The difference is that the AI need to be able to specifically target the hidden model with an action in order to test for detection and any subsequent attack; ergo a melee-only Adversary wouldn’t be able to actually move, engage and attack a hidden model (assuming it was within its movement/sprint range of course) unless that detection test is instead carried out at the point of determining the Response target priority (which is what it is suggested happens and, I believe, @Modiphius-Dom also clarified). If they fail that detection at that point, they can still carry out the response, the target is just now the next one according to their priority; the detection test is carried out before the actions are actually carried out.
A player model who was attempting to engage in melee works differently as the detection timing doesn’t seem to be linked exclusively to an action as such; they would move towards the hidden model, attempt to detect and, if successful, engage. If they fail (and presumably stop 1/2 an inch away), they can carry on their movement (and engage another model instead if they like).
If the hidden model is 34" inches away then yes, an adversary who doesn’t have a ranged effect (or without one of sufficient range) can’t attempt to detect them because they can’t target them with an action. Which seems…reasonable?
I think I don’t understand this . Then, why @Modiphius-Dom says you need a valid action interacting with the hidden model to get the detection roll, and you say, without engaging, is a valid action wich interacts with the hidden model just the fact of determining the target priority list? Only because the adversary in the next round maybe could move again to engage or attack if it is already engaged? In that case, every single adversary would try to remove hidden tokens when activates, regardless of its status, responses, or actions…
The detection test is carried out when the model announces a valid action using the hidden model as a target, as far as I know, and before that action is carried out, but already announced (so they cannot cancel that action and use a different one until every legal target has been checked and it is impossible to carry out that action)
Moving to a point on the battlefield is not an action wich interacts with the hidden model (without engaging).
According to you, I would not know at wich distance the adversary would get a detection roll, you say not at 34", even if the hidden model is the target of the response, and therefore, the 2 actions of the adversary, so the same at 12"…
Obviously, if with the action it is going to engage the hidden model, then yes, it would get the detection roll, but the rest of movement situations, then not, it would not matter any distance.
To my understanding (and I may well be corrected on every part of this):
A Player controlled model would make its detection roll when it declares the hidden model a target for an Action. Engaging in melee is slightly awkward because it’s not 100% clear when you make the test as the engagement is part of the movement (which doesn’t need a target as such). This could be better clarified I agree. My personal instinct would be that the detection test is taken just prior to the actual point of engagement (i.e., the activated player model stops just short of base contact, takes the test and, if they pass, carry on into base contact, if they fail, they can carry on moving (if they have any movement left) and engage a different target if they like (or not engage/attack at all and do something else if there’s no other valid target). This is the best way I can find to reconcile the wording on page 27. with how engaging into melee works; (one of the main stumbling blocks in the timing being that there’s no mention of having to actually declare an attack action until after a model has already engaged, at which point the ‘target of an action’ is a moot point).
An Adversary model makes its detection roll when it targets a model for a Response (not the target for an action, even though that Response obviously includes taking actions and the target for those said actions is likely to be the same model).
Correct; although, as mentioned above, strictly speaking engaging doesn’t in of itself have ‘a target’ hence why I’ve tried to find a way to make this workable above and yes, I agree this should be clarified in the new version of the rules.
You’d get the detection roll for an adversary IF, as part of their Response, they have actions that would be in range to target the hidden model (so it depends on A) the Response and B) the range of whatever actions/attacks that adversary has). What’s more you’d carry out that detection roll at the point of determining the target for the Response (so before carrying out any actions).
So, for a non-specific example, say an adversary activated and it’s response was to Move to the nearest enemy (sprint if able) and then perform a Ranged attack (and its attack has a range of 16"). There is a Hidden enemy within 20" and another, non-Hidden enemy within 35".
Adversary activates and its nearest (20") enemy is Hidden. So it takes the INT roll for detection. It passes and so it moves 6" towards the previously hidden enemy (which is now only 14" away) and can make its ranged attack against it.
Or
Adversary activates and its nearest (20") enemy is Hidden. So it takes the INT roll for detection. It fails and so it can no longer target the Hidden enemy. So it checks for the next nearest enemy which is the one 35" away. As that model is not hidden, it can still be the ‘target’ of the response and so the adversary will sprint 9" towards that enemy. However, as that enemy is still 26" inches away, it’s not in range to attack, so the adversary’s second action reverts to something else (usually a Block or their activation just ends).
Sorry that was a long response and it’s a wordy issue (I’m trying to explain it to the best of my knowledge and be as clear as possible with the example but there’s a lot of variables to try and cover). To reiterate, I do take your point on the timing of the detection roll not being abundantly clear currently and the lack of a ‘target’ for engaging in melee does make it quite murky to try and decipher properly.
If this is true, me, as a Dark Brotherhood player, when I play against the AI, I deploy with Obscured or blocked LoS to get the hidden token at the beginning. Let’s say the AI begins the game, and they roll M or A, they are at 30" for example, and if they have some ranged attack, they are all going to get a detection roll (if their weapons reach after moving, according to your words) because they have 1. Nearest in their priority list, so… how could they have any “nearest” target to go and do something if they fail the INT test (Draugr roll against INT 1)?? would they move with scatter direction? What should they do if they cannot roll or they cannot remove the hidden tokens? do they block/hidden and stand there, don’t they move with the M response?
I agree all this should be accurate in the rulebook…
In that situation then the adversaries who rolled M or A would probably just stand still and block/attempt to hide if able, as their response dictated or their activation ended. Which could be a bit boring for the first Round but in most scenarios you’d be moving towards an objective or similar and getting close to the Adversaries and giving them more chance to detect the hidden models.
I’ve been trying to read round it a bit and I am wondering if, when a Response requires an adversary to move towards a model (so most Move, Attack, Defend responses etc.), it will do so, even if that target is Hidden and then it does the check for detection if its second action would target that model. This also works most of the time except, again, when you’re getting to the Engagement issue because that doesn’t have ‘a target’ in terms of movement as an action.
There is a paragraph mentioning that an adversary’s target priority can alter mid Response if its first action causes there to be a change; such as if it moved near a model which then ran away due to Terror and was no longer the nearest.
I’ve honestly talked myself in a circle here and I can’t reconcile the Adversary procedure with what’s written unless my assumption (much) earlier above is correct: that a Hidden target can only be detected by an Adversary if it’s eligible for the INT test at the point of allocating target priority (i.e. immediately after the Response has been rolled but before the Response’s actions are carried out). Otherwise it gets pretty convoluted (especially when it comes to Engaging). Regardless it could do with clarifying in the updated rules.
Yep. That would be my understanding of it (or would it actually… hold on).
Maybe not as it doesn’t have a valid action to target the hidden model with unless it detects it (as per @Modiphius-Dom’s comments above); you’d still have to check against its potential second action to see if it could detect it via that action (i.e. the potential attack), otherwise you go back to assigning target priority to the next eligible enemy.
My understanding is, until now, the adversaries would roll to detect the hidden model if they can engage or attack at any point of their response, after announce the action, and before carrying it out, so, if they need to move to the nearest model at 20", and that model is hidden, and they cannot potentially attack or engage at any point of their actions, they would directly ignore this model before moving, and try the same with the 2nd nearest model… in the case they cannot find anyone, because they are too far, or near enough but they fail the INT test, they stay there and block.
In the case of that first round hidden, as I like to do with my Dark Brotherhood models, they would stay there and block, if, before moving (because they don’t know where to go with M or A) you calculate as impossible any actions wich could engage or attack
@DK-dark has it exactly right here in terms of when you do your detects - after announcing the action but before carrying it out. This gives you the chance to swop if if the Detect fails.
Movement doesn’t target but you would need to do your Detect at the same point as with other Actions - after announcing but before doing it.
Its also worth remembering that you still need to have the possibility to see a Hidden model - you can’t detect a model that is Hidden if the line of sight is Blocked - it has to be clear or obscured. If @DK-dark Dark Brotherhood start with a Hidden token and they are behind cover, they will most likely also have Blocked line of sight. The Hidden status reduces your Height by one, remember. If you are hiding behind walls or barrels or bushes that are also Height 1, or are completely behind higher cover, the enemy can’t make a roll to detect you as there is no line of sight. They have to keep ambling towards you with no chance of detecting you until they get within 6". If you are hiding out in the open, without anything to hide behind, then they can try and detect you as soon as there is a possible Line of Sight.
It’s not explicitly stated in the current rules, but the AI forces are assumed to have a general idea of where their enemies are, so they can still move towards the Player’s models without having seen them. This was done for simplicities sake and to speed up the time it takes for engagements to occur. We do have some scenarios where enemy models will follow patrol paths, or stand on guard, but we found that random movement for AI models slowed the game down a lot and resulted in a few turns were nothing much happens at the start of the game. If you only have, say, 6 turns, then waiting for 3 of them until you can start engaging the enemy was unsatisfying.
However, with the introduction of the Dark Brotherhood - well, those guys sneak around for fun. We’ve got our next Histories of the Empire supplement in the works, so you may see some additional expansion on the stealth mechanics in that for games focussed on sneaking around.
Thanks Gavin (and @Modiphius-Dom) for the further clarifications.
Just for the sake of being concise;
An Adversary doesn’t need to have detected a hidden model in order to move towards them as part of a response (just to target them for an action).
Detection doesn’t have a range (as such), there just needs to be clear or obscured line of sight; i.e. it doesn’t matter if the active model doesn’t have an action/attack that would feasibly be in range?
Adversaries roll for detection at the point of assigning target priority (so before they carry out any of their response; they wouldn’t, say move towards a hidden model and then attempt to detect).
At what point does the detection test take place when attempting to engage in melee (this is pertinent for player models too as they’re allowed to carry on their move if they fail to detect)?
Yes, the explanation was assuming they have obscured LoS with the hidden model.
But according to @Modiphius-Dom , they don’t take into account the hidden models with blocked LoS when you look at the target priority list on the faction card, then it could bring an odd situation against the Dark Brotherhood in the 1st round, so you would need to know where they go in that case. But according to you, they would move to the nearest, for example, without rolling, wich seems reasonable for game purposes (not from a RPG point of view, they go straight to the hidden models at 30"), and with that info, all this problem would be solved, because you would not need to roll for detect them when moving, just when engaging to consider them engaged
Here is where I get lost After reading @Modiphius-Gavin , why could they remove your hidden token, one by one, just for selecting the nearest enemy? According to the point 2, they don’t need a potential action/attack to get the roll…
Well I think that’s the point yes; they would attempt to detect them at any range, provided that they had obscured or clear line of site (which does make logical sense if they can still see and react to enemies, even if they’re too far away to do anything directly).
They still have to actually detect them and, as Gavin says, if the Hidden enemy is behind even Height 1 cover, they’ll still be in blocked line of sight so the adversary wouldn’t actually be able to attempt even that.
. Then, why 
What should they do if they cannot roll or they cannot remove the hidden tokens? do they block/hidden and stand there, don’t they move with the M response?