Tomas: Okay, I’m going to search the room.
GM: Give me a Search check.
Tomas: (rolls dice) Aw, shoot. I rolled a 4, which only gives me an 8 on the check.
GM: You spend a couple minutes tossing the room, but you don’t find anything.
Maria: Oh! I want to check, too!
Steve: Me, too!
Samantha: I got a 17!
This is dogpiling the check. It’s not always a bad thing, but when it crops up during play it often just feels… wrong somehow. Are the players pulling a fast one? I mean, it just makes sense that Maria and Steve and Samantha could help search the room. So why can it be so frustrating when this happens?
There are actually three different problems here, and it’s probably useful to split them apart and look at each one separately.
First, there’s a metagaming issue. If Tomas rolled a 30 on their Search check and the GM says “you didn’t find anything,” Maria (and Steve and Samantha) don’t pipe up to say, “I’ll try, too!” It’s only when the thief muffs his roll and fails to find something that the rest of the party uses that metagame knowledge to have someone else make the check.
If this is your concern, you can address it by simply making these types of checks secretly: Since the players don’t know the check result, they can’t metagame it.
This won’t necessarily stop the dogpiling, though. The uncertainty about how well Tomas did can actually motivate the players to always dogpile the check. Other types of action checks may also make it more immediately obvious if Tomas failed or succeeded (e.g., he either picked the lock and can open the door or he didn’t) and have the other PCs queuing up to try, which isn’t metagaming but still dogpiling.
Which leads us to the second problem, which is a pacing issue. Is it really interesting to have everyone sit there sequentially rolling the check? What a huge drag! Let’s wrap it up and move it along! (This is particularly true if there really isn’t anything for them to find in the room and Tomas’ check result is irrelevant.)
You can often bypass this issue by calling for all of the PCs to make a check at the same time.
GM: You’re ushered into a grand ballroom. On the wall hangs a huge, heraldic banner of a red stag rearing on a checkered field of blue and white.
Samantha: Do I recognize the heraldry?
GM: Everyone give me an Intelligence (History) check.
In fact, you should often be anticipating this type of check: All of the characters can see the heraldic banner and they either recognize it or they don’t; it’s not something requiring active study, so you should be immediately asking everyone to make the check. (We call these reactive checks.)
Even if it’s a non-reactive check (like searching a room) and the player is the one proposing the action, you can still try to get ahead of the dogpile: “Is anyone helping Tomas search the room?”
But the third problem is a balance issue.
Let’s say that a particular check has a 70% chance of success for the first character, but a 50% chance of success for the other four PCs. This probably falls into the range of checks that’s both interesting and relevant to resolve. But if the check is dogpiled (with all of the PCs rolling and only one needing to succeed), that 70% chance of success suddenly becomes a 96% chance of success, at which point you have to ask yourself why you’re even rolling the check in the first place.
Compound probability adds up quick, and this is particularly true in systems where the range of the die roll is larger than the skill bonuses: The shift in the average die result over multiple rolls rapidly makes skill almost completely irrelevant. (For example, a DC 15 check where most skill bonuses are +1 to +5 and you’re rolling a d20. The successful check is more likely to come from whichever player rolls the highest number on the d20 than it is to come from the PC with the highest skill rating.)
DOGPILING vs. GROUP ACTIONS
Dogpiling a failed check isn’t the same thing as a group action (where multiple characters are working together). Many games already include effective mechanics for resolving group actions. In D&D 5th Edition, for example, you have group checks (everyone rolls and the group succeeds if at least half the checks are successful) and the Help action (the character with the best check modifier makes the check with advantage if there are other characters helping them).
For more on resolving group actions in any system, check out Part 14 of The Art of Rulings.
FAILURE MUST BE MEANINGFUL
Somewhat ironically, the solution to dogpiling largely goes back to the first principles for framing a check: If you’re rolling the dice, then failure should be interesting, meaningful, or both.
So whenever a PC makes a check, there should be a penalty or consequence for failure.
But yet another reason that dogpiling can feel frustrating is that it can trivially bypass what initially seemed like a meaningful consequence. For example, failing to pick a lock on a door is meaningful because it means the PCs need to find another way to get through the door.
Of course, this also reveals that dogpiling is, in many ways, indistinguishable from one PC repeatedly rerolling the same check.
We could start by reviewing the three techniques described in Failure for the Beginning GM:
No Retries. This obviously solves the dogpiling problem by definition. Whoever made the initial check represented the group’s best effort, and no subsequent checks will change that outcome — e.g., we have established that this door cannot pick and you’ll have to find a different way of getting in.
If your group is used to using narrative resolution, this may be all that you need. But frequently it will just leave people scratching their heads, “Why can’t Samantha search the room after Tomas or at the same time as Tomas?”
A technique I’ve been experimenting with in D&D to “soften” the concept of No Retries is a gradated group check. Basically, it interprets the dogpile as a retroactive group check. The group check requires half of the people attempting the check to succeed, so the second character to make the check has a chance to snatch victory from the jaws of defeat. But if the second character also fails the check, now they’ve dug a hole and it would take two more successful checks for the group to work their way back to success.
The effective application of this technique can be a little limited, but I’ve found it can be a great mental model for social scenes: Tomas has clearly screwed up the negotiations with the vizier. Can Samantha step in and smooth things over? If not, then it will be quite difficult to salvage the situation.
Failing Forward. This preempts the problem because the first character to attempt the check can’t fail, they can only suffer a consequence — Tomas finds the hidden jewel, but triggers a trap; he picks the lock, but his lockpicks break; etc. Since the initial attempt didn’t fail, there’s never a reason for the other PCs to dogpile.
Progress Clock. When all else fails, start or tap a failure clock. The flexibility of this option is great, making it easy to default to when all else fails.
To sum up: Once the dogpile starts, you’ve just got to start looking for additional consequences. Not in a punitive way, just in a, “Okay, you’re spending a lot of time on this lock… what does that mean in the context of the wider world?” way.
DO YOU DARE TO DOGPILE?
With that being said, you also don’t want to discourage the second-best skill rating in the group from flaunting their stuff occasionally. So you may also want to think about how each additional success on a check could provide a benefit:
- A gather information check where each successful check gives a different piece of information, but a failure results in the bad guys becoming aware that the PCs are asking questions.
- A check to craft a magic item where each successful check adds a unique feature to the item, but each failure results in a curse.
This may require a more complicated framing of the check, but it carries the benefits of extra successes needing to be weighed against the increased risk of failure for each additional participant and forcing the group to make a strategic decision.
I should try to fail forward more often for situations that lend themselves to dogpiling, that’s just not something I think of.
My usual solution for dogpiling is not something I see mentioned though: I authorize a maximum of two checks. The first is by the person that initially had the idea (no reason why they shouldn’t have priority, and that gives a chance even to people with low skill), and then if someone else wants to try only one of the characters with the highest skill can try, and only once. This avoids any situation where the super strong guy fails to lift a stone but everyone else succeeds, and it keeps dogpiling to two checks maximum. Of course if the person that had the initial idea is also the most proficient they can’t reroll so only someone as proficient can attempt a second check.
It probably doesn’t solve all issues but I find it a good compromise in practice and it keeps things focused without becoming a rollfest.
Searching is an awkward example, because it’s an activity where just trying again *does* actually make sense. It’s what we do in real life when we lose something, we search again and again, until we either succeed or give up.
I think the no-retries solution is my second favourite here. Asking beforehand how many people want to search and for how long, and adjusting the bonus accordingly, makes everyone involved and the resolution quick and final. The key to make it work is that spending time should have a tangible drawback, in any case.
My favourite solution, however, is another one: not rolling at all.
Tomas: Okay, I’m going to search the room.
GM: Sure. Where do you look?
I was really excited for this article given that I struggled with this too, but… it doesn’t actually provide an answer, it seems?
The proposed solution is that “failure should be meaningful” but this doesn’t really work – or at least I don’t understand how would it.
My problem is with this: “dogpiling is, in many ways, indistinguishable from one PC repeatedly rerolling the same check” is usually not that true. And it’s particularly noticeable on the opening example. Issue is with PCs doing stuff together. It usually makes sense that they do – if anything, it’s weird that example sorta-assumes others just stood in that rooms for these minutes, doing nothing.
Say, party searches the room. They are doing so at the same time. Let’s say that the risk is that they waste too much time so some clock until random encounter ticks at failure or something like that. They all roll, 2 successes, 2 failures. So… what does that mean then? 2 PCs succeeded. So, they found whatever was the something important in the room. Did the other 2 PCs somehow stalled their progress and made the clock tick, ruining the result? Is this 2 bad clocks ticks somehow? That wouldn’t even make much sense.
Generally, when characters work together I’d assume they can’t make the rest perform worse by helping, save for some very specific cases like stealth (in which, ironically, the issue is usually that PCs want not to roll, but can’t without splitting the party). This is becasue usually helping doesn’t make things worse and also becasue that would skew things even further into the ‘no one but the most specialised character should ever try this’ situation – so that seems like a poor idea both in fiction and also from the gameplay standpoint.
“Whoever made the initial check represented the group’s best effort” can work, but falls apart if there are some mechanics more intricate than a basic skill check that might work differently for different PCs. Plus, why won’t helping, well… help?.. It should, right?
I don’t have a neat little perfect answer to this quandary, so I’ve simply started declaring that if multiple PCs try doing a thing that lends itself to dogpiling (e.g. searching a room), regardless of how many are helping, it’s equivalent to giving one PC advantage on the roll. And that’s all. Keeps things simple and quick while also giving a meaningful bonus that doesn’t necessarily guarantee success.
I know this solution doesn’t address things like allowing the PC who’s second-best at a certain skill being able to get the “glory” of doing the thing. But honestly…it’s easy enough to narratively bend the story a bit. Even though the PC with the highest Search bonus is the only one making the roll with advantage, it can still be narratively explained as that character exhaustively searching everywhere *except* the inconspicuous drawer in the corner, and ta-da! The GM declares that actually finds the important object.
I know this will probably upset a lot of folks who dislike the method, but it’s worked just fine for me so far.
If everyone is working together at the same time, that’s not dogpiling on a failed check: That’s either a group check or assistance, and there are usually just mechanics for that.
I erroneously thought that was obvious, but I’ll add something to the article now.
In the past, the game had a nice solution for at least this specific Search problem. Searching a room took ten minutes, and it triggered a wandering monster roll. Reincorporating a consequence like that (time spent and risk of a wandering monster) could help. I think, like you mentioned, it comes down to the stakes, and it’s a good reminder to make sure there are tangible stakes that actually apply to this issue.
I usually tell the group that they can do one check, up to them to choose which character is used for the roll.
Another idea.
Player 1: I search the room
DM: ok, Player 1 is searching the room, what are the rest of you doing while they are doing that?
Ask this before rolling for success/failure of the search. Make secret search checks for every character which is searching in the proper location.
Make alternate non-search actions relevant. A Player stating that they watch the entrances while the search is occurring, or binding wounds, studying that scroll they just found.
At the end of the day though, time needs to be meaningful. Either through torch/spell durations, wandering monster checks, fatigue, etc. When this is the case, the decision to keep searching, or taking alternate tasks during the same time others are searching becomes a choice with actual consequences.
For those complaining that this doesnt make sense because a person can search something over and over:
You are absolutely correct! You _can_ search a room over and over. You can search it a hundred times, and still find _nothing_! Or is someone here one of rhose miracle people who has never, ever lost anything because they could always look again?
The Roll is a contest: did he hide it so well that I cannot find it? If the roll is failed, then _yes_; he did. If the best skill fails, then no one else should have chance, realistically.
Try reversing it: the party thief hides their (rolled like 30 or something) treasure so they arent caught with it in town. Ten minutes later, the innkeeper and his family come traipsing by with their treasure. How? Because they just kept looking until they found it. Fair is fair, right?
I have my group roll a Wisdom save same DC as initial check to retry after a failed check to represent resolve at retrying the task. Multiple attempt’s the Wisdom save increases with each additional attempt.
I have my players roll a Wisdom save same DC as initial check to represent resolve at retrying a failed check. Multiple attempt’s the Wisdom save increases with each attempt. If they can come up with a different skill check (or ability modifier) to represent a different way of looking at the challenge, I’d let them try that without the Wisdom save.
Quite a lot of this can be avoided if the GM is in the habit of checking what the other players’ characters are doing concurrently with the one player character acting.
If a player says their character searches, the first step is to find out what else is happening at the same time. If the other players indicate they want their characters to search too, then e.g. in 5e rules you know you’re calling for one roll with advantage, or whatever. If they choose to stand guard or go search somewhere else or whatever, then that’s what they do.
I thought 3.5 had a decent solution with Taking 10 and Taking 20. Taking 10 representing that you spend 10 minutes to get a roll of 10, and Taking 20 being 20 minutes for a 20 result. Basically once the dogpiling starts I would ask the party if they want to Take 10 or 20, and they have to decide if it’s worth wasting 10 to 20 minutes in one area for something that might not even matter. Then the decision becomes “is it worth it”? And it eliminates all the rolling.
Then for a consequence, you can either have a random encounter as someone noted above, but I prefer to simply indicate to the players that their time spent searching a room, or trying to open a door, has exhausted them a bit mentally, put them behind their schedule, made it so the escaping goblin is even further from them, put the kidnapped child in further danger, etc. Just find some small thing in your narrative that makes wasting time a bad thing, and then bring it up after they’ve Taken 10 or 20.
As Mark said above, always ask what everyone is doing before any checks are rolled.
If there aren’t enough interesting things going on for your whole party to be considering their actions, then it’s likely this should be a group check or no check at all.
I resolve this issue in the following way when a group wants to accomplish a thing for which any one successful roll will provide overall success and multiple, independent rolls might be made by those attempting the task (searching, deciphering an ancient script, determining if an NPC is lying to the party…)
The most-skilled person (highest total bonus in the relevant skill) makes the “lead roll” against the target DC. Anyone that wants to help makes the same roll using their skills against the target DC at the same time. The leader (the “smartest/most-capable person working the issue”) needs to meet or beat the DC on their roll, but can get a +1 (or sometimes I use +2 I haven’t settled on the right value) for each helper that _also_ meets the DC on their roll. I sometimes allow helpers to use an ancillary skill if it is applicable to the issue at hand, though the DC remains the same.
This allows for a helper (or blind luck) to provide aid to the most-capable person without running into the problem with binomial probability (called ‘compound probability’ in the article). This also prevents the most-capable person from looking clueless when they roll poorly; they are the only person equipped to deal with the issue, and no one is really in a position to out-perform them regardless of their roll.
The way OD&D does it, with searching taking a dungeon turn, repeating the action can be done, but also invites wandering monster checks. At the same time, during the example of gameplay given in book 3 Gygax has the caller say 3 of the characters will be searching, and so 3 search checks are rolled. Since they are searching at the same time, it happens during the same turn, but their actions are declared before anyone gets their results. In a more time-abstract game where time-pressures aren’t involved, then having everyone declare their actions, and then not getting a re-try works because it simply represents their best effort over whatever amount of time.
Just get everyone’s actions declared before any actions are resolved. That usually solves these issues.
I had this Dogpiling problem a lot in 3e.
4e solved it (and, for once, as you noted, 5e didn’t roll it back) for me: group checks.
When the players Dogpile, it just becomes a group check.
Very often, if I expect it’s that kind of check, I’ll ask up front, “do you want to pick an ‘expert’ and trust his result, or do you want a group check?”
I still use Skill Challenges, too, the structure of which heads off the problem, amyway.
For the decision to be meaningful, there must be some kind of cost to it. If there’s no reason for the players to _not_ take a particular action, then the decision is already not meaningful, and no mechanics would change that. I would assume that fixing the cause (searching has an opportunity cost, because it requires time) would solve the issue without requiring additional rules.
Burning Wheel addresses these issues quite elegantly. A primary character would test to search the room, and the others would lend helping dice according to their skill level. Burning Wheel’s Let it Ride rule discourages retesting unless conditions significantly change. There would only ever be one roll. Finally, if nothing is at stake (important to the narrative), no roll occurs. Burning Wheel also assumes fair amount of player maturity around metagame knowledge so there’s less need for secret keeping by the gm.
Opportinity cost is my answer – if you’re doing one thing, searching the room, say, then you’re not doing another thing – perhaps staying fully alert to danger like patrolling guards or a prisoner trying to escape (disadvantage on any perception checks for anything else).
The beauty of this is that once you make your players aware of what you’re doing, they start choosing not to dogpile themselves – if you do it right, you don’t have to do anything at all.
When it comes to searching the room, I tend not to care if players dog pile. But when it comes to certain checks, like recalling lore or picking a lock, I often require the character have proficiency to even make the check. That tends to drastically cut down on the number of players making checks after someone fails.
Sometimes when it comes to lore that I want the party to know in order to move the plot forward, I don’t call for checks. I pick there player who I think is most likely to know that lore based on their background, I slip the player a note with the info. When it comes up in the session, I give them inspiration for sharing it with the party.
Several of the comments have alluded to a fourth issue which the article doesn’t mention: the issue of niche protection. Due to the swinginess of the d20, dogpiling can lead to “silly” results like the barbarian knowing more about arcana than the wizard, or being more diplomatic than the bard.
In this case, I think the optimal solution depends on the roll involved. For knowledge skills, you can simply say that non-proficient characters roll with disadvantage. For diplomacy, you can justify it by saying that the first PC rubbed the NPC the wrong way for some reason, but perhaps someone else might have better luck with a different approach. For some checks, if time isn’t a concern, just have the strongest person/best tracker/most skilled thief take 20. And for search checks, this issue is unlikely to come up, since nobody’s class identity really revolves around being “the best at searching”.
@Ron: Taking 10 actually *doesn’t* require extra time, it just requires that you not be in a high-stress situation (e.g. combat). It’s meant to prevent PCs flubbing rolls that should be routine for them, such as a character with +5 Medicine applying first aid.
Taking 10 actually isn’t ideal for search checks, since it’s statistically slightly inferior to rolling the die. You might use it for your *first* attempt, just to make sure you don’t waste time by rolling a 1, but if you really want to make sure you’ve found everything you possibly can, that’s what taking 20 is for. (Which doesn’t necessarily take 20 minutes, either — it takes 20 times as long as normal. For a skill check that’s normally accomplished in one round, like picking a lock, that’s two minutes.)
Okay, first, I think an important part of the solution to the metagaming and pacing sides of the problem is to make certain that everyone has declared their actions *before* any Search checks are resolved (maybe even make it a rule that if the player rolls their result before everyone has declared actions, it doesn’t count and they have to reroll), so that players have to make their choice before they know what the results of the Search check are. Of course, that depends on there being some reason why everyone won’t simply declare that they are collectively searching the room, which leads to my next point.
Second, I think searching is a bit of a weird example here, because searching a room or a building is very much the kind of activity that lends itself to being dogpiled. In real life, if you are looking for something and have friends and family around, you ask them to join the search, and they all look in different corners simultaneously. The chances of finding the thing (and especially of finding it in a hurry), go up exponentially, and there is essentially no opportunity cost or risk to having more people search. In addition, the swinginess of the d20 roll compared to the bonus also matches real life here: if several people are searching the room for a thing, the person who finds it first is not necessarily the one with the best eyes but is more likely to be the one who randomly happens to look in the right corner first.
One thought is that (as you said), dogpiling a check is in many ways equivalent to the same player getting to roll several times. So we can apply similar principles to dogpiling as we do to taking 20: If there is no opportunity cost, resource expenditure, or risk from failure (and, of course, if the nature of the task is such that multiple people can attempt it simultaneously), then let the players dogpile. And if there is some kind of opportunity cost, some kind of risk, or some kind of resource that has to be expended, then again let the players dogpile if they want, and assess the consequences appropriately.