D&D 5E Exploration Rules in latest playtest packet - is surprise to difficult to get?

[zoroaster100]

After reviewing the exploration rules in the new playtest packet I was a bit confused about how the rules for Keeping Watch vs. Sneaking are supposed to interact with the rules for surprise in the Skills section. I was also concerned that it seems too easy to avoid being surprised and too difficult to surprise the other side.

If I understand correctly how the exploration rules are meant to be used, then in a dungeon setting it seems these rules will supersede the normal rules for spotting vs hiding in most cases, except for someone trying to hide in the middle of an ongoing encounter. And the exploration rules say that when moving at a cautious pace, which I think would be the default for most parties during normal dungeon adventures that are not under time pressure, the exploration rules say there is no chance to be surprised. So essentially these rules write surprise out of the rules, except in the very rare situation when a group has to move faster than at a cautious pace.

Even when someone is moving at a moderate pace, surprise seems unlikely. Because each person keeping watch can roll a Wis check and any one of them spotting the foe negates surprise, while those sneaking each must roll a Dex check and any of them being spotted negates surprise. So with a party of several players and several monsters, either side gaining surprise is unlikely.

 

[Sadrik]

I will go back to, remove contested rolls they are too random and offer little value. In fact they only serve to create situations like you elegantly put above.

Sneaker rolls to hide in environment.
Spotter rolls what amounts to a surprise check. (DC = lighting, distance, hidden/not hidden, and skill/ability of spotter.

Done same number of rolls but removed the wildly swingy contested roll.

 

[Starfox]

If by contested roll, you mean when 2 parties make a skill check and the highest wins, it is actually less swingy than a standard roll. 2d20 generate a bell curve (well, actually a triangle/pyramid) and such a curve has much less swinginess than the linear distribution of 1d20. The end results are more extreme, true, but they are also quite unlikely.

What the OP is complaining about is not opposed rolls per se, but that it pits the lowest roll on one side (sneakers) against the highest roll on the other side (watchers). A way around this is to make each party have a dedicated advance scout, that makes the rolls and then returns to his party, warning them of impending danger and possibly maintaining surprse. But generally speaking, it is hard for 2 large groups of people to surprise each other - there is always likely to be some random noise.

 

[Sadrik]

If by contested roll, you mean when 2 parties make a skill check and the highest wins, it is actually less swingy than a standard roll. 2d20 generate a bell curve (well, actually a triangle/pyramid) and such a curve has much less swinginess than the linear distribution of 1d20. The end results are more extreme, true, but they are also quite unlikely.

Think of it this way. Would you want your saves to be contested rolls? Your AC to be contested rolls? No way, because you want to have expected results and tailor the system to be able to account for a base line of defense. with an opposed roll you get no base line you get either amazing success or horrendous failure and occasionally an average roll which is not the norm. It makes sense to roll against the party, however this is purely a math issue.

Over the course you are likely to average a 10.5 roll on both sides of the opposed rolls, however the success rate in any single task is highly volatile. Rather than 1, side of volatility you have two. A low roll of 2 can waste a roll of 20, winning by a huge margin, but the margin of success is not important, thereby wasting the roll.

In the case of hide, grapple, diplomacy, and certain combat maneuvers contested rolls come up. Very swingy as illustrated by the OP. Sneaking often is the bigest no no, it often makes the sneaker roll against multiple onlookers, based on the volatility of the contested roll, no matter if you are a grandmaster at stealth you will fail due to the math of the contested roll and swingyness of contested rolls.

 

[Starfox]

Yes, in my homebrew I use multiple-die rolls (2d6, exploding) for everything, including the equivalent of saves and attacks. I very much prefer this to a linear probability distribution like 1d20.

As to repeated rolls, those suffer an increased chance of failure whether you use a linear or curve probability distribution.

 

[jadrax]

Think of it this way. Would you want your saves to be contested rolls? Your AC to be contested rolls?

A lot of people actually do want that*.


*I am not one of them.

 

[the Jester]

Think of it this way. Would you want your saves to be contested rolls? Your AC to be contested rolls?

Active defense is a pretty well-established combat variant in D&D, actually- it appears in 3e in either the DMG or Unearthed Arcana, I can't remember which and IDHMBIFOM.

 

[Obryn]

If by contested roll, you mean when 2 parties make a skill check and the highest wins, it is actually less swingy than a standard roll. 2d20 generate a bell curve (well, actually a triangle/pyramid) and such a curve has much less swinginess than the linear distribution of 1d20. The end results are more extreme, true, but they are also quite unlikely.

With opposed rolls like this, any curve generated is incidental compared to the greater amount of variability you're entering into the equation. Instead of 1d20 with a 1-20 possible range, you're looking at (essentially) 1d20 minus 1d20, which has a range of -19 to +19, with 0 still occurring exactly 5% of the time. Yes, there's a peak in the middle, but it's no higher than your base rate of success under 1d20.

http://anydice.com/program/21ae

I mean, let's take a pretty trivial example. I'm rolling 1d20 with a +5 bonus against a DC of 10.5 (so I need an 11 to beat it). If you want to simplify it, I am rolling 1d20 vs. a DC of (11-5) or 6. I "win" 75% of these rolls

If I'm rolling 1d20 with a +5 bonus against someone else with a +0 bonus, I "win" as long as... (my d20 - his d20) is -5 or higher, assuming of course that I win on ties. I "lose" if the d20-d20 comes up -6 or lower. Checking the math, this situation occurs 26.25% of the time. If ties are handled otherwise, it gets funky.

Keeping down this path, if I'm rolling with a +10 bonus against someone with a +0 bonus, I "win" if the total is -10 or better. This occurs less than 90% of the time. It's an 11.25% chance I lose, which is hardly what I'd call "rare." (Compare to a static DC, where this would be 1d20 vs. a DC of [11-10] or 1, or a 100% chance of success.)

As far as I can tell, at no point is d20-d20 better for a character with the higher bonus than a straight d20 roll vs. a static DC. In other words, curve or no curve, it's substantially more swingy, not less.

Now, if you're changing things around so you're rolling (say) 2d10 instead of 1d20, the curve absolutely works in favor of the guy with a higher bonus. That's not what you're proposing, though.

As to repeated rolls, those suffer an increased chance of failure whether you use a linear or curve probability distribution.

Only if you're using a "one failure and you're done" model. In which case, every die roll does indeed add more chances of failure.

If you're using a model like 4e's skill challenges where you're counting successes vs. failures, the math works out differently.

-O

 

[Sadrik]

Excellent analysis

Lets hope the game designers pay heed to the notion. Fiddly contests should go the way of the dodo.

I honestly think that many of the complaints for things like grappling and sneaking from past editions are directly attributable to them being contested rolls. The other issues were only partial issues. For instance in 3e, we had huge grapple bonuses to try and make a huge swingy die mechanic work. So that is why size bonuses ramped up so much because you would not want the halfling to accidentally grapple the ogre, that would be a system that did not work- hence massive grapple bonuses.

 

[Libramarian]

As far as I can tell, at no point is d20-d20 better for a character with the higher bonus than a straight d20 roll vs. a static DC. In other words, curve or no curve, it's substantially more swingy, not less.

-O

True, but you could have determined this more easily: the second number (after the average) above the graph on anydice is the standard deviation. This number measures the "swinginess", taking into account both the curve and the range.

2d20 has a standard deviation of 8.15
1d20 - 5.77
3d6 - 2.96