppose I've got an ability rating of 10 and we're rolling 2D10s. Each HP I spend increasing my ability from here generates a reduced payback untill I hit an ability rating of 20. After that, each HP I spend generates an increased payback untill I reach a rating of 10w.

I´m sorry. I forgot to mention that I got rid of masteries. They are a strange (interesting, but still) way of being able to keep a "roll under a target value"-system and having one range of ability score.
So what I basically did was to change to a "roll,then add skill: must be over target value"-system.

What you get is something like this:

2W3 may FEEL like your only rolling against a 3 on 1d20, but in reality, you're rolling d20+43, with every increment of 20 being a success level.

Same thing, different feel. You roll 2d10, add the skills (with masteries being +20 each). Then every increment of 10 is an extra success-level (ok, the shift from 20 to 10 is intentional).

2) HQ is really a die pool system in disguise.  
[...]
Where d20 breaks down is when it shifts to non combat where the entire task is handled with a single roll, and you don't get this faux pool effect.  This is why Take 10 and Take 20 were invented.  Basically to patch the weakness of using single rolls for everything other than combat as a way of increasing the likely hood of getting the expected result.

HQ solves this problem by using the same degree of resolution for everything.  What you have to keep in mind is Augment Augment Augment.
[...]
So ultimately.  Hero Quest really doesn't suffer from the usual foibles of a linear curve resolution.

Well, it does if you use fixed augments. One roll with a d20 under one number (including fixed modifiers). Using multiple dice could possibly fix that without changing much in the standard HQ resolution.

Nicolas

[Mike Holmes]

I've said this before, but apparently I have to say this again. HQ is not linear. Two dice are always rolled. The results are pyrimidal.

It looks linear, since each side only rolls one die, but that's decieving. Hence there is an expected value determined by only one roll - which is what I think people like about curves. That is, if I roll 2d10 against a TN, the expected value of the roll is 11. With HQ, the expected value of the 2d20 is 21. No that number isn't used as is, but all of the 400 potential combinations of the two dice do fall into one of ten results.

Let's say we've got two guys each with 10W fighting. That becomes 10 for each with the masteries cancelling. The odds of each result are (for either individual in this case):

Code Select Expand

00.25%  Complete Victory
02.50%  Major Victory
25.00%  Minor Victory
19.75%  Marginal Victory
04.75%  Tie
19.75%  Marginal Defeat
25.00%  Minor Defeat
02.50%  Major Defeat
00.25%  Complete Defeat
00.25%  Mutual Fumble

See the curve in there? Yes the curve gets all weird as the abilities move apart, but in general it has areas with higher and lower outcome probabilities, which is what we're seeking. The important thing to do is just ignore the "success" and "failure" results. I mean, if you fail, and I critically fail leaving you with a Minor Victory, that's not much of a failure, is it? Especially when bumps and cancelled masteries were involved.

Note how the tails are very small already? With more dice they practically vanish from play, and the far end results become so rare as to make bumping to the most extreme results very rare.

Mike

[Nicolas Crost]

Since Mike posted already, I will add it here... :)

One more thing is strange with the standard HQ resolution:
If two opponents both have the rating of 19, you will get lots and lots of ties. Both have 90% chance of rolling a success leading to over 80% chance of a tie (both successes) all in all.
Well, you could argue that with two people being equal, that would be expected.
But look at two opponents with 10w (cancelling out the masteries): both have a chance of 45% chance of a success (20% tie) and 45% of a failure (20% tie). So, added up they get a bit more than 40% ties.
That doesn´t make any sense. Basically the distribution of ties changes if you move from the middle of the 1-20 region to the outer ends.
And that is not all, the same thing holds true for different types of successes. With 19 vs. 19 it is almost impossible for someone to score a minor victory (critical success vs normal success: under 5%). With 10w vs. 10w it will be rather common (around 40%).
What the??
Now talk about screwy distribution and different payback of increases in different parts of the range!
I have to add, that I love HQ and think that it is one of the best systems available, but the distributions of the resolution system are ... a bit whonky (is that even a word? sorry, my english). And nothing is as good as to be unimprovable (no word for sure).

Nicolas

EDIT: Damn, sorry again. Got confused about the ties. Forget about that. And the marginal victories should be minor victories (i changed that already).
Mike, I don´t know how you did that neat table in your post, but could you repeat it for a contest 19 vs. 19? My statistics-skill seems to be getting worse...

[Paul Watson]

Keep in mind that in the case of a tie, the lowest roll wins with a marginal victory. In the case of 19 vs 19, the chance of an acutual tie is 5%.

[Paul Watson]

I suck at working out percentages through math, beyond the very basics. What I did instead is write a simple program that, given a certain ability and resistance, runs through all 400 possibilities tabulating the results.

For the curious, I have a CSV (comma-separated values) file with the results for all combinations of ability and resistance 6 through 10W6, showing all the percentage chance of all levels of defeat and victory. Its a smidge over 1MB. If anyone wants me to mail it to them, just send me a PM.

Edit: Its about 161 KB zipped.

[buserian]

Hey, everyone, I was just speculating! I have no problem with the current system, just wondered how it would be change, and whether it would appeal to anyone.

Having said that, I turn to what Mike said:

Let's say we've got two guys each with 10W fighting. That becomes 10 for each with the masteries cancelling. The odds of each result are (for either individual in this case):

00.25%  Complete Victory
02.50%  Major Victory
25.00%  Minor Victory
19.75%  Marginal Victory
04.75%  Tie
19.75%  Marginal Defeat
25.00%  Minor Defeat
02.50%  Major Defeat
00.25%  Complete Defeat
00.25%  Mutual Fumble

Why is mutual fumble relevant? It was in HeroQuest, but as far as I can tell it is not in HeroQuest -- it is a tie like any other.

buserian

[Mike Holmes]

Nick, you're not playing right. Re-read the rules. The distributions are odd, but in a good way. Play a lot, and you'll see what I mean. Really, the system is downright brilliant. At first, I too, thought it was just some odd hack, but if you look at it carefully and in play, you'll find that it really is substantially beneficial in it's normal form.

Why is mutual fumble relevant? It was in HeroQuest, but as far as I can tell it is not in HeroQuest -- it is a tie like any other.

There is a slight difference in that if both fumble in a Group Extended Contest, then the book says that they should both lose with respect to all the other sides in the contest. Kinda vague, but I'm seeing them both losing the bid or something.

I just wanted to be thorough. :-)

Paul, I do exactly what you do, but with a spreadsheet. I just plug in the two values, and it tells me what the odds are of all of the results. Someone here or on HQ-Rules did the same. I think that, given the complexity of the system, that you probably couldn't use a simple formula or set of formulae.

Mike

[Brand_Robins]

There is a slight difference in that if both fumble in a Group Extended Contest, then the book says that they should both lose with respect to all the other sides in the contest. Kinda vague, but I'm seeing them both losing the bid or something.

The chart makes it a little clearer. Both the single participants in the roll lose 1/2 of the bid. This means that they stay relativly close to each other in proportion, but lose out compared to everyone else in the combat.

So if Joe and Bob are in a group melee involving Joe, Bob, Cletus, Cletus-Bob, Billy-Bob, Billy-Joe, and Joe-Bob, and Joe swings at Bob and both fumble, they both lose 1/2 of what Joe bid. Thus Joe and Bob both lose out because they are screwups, while Cletus, Cletus-Bob, Billy-Bob, Billy-Joe, and Joe-Bob don't lose anything -- and thus are in a better place with regards to the two fumblers.