So, according to the PHB, when a group skill check happens, if at least half the characters involved succeed, the group succeeds. Where I would expect this to come up most often is Stealth checks when moving as a group. Success would be overcoming the passive Perceptions of any possible foes. My question is that the wording reads "at least half." So since in DnD you always round down, if you had say 5 PCs, do they need only 2 successes? It feels a bit too easy. I'm not asking RAW. I'm asking what people think.
I think the question at hand has already been answered, and I agree that "at least half" means no less than 50% without rounding, so 2 out of 5 is not enough, but 2 out of 4 is.
---
On the more general topic of group checks, I think it's a great 5e rule, even though I do not use it in every circumstances.
My quick way to choose whether to use it or not is to ask myself: "is the group going to succeed/fail as a whole?" and if the answer is yes, then probably a group check is the way to go.
Some examples...
1) The group is trying to climb over a castle wall to get inside -> group Strength(Athletic) check vs fixed DC
The idea is that if the purpose is for the whole group to get past the wall, they either all make it, or even those who would make it decide not to, in order not to split the party. The group check represents the best climbers helping out the worst climbers, but possibly being dragged down by the effort.
2) The group is trying to convince the king to provide them shelter -> group Charisma(Persuasion) check vs fixed DC
Here the idea is that even if only one person does the talking, the behaviour of the comrades matters. Maybe the king notices that one of the allies looks suspicious, has bad manners, or just farted... all of which can negatively condition the talking. If they are instead behaving properly even if not talking in first person, they can subtly increase the chances of success by making the whole group look more believable.
I like using group checks in social situations because of a few reasons. One of them is that it helps avoiding the old cliché when one PC maximizes Charisma and everybody else dumps it, viceversa it rewards every point spent by anyone on that Charisma score. Then it also encourages
the players to participate in roleplaying the encounter without fear that their not-specialized character will individually be responsible of failure, because their own roll result is lumped together with the others. And no, it doesn't help to send only the Bard do the talking and everybody else doesn't even get into the room to avoid rolling, because numbers make might and 1 person instead of a small army of 5 trying to convince the king means the DC will be higher in that case.
3) Navigating a dungeon/forest when lost -> group Intelligence/Wisdom(Survival) check vs fixed DC
Should it happen that the group gets lost, instead of roleplaying an hour of randomly trying directions, I give them a group check to find the right way. This is very much a group effort, because even if the best PCs would more easily find the way if they were alone, since they are in the group the others will want their opinion to matter, and can lead even the ones who would get it right to be convinced into making a mistake.
4) Sneaking past enemies to avoid a fight -> group Dexterity(Stealth) check vs passive Perception (if routinely aware) or Perception checks (if actively searching for the PCs)
If the purpose if to wholly avoid a fight, then the group succeeds or fails as a whole, and it doesn't matter who spoiled the effort.
Here the slight complication is that, when there is more than one enemy, they might have different passive Perception scores or check results. But as a matter of fact, it only matters if ONE of the enemy notices the PCs, so I only need to compare the group checks with the highest Perception score/result of all the enemies. Then the sneaking effort is spoiled, the group is discovered and the enemies alert each other. If it turns into a battle (it doesn't always have to), nobody is surprised.
5) Making an ambush -> group Dexterity(Stealth) check vs passive Perception or Perception checks
This is different: the PCs are trying to stay unnoticed but ALSO try to get an advantage when they start the fight. Barring rare cases when they might decide
not to attack after all at the very last instant, they WILL attack. For surprise, it does matter whether each one of the enemies notices them before it's too late or not, so the group check results are compared to each one of the enemies passive scores or checks. Which one between passive or roll, it depends on whether I care for some variety or not... most of the times I prefer the monsters to roll, but if there are too many of them, then I might roll once per group of monsters with the same Perception score (e.g. one roll for all orcs, one for all bandits etc.); other times I just don't care and I use their passive scores even if that means less randomness.
This case works notably differently when the monsters are making an ambush against the PCs. In that case I
hate that the best PCs are never surprised if the worst PCs are not, and viceversa that the worst PCs are always surprised if the best PCs are, which is what you get when using passive Perception, so I
always ask the players to roll Perception for surprise.
---
A notable case when I DON'T use group checks is with Knowledge skills. You can use them if you like, but personally I always treat knowledge a bit differently from all other skills. There is always a potentially immense amount of lore that may or may not be known, but at the same time I like having fantasy worlds based on scarcity of knowledge source (very unlike our real world). I want there to be a REAL difference between characters who study a field of knowledge and those who don't, not just a few points of difference on a d20 check. Because of that, I enforce rule 0 to not even allow a Knowledge check to a PC that is not proficient on the specific field of lore: you either invest in the proficiency, or you're not going to answer any question except the most basic. But then for basic common knowledge such as "who is the goddess of Magic?" or "what's the nearest city north of Waterdeep?" I won't even require a roll. That means I hardly even have to consider group Knowledge checks, but if it happens than multiple PCs have the same Knowledge proficiency, then I still don't use group checks but instead reward them more by having EACH of them roll separately so that ONE succeeding is enough.
