A discussion of Keith Baker's post regarding the Skill Challenge system

[the Lorax]

I don't think he used any "math" just good DMing skills and ended up with some anecdotal advice.

Indeed, it seems so. I agree with the ideas and advice he was giving, and some of the things Keith said make excellent house rules, rules which are basic enough that they SHOULD have been put into the rulebook.

As written, the Skill Challenges seem to be tougher than intended or reasonable for a moderate success rate.

 

[Roxlimn]

If I may, I have to ask why the community is expecting so much of this skill challenge framework. I mean look at it! The Skill Focus benefit seems to have been pulled out of nowhere, and the power benefits are much too rounded-off to have been done with math to any considerable degree. Contrast that to the numbers for the combat damage and it's obvious that the math here wasn't done with any kind of exhaustiveness or polish.


This skill challenge thing isn't actually new. Snippets of it started appearing in adventure design around late 2005, IIRC, and definitely was included in the Red Hand of Doom superadventure. Certain frameworks of obtaining crucial information appeared in adventures as early as Tears for Twilight Hollow and The Dead Gate.

So really, it was a good thing this was codified in the 4e for all of us to have a common reference point, but this is really passe at this point.


These days, I'm doing variations on the skill challenge structure, also using partial successes as Baker does. This was actually the norm for when you used multi-Knowledge checks and Gather Information in 3e to root out important clues.

I'd like to smooth out the math on it, but winging it has so far worked for me. The fundamental problem with the basic skill challenge structure outlined on page 72 is that its length and ultimately its chances for success are based on failure rates. This lead to extremely swingy mathematical models where you either have massive success or massive failure just based on a few moderate skill adjustments. It's like a test that's graded with right answers-wrong answers. You either get really, really high, or your grade plummets like a rock.


The solution I've found to counteract this is to NOT base the length of the challenge on failures but on complexity itself and time constraints. If you only have 10 minutes to fast talk a bugbear out of attacking you, for instance, you only get so many checks. Each failure counts against you, but only in that it's not a success. There's no chance that a string of failures will end the encounter prematurely, and each successful check has some small measure of consequence. Figured this way, success rates are more stable and easier to manage.

Moreover, you can adjust that. If the players are really getting into the challenge, I reward clever ideas not only with bonuses on certain checks, but sometimes with the opportunity to make MORE checks. Since failure doesn't kill the encounter, each chance is another swing at the treasure and players are well motivated to look for ideas to extend their chances at success.

Again, this is easy to manage. Each additional check has a flat rate of success and it's easy to figure on the fly how much successes on average a given number of checks will add.


There is also the "reverse challenge." You can use these for extremely high DC challenges. It creates the atmosphere of "we barely made it." Essentially you swing the thing around. You only call for a small number of successes which will then end the challenge. Failures likewise do not count towards ending the challenge, but there's only a certain number of chances you get, modified again by whatever clever idea you can come up with.

This mode of success manipulation and management is actually nothing new. Game shows have been using this forever. You just need to select the manner of framework you want to evoke the mood that you want.

 

[ObsidianCrane]

Adding in partial successes is a rather serious tweaking of the binary pass/fail system in the DMG.

The DMG doesn't have a binary Pass/Fail system.

Binary Pass/Fail would be - win the skill challenge or stop the game, analogous to a TPK in a combat encounter.

The DMG has a variable Pass/Fail system. This is far more analogous to water running down hill. Water running down hill always takes the easiest route first (Pass scenario) however, when that route is blocked it finds another way around the obstacle (the DMG's Fail scenario).

Now the DMG does a pretty poor job of explaining this admittedly.

It also does a poor job of explaining the use of Group Skill Checks (which I see no reason not to use as Hellcow has been doing given the RAW. The climbing check is an example, not an exclusion.)

I do think the Skill Challenges suffer from author blindness - that thing that happens when you write something out, and when you re-read it you add missing words or even sentences which significantly change the meaning. The meaning they are intended to convey is very similar to what Hellcow's blog suggests I suspect. What they actually say is not always exactly that.

Now I only have the PDF atm (Amazon needs to learn how to ship to Australia faster *sigh*) so I have to wonder did the table on Page 42 change the Moderate DC for level 1 to 20 from the 15 it says in the PDF?

I see lots of people arguing off a DC of 20, when I read the usual DC as 15 for a 1-3 test per the stuff on page 73. Something that should be 50/50 or more favorable for a skilled character, heck even some Unskilled characters could be looking at better than 50% success rate.

(I'm liking the idea of the Half Elf Paladin Diplomat with a starting Diplomacy check of +14!)

 

[Tuft]

I see lots of people arguing off a DC of 20, when I read the usual DC as 15 for a 1-3 test per the stuff on page 73. Something that should be 50/50 or more favorable for a skilled character, heck even some Unskilled characters could be looking at better than 50% success rate.

It's been covered a few times in these threads, but it's so confusing that it might as well be covered a few times more.

As I understand it: Skill challenges on page 73 refers to the DCs on page 42, which are generic DC:s for the given level ranges. That is, they apply to all kinds of rolls, like attack rolls and pure attribute rolls. Since the possibility of skill training makes you in general have a higher skill than attribute, you have to apply the +5 in the footnote for all rolls that involve skills. Thus 20 instead of 15.

The important part is the footnote: "For skill checks: Increase DCs by 5", and the text "you pick an easy DC: The table says DC 15, but it’s a skill check, so make it DC 20." in the middle of the right column on page 42. As page 73 just refers to page 42 for DCs, those modifiers would apply.

But, I admit, that modifier is a bit funny if you think about it. Why should something be harder because you chose to use a skill on it, rather than brute force? But, that is what the RAW say...

 

[Transformer]

I think a lot of people are forgetting the core problem with RAW skill challenges as identified by Stalker0: variability.

I can add all kinds of clever and interesting DM fiat bonuses, aid another usages, and original player options in my skill challenges, but that doesn't help with the variability. Low complexity challenges will still often be far more difficult than high complexity challenges. The chance of success of a 3 member party will still be very different from that of a 6 member party. The chance of success for a level 1 party will still be very different from that of a level 15 party, which will be very different from that of a level 30 party.

The problem isn't that I have to use lots of circumstance bonuses, aid another opportunities, and special options to give my players a reasonable chance of success. The problem is that I have to include drastically different amounts of circumstance bonuses, aid another opportunities, and special options to give my players the same target chance of success when the party level, or complexity level, or party size changes. That makes skill challenges very difficult to design.

How do I know that the amount of help I gave that 6-person, level 15, average-skill-modifiers party on their complexity 4 skill challenge to get them a 50% chance of success will also net my new, 3-person, level 1 average-skill-modifiers party a 50% chance of success on their complexity 2 skill challenge? I don't, because it won't. The base system should be mathematically stable so that DMs can easily do all the cool stuff that Keith is doing. But the system just isn't; there's no way to predict how much the help you're giving the players is really going to help them. A DM has to significantly change the amount of help he gives as the party levels, if the party changes size for a session, or whenever he changes complexity. He has to completely re-asses how much help he gives when he gets a new group. A DM just shouldn't have to do all that.

Keith's "partial successes" house rule help with this a little, but not all that much. If the same amount of special options and circumstance bonuses gives one party a 25% chance of success and another party a 65% chance of success because of party size, party level, or some other factor, even when the parties both have perfectly average skills for their levels, that's still a massive disconnect. That first party is going to be getting much weaker partial successes, even though their skill modifiers are no worse than those of the second party. Why? Because the system is fundamentally broken.

 

[Spatula]

The DMG doesn't have a binary Pass/Fail system.

Binary Pass/Fail would be - win the skill challenge or stop the game, analogous to a TPK in a combat encounter.

Errr, no. You either win a skill challenge, and get XP, or you fail the challenge, and you get nothing. The ramifications of success or failure are unimportant from the viewpoint of the challenge itself. Did you get XP? You passed. No XP? You failed. There is no in-between state.

 

[Harr]

The problem isn't that I have to use lots of circumstance bonuses, aid another opportunities, and special options to give my players a reasonable chance of success. The problem is that I have to include drastically different amounts of circumstance bonuses, aid another opportunities, and special options to give my players the same target chance of success when the party level, or complexity level, or party size changes. That makes skill challenges very difficult to design.

How do I know that the amount of help I gave that 6-person, level 15, average-skill-modifiers party on their complexity 4 skill challenge to get them a 50% chance of success will also net my new, 3-person, level 1 average-skill-modifiers party a 50% chance of success on their complexity 2 skill challenge? I don't, because it won't. The base system should be mathematically stable so that DMs can easily do all the cool stuff that Keith is doing. But the system just isn't; there's no way to predict how much the help you're giving the players is really going to help them. A DM has to significantly change the amount of help he gives as the party levels, if the party changes size for a session, or whenever he changes complexity. He has to completely re-asses how much help he gives when he gets a new group. A DM just shouldn't have to do all that.

Keith's "partial successes" house rule help with this a little, but not all that much. If the same amount of special options and circumstance bonuses gives one party a 25% chance of success and another party a 65% chance of success because of party size, party level, or some other factor, even when the parties both have perfectly average skills for their levels, that's still a massive disconnect. That first party is going to be getting much weaker partial successes, even though their skill modifiers are no worse than those of the second party. Why? Because the system is fundamentally broken.

QFT, this is, of course, the actual issue here. And the reason why some people (which is to say, the people who have actually played and ran a number of skill challenges with an awareness of what's going on) get annoyed when they see something to the tune of, "Dude skill challenges are fine, just add +2!". No, they are not fine if you add +2, +12, +342, or whatever.

Of course, in order to understand this, you need to be someone who plays, runs or simulates challenges a whole lot. You need to see the averaging and the variance in action. As of now, there are exceedingly few people in this position (though there are plenty who posture to be ) hence why all the off-hand dismissals and all the attempts at simple anecdotal resolutions. They are not enough. But people in general will not be convinced of that anytime soon.

 

[ObsidianCrane]

Errr, no. You either win a skill challenge, and get XP, or you fail the challenge, and you get nothing. The ramifications of success or failure are unimportant from the viewpoint of the challenge itself. Did you get XP? You passed. No XP? You failed. There is no in-between state.

The purpose of a Skill Challenge is not to give the PCs XP, its to highlight a part of the story as important (yet, in the RAW, not critical).

Getting XP is a mechanical reward for success, its not the goal of a Skill Challenge.

The goal has everything to do with the overall story of the adventure. Success at the challenge should be important to the adventure, but not essential. You don’t want a series of bad skill checks to bring the adventure to a grinding halt. At worst, failure at the challenge should send the characters on a long detour, thereby creating a new and interesting part of the adventure.

Thus Skill Challenges do not have a binary Pass/Fail situation. You can learn things in the process of failing to make it possible to find another path to success for example.

Oh and thanks Tuft for the clarify on that bit...off to house rule land I go...
[Sblock=House Rules and Page 42]At low levels the numbers seem tough for trained characters, but right for highly trained ones (Stat 16, Racial +, Trained, Skill Focus). At high levels they are easy to automatic for trained characters, and even worse as DCs for attacks etc.

Yet if you look at the monster design rules on page 184 you get a DC of 42 for level 30 Moderate Attack DC using the +2 for Attacks guideline in reverse (ie 44 -2 => Moderate DC = 42). Compared to the page 42 table's DC 29, DC 42 is a reasonable target for a 30th level character and still offers a chance of failure.

DC42 also exactly matches the "Other Defenses" section of the pg 184 table; "Other Defenses = Level +12". That looks like a good starting point for comparison to skills. So I'd go "15 + level" for moderate skill checks. (ie the DCs on pg 42 are off considerably)

Yes that still makes it easy to very easy for people that are "focussed" on a skill, but keeps the difficulty on par for everything else. Trained characters will need in the 7-12 range, and untrained characters can still succeed. At higher levels you will need items with bonuses for the skills and/or higher stats to keep making the checks reliably which seems on par with the combat rules.[/sblock]

 

[Spatula]

The purpose of a Skill Challenge is not to give the PCs XP, its to highlight a part of the story as important (yet, in the RAW, not critical).

Getting XP is a mechanical reward for success, its not the goal of a Skill Challenge.

You're confusing the mechanical system of skill challenges with their place in the overall adventure. Skill challenges propose that if the party makes 2X skill checks before failing X skill checks, they have "beaten" the challenge and get XP (and they fufill some goal within the story). Otherwise they "fail" the challenge and don't get XP (and they have to find some other way to get to their goal). Those are the only two possible outcomes in the system as outlined in the DMG. All or nothing. 1 or 0. On or off.