D&D 4E Showing the Math: Proving that 4e’s Skill Challenge system is broken (math heavy)

[Rystil Arden]

In the context of skill challenges, if the DM has preplanned the room in detailed and History is an allowable skill, then it should either 1) find the hole that the DM put there; 2) find something else that the DM put there that would let the PCs escape. Lack of narrative fidelity is a DM thing, not something due to skill challenges.

This is quite true--and when the cell is pre-planned doesn't have something for someone with history to find, History would not be an allowable skill, and that's perfectly fine, fair, and reasonable within the bounds of that skill challenge. I was responding to a post that pretty much implied that you're playing it wrong if some skills are ruled out from the skill challenge (despite the 4e DMG doing so in the examples).

 

[LostSoul]

???

Uhm.....why do you think the "+5 doesn't seem to apply"? Because of the page title??

If you need more help, try page 61. Both pages 42 and 61 are entire consistant. ...and entirely oblivious to the underlieing chance of success.

Well, hey. That's interesting.

I won't add the +5. Maybe it's not RAW, but it's a fine house rule.

 

[biotech66]

I saw the math portion of the article, but I thought that skills were to hard when I considered simple die rolls and 3.5.

Back in 3.5, a base DC was 12+party lvl. It seemed good to me then and it meant a ~75% chance of success with a trained skill check.

The only thing I am still thinking about is the definition of failure. If you only fail by less then 4, the trap does not activate. Maybe this should be consider for all skills? Anybody run the numbers with a less then four failure not resulting in a failure in the check?

 

[Harr]

Well, hey. That's interesting.

I won't add the +5. Maybe it's not RAW, but it's a fine house rule.

Heartily agreed

 

[abeattie]

Hmm.. this thread became monsterous before I got through it -- but let me suggest the following.

1) if we're going to assume PCs are buying up to 18s, then the starting PC will likely have a 20 in their "best" skill -- assuming a racial link-up.

2) further -- most of the races give +2 to two skills, bumping the "best" skill by another 4 levels worth of bonus.

3) If your DM is skill-challenge happy, may I suggest skill focus to get that +3 (6 levels of bump).

I will include 1 and 2 and discount 3 as too much of a commitment..

so the starting PC (at level 1) is rolling d20 + 5 for training, +5 for attribute, +2 for race. (+12)

Other factors may come into play to move it further (+1 to everyone's diplomacy if a half-elf is in the group, for instance).

So starting out there is a flat 60% pass rate per roll against a moderate DC. (75% passage rate for the skill focused PC.) Yes -- if you want your first level party to pass without minmaxing their PCs, either they need to get luck, or you need to softball the encounter. (or let them fail).

Where things shift is at 2nd level. Many of the classes gain abilities that give them bonuses to a skill check, let an ally reroll a skill check, let the PC re-roll, etc. In other words -- if you want to succeed at skill challenges you need to actually use utility powers. The extra plus one kicks in at 2nd level as well -- pushing odds of success on individual rolls on the "best" skill up to 65% base -- with one or two rerolls, or a skill bonus.

I won't commit to calling it a balanced system, but I don't think it is as off as suggested.

 

[Morrus]

They explain aid another in one section, but don't go into with their skill challenges sections. Even in their example they don't use aid another, so I'm not sure when you are supposed to add it in.

1) You can use it for any skill check; i'ts thus an integral part of the number-crunching. Not incoprprating it into your model doesn't provide a result which accurately reflects 4E D&D, merely one which reflects a similar system you've based on it.

2) In addition, there's nothing to stop PCs choosing "easy" for every single check.

3) Additional modifiers will mean that the highest skill score will often exceed 9. Racial bonuses to both the ability score and the skill itself; bonuses granted to other characters; basic equipment (thieves tools, climbing kit, all of which add 2).

What you've done is identify one disadvantaged strategy for the PCs, assuming they do nothing to help themselves. In which case, yes, the odds are against them.

In short - your mathematics is good, but it doesn't go nearly far enough in simulating the system as outlined in the rules. It skews the situation against the PCs at every possible juncture because it forces them to act like klutzes; and thus the result is unsurprising. You're missing modifiers (at the other extreme end - but you've taken the lowest extreme, so I'll take the highest to illustrate the possible range) of: +4 skill; +8 aid another; easy checks only giving a DC 5 points lower, a MASSIVE difference of 17 points by simply making optimum choices.

My belief, then, is that you have failed to demonstrate (yet) that the system is broken.

 

[pemerton]

You don't want to elevate anecdotal evidence over the math.

This is true - the mathematics is accurate. But the mathematics is assuming certain percentage chances of success on skill rolls. To me, this gives rise to the question - what are the expectations about modifiers to the skill checks? In particular, is there something that the critics of skill challenges in this thread are missing?

People keep going around saying essentially "our quick five-minute analysis is perfect, so obviously WotC's several months of detailed design and playtesting was a total joke". Meanwhile, attempts to show that the core assumptions of the five-minute analysis might be flawed (the +5 rule) are shot down without detailed examination.

I'd add to this: I was able to work out the odds of success for comlexity 1 challenges using my memory of the high school maths (probability trees) that I learned 20 years ago. It took me about 20 minutes. Presumably some of the WoTC designers can also do high school maths and had 20 minutes to spare, or even more, to look into the issue.

So the real question becomes - what are we missing about either DCs or skill bonuses?

Its not a question of the 4e playtesters understanding the math. Its this: assume that someone out there did 20 playtest skill challenges. According to this thread, they should have lost the vast majority of them. Did they?

Related question- people have been using skill challenges for a while. Are they losing all of them? Are they adjusting the DCs in some way?

Basically, it seems very unlikely to me that everyone using skill challenges has been losing the vast majority of them. This suggests that something else is going on that is not accounted for in this thread. Perhaps it is DM intervention regarding skill challenge DCs. Perhaps it is players optimizing their choices more effectively than this thread suggests. Perhaps it is something unknown.

I'd like to know the explanation, to help me in my own work as a DM.

Cadfan, I agree entirely - these are the questions that need to be asked and addressed.

As per my earlier post, to get around a 75% chance of success on a complexity 1 challenge you need an 80% chance of success on each check, ie, any roll of 5+ on d20 is a success. Given that the DCs range from 20 (at level 1) to 34 (at level 30) this requires a bonus of +15 to +29, or around +14 after level adjustments.

Subtract 5 for training and 4 for stat - this means that another +5 is needed. Where is that going to come from? Either you can drop the DCs by 5, as Harr, Lost Soul and some others are advocating, or you can introduce some circumstance modifiers, or you can introduce a bit of Aiding Another (gives up to +8 according to the relevant sections of both the DMG and the PHB), or you can introduce some racial bonuses (+2) and Skill Focus feats (+3), or you can introduce some auto-successes to help even the odds. Magic items will also help, obviously.

What I would like to hear is a bit more about what techniques are actually being used in play to make skill challenges work.

I won't commit to calling it a balanced system, but I don't think it is as off as suggested.

Agreed.

 

[Saeviomagy]

Trapped in 3e? No. Being reasonable? Yes. And if you look at the 4e example Skill Challenges in the DMG, they also make choices that certain skills will not work (there's an example of a low-level challenge where Intimidate automatically generates a failure, even if you roll a 50).

Which a change to the maths won't help.

It of course depends on whether the GM has planned out the room more carefully. I chose an example that someone actually used in a pre-4e discussion thread. They argued vehemently that History should make a hole in the room even if the GM had pre-planned the room in detail without the hole. This is ludicrous to me. And this isn't a 3e/4e thing. The 4e DMG specifically indicates that sketchy skill uses that don't really fit should be automatic failures, result in penalties to later rolls, or have extremely high DCs. I have every indication from the 4e DMG that 4e is approaching this in a reasonable fashion.

I totally agree with you - if the DM has designed something down to that level, then it's crazy for a hole to suddenly appear due to a history check. If the cell truly has no history of being compromised, then I would expect a history check to tell me that. Making it an automatic failure however would be a bit silly though.

The GM agreed that we could get clues to this information with Perception. But we had specified what we wanted to learn ahead of time (he asked us to state it before setting up the challenge), and those things were not among them.

That seems a little odd. I would have thought the goal would be "get some information out of this guy". Making it a lot more focussed than that seems to severely limit what scenarios a skill challenge is actually any good for.

I agree that in some cases, there are more skills that could help than in others. In our case, we knew already where the cult was based, etc. We just wanted to know about the identity and abilities of the higher-ranked members and their end goal. Perception would not have helped with that (a good example of where it might have helped was if the cultist had a hidden object on him with a symbol of Orcus and we didn't know it was a cult of Orcus or something, but we knew, and he didn't). This is from KotS, which seemed to have pretty detailed information of what this guy had and knew.

It is indeed the numbers--some skills do not have a place in some skill challenges (or if they do, they have a place as defined by the DMG as auto-failures, penalties to later rolls, or exceedingly high DCs). The GM was doing us a kindness by telling us this ahead of time for each suggested skill use (since we weren't veterans of skill challenges and were trying to feel it out) rather than just letting us roll and then failing.

Auto-failures aren't really working with the numbers. Personally I think they're a terrible idea unless they're incredibly well justified, for precisely the reasons that you suggest.
Interestingly enough, the DMG
1. Makes no mention in the text of skills that cause automatic failures, only in an example
2. Seems confused as to whether a secondary skill should be harder, be only useable once or both (ie - one passage says that a secondary skill should be harder OR useable only once or both and gives an example of an easy check that's useable only once, while the other says it should be harder AND useable only once).
3. Offers the possibility of a +/- instead of 2.
4. Suggests that actions in a skill challenge should possibly have secondary effects (ie - learning extra information that was not specifically set up as the goal of the challenge)
5. Suggests that ideas that line up with your preconcieved primary skills should probably get the same benefit.

It seems to me that a DM who wants skill challenges to be fun will, according to the other DMG advice, go with the more lax options.

 

[Lurker37]

Hmm.. this thread became monsterous before I got through it -- but let me suggest the following.

1) if we're going to assume PCs are buying up to 18s, then the starting PC will likely have a 20 in their "best" skill -- assuming a racial link-up.

2) further -- most of the races give +2 to two skills, bumping the "best" skill by another 4 levels worth of bonus.

1) Assumes that the character's prime stat AND racial bonus apply to a skill they trained,m and that this skill will always be useable in every skill challenge. That's hardly a safe assumption. In fact I would go so far as to declare it unlikey.

2) Further assumes that this character also got their racial bonus. If the stars did align to not only allow this but also to let this happen in a skill relevant to the current challenge, then I'd expect that character to have a far, far better than 60% chance of success. 80% sounds better (95% with +3)

And it ignores that having a single skillful character make all the rolls is exactly what the skill challenge system is meant to be avoiding.

 

[gribble]

I find it hard to believe that no party members had skills that could be used aside from one or two characters with good scores in prime abilities. I think it's more likely that your players are falling into the trap of treating a skill challenge like a 3e style "make a roll on this skill and only this skill. Anything else will fail. Only the highest score need apply" skill contests.

Not at all. We all tried to come up with "creative" uses for the other skills the characters had (Intimidate - no, expressly forbidden by one of the skill challenges; Althletics, Acrobatics, Stealth, Perception - none really seemed terrribly applicable to any of the players - or at least they were unable to come up with any creative uses of the skills that weren't just stupid, given the circumstances).

Besides, as someone else pointed out, all of these secondary skills would have used the hard DC (25). Even with a +10 in a skill (unlikely at 1st level), that's effectively only a 1-in-4 chance of success. As I already stated, the rules made those players afraid to try, because it was much more likely to lead to failure than success...