Engagement Mechanics- Problems and Solutions

[Sensuki]

I enjoy moving in combat, but it sucks in Pillars of Eternity, standing there and doing raw damage is always superior. The Chanter push back spell is a joke compared to the other good ones they can cast IMO.

[Shevek]

It depends. I pause, switch to disengage song, and do the push back then move. Can be useful. Cipher also has a multihit push back ability (cant get it to work right, must admit). I think fighter gets an aoe knockdown at higher levels too.

[Jon of the Wired]

That's right, you can. However since prone, stunned and paralyzed units can't fight back and get huge penalties to defenses ... why on earth would you stop attacking them?

 

Retreating isn't the only reason to disengage, though, you may also be trying to get to a squishier target.

[Sensuki]

I have over 120 hours in this game, and I only speak from practical examples. Maps are small, pathing space is usually limited and the current enemy AI usually target the same unit, so they are often if not always close together. Encounters don't go for very long and AoE disables are usually cast by Priests, Druids and Wizards from the back and AoE disable spells have small AoEs.

 

One of the only encounters with "squishies at the back" is the Dragon Egg fight and I have never seen the use of anything remotely to do with disengaging in that fight.

[Shevek]

Well, the beta is a tiny slice of the game and the encounter dynamics shift wildly as they mess with AI. I have used disengaging tactics to both retreat and to go after ranged opponents fairly often in the final dungeon but also sparingly before then. I notice that I am using them more now since the AI seems a bit better. Ive played like 70 hours but mostly in the last 2 patches.

[J.E. Sawyer]

In the next BB update, you should see the following changes:

 

* Defender no longer increases Engagement range.  It was the only ability in the game that allowed an increased Engagement range, so with this change, all Engagement ranges will be the same.

 

* When a character breaks Engagement from a defender, they go on a list for that defender of "recently Engaged".  The defender is prevented from re-Engaging that target for a constant amount of time.  We will adjust this constant until it feels appropriate.

 

Not quite related, but connected to an issue that Sensuki highlighted: AIs were not updating their targeting preferences (of any sort) while moving, which made kiting extremely easy.  In the next BB update, moving AIs will periodically re-evaluate targeting preferences within a narrow range of time.  We will tune that time range as necessary and, if other common exploits become obvious, we will consider how to address them.  However, a micro-oriented player with a party of high movement characters devoted to kiting presents, if not an unsolvable problem, one where investing serious programming time to it has diminishing returns.

[IndiraLightfoot]

Out of sheer curiosity, since I'm playing DA:I right now, will there be more monsters in PoE like the beetles, with teleport-like abilities?

I mean, since that certainly puts the pressure up on any glass cannons or poorly armoured party members hunkering in the back row.

Edited December 5, 2014 by IndiraLightfoot

[J.E. Sawyer]

There are more monsters than just the stone beetles that have teleport abilities, but not a ton of them.  In lieu of teleporting creatures, ranged enemies often target spellcasters or low DT/low-Deflection characters above all other enemies (according to QA it's a little too nasty right now in some spots).

[Namutree]

In the next BB update, you should see the following changes:

 

* Defender no longer increases Engagement range.  It was the only ability in the game that allowed an increased Engagement range, so with this change, all Engagement ranges will be the same.

 

* When a character breaks Engagement from a defender, they go on a list for that defender of "recently Engaged".  The defender is prevented from re-Engaging that target for a constant amount of time.  We will adjust this constant until it feels appropriate.

 

Not quite related, but connected to an issue that Sensuki highlighted: AIs were not updating their targeting preferences (of any sort) while moving, which made kiting extremely easy.  In the next BB update, moving AIs will periodically re-evaluate targeting preferences within a narrow range of time.  We will tune that time range as necessary and, if other common exploits become obvious, we will consider how to address them.  However, a micro-oriented player with a party of high movement characters devoted to kiting presents, if not an unsolvable problem, one where investing serious programming time to it has diminishing returns.

Glad to see the exploits are being dealt with.

[IndiraLightfoot]

As long as you keep it nasty enough, I'm fine with any such fine-tuning.

[Quadrone]

In lieu of teleporting creatures, ranged enemies often target spellcasters or low DT/low-Deflection characters above all other enemies (according to QA it's a little too nasty right now in some spots).

 

Haha, yes. If you do the ol' naked Mage/Rogue in Lle a Remen the group of 4(?)Spectres there can kill him in 1 - 1 1/2 volleys.

On that point, is it supposed to be that way that their projectiles follow you across any distance and through obstacles?

Edited December 5, 2014 by Quadrone

[Gairnulf]

Not quite related, but connected to an issue that Sensuki highlighted: AIs were not updating their targeting preferences (of any sort) while moving, which made kiting extremely easy.  In the next BB update, moving AIs will periodically re-evaluate targeting preferences within a narrow range of time.  We will tune that time range as necessary and, if other common exploits become obvious, we will consider how to address them.

Adding a loop for reevaluation of targeting preferences (implying there will be a number of such preferences) is what I expected would happen. To think of all the hundreds of kilobytes collectively written on how exploitable Engagement is, addressed in a couple of sentences...

[J.E. Sawyer]

Adding a loop for reevaluation of targeting preferences (implying there will be a number of such preferences) is what I expected would happen. To think of all the hundreds of kilobytes collectively written on how exploitable Engagement is, addressed in a couple of sentences...

 

AI reevaluates on events (e.g. being hit, completing an action, etc.), but a moving character that does not stop often has no new events to re-evaluate.  It starts moving and continues moving without being able to reach the target, so it keeps on truckin'.  At least, that's how it worked before.  In the next build, the main difference is that a moving AI will say, "Hey, I'm moving, so I will periodically evaluate if I should keep doing this or peel off and attack someone else."

[Gairnulf]

I remember how the IE games' characters would stop and twitch in the middle of walking, and I knew that was when tehy were re-evaluating their path, but in combat the twitching was also resulting in changing targets sometimes. I never researched the IE games' mechanics in mich detail, but I guess that was the equivalent of what we'll get now in PoE, minus the twitching Anyway, since I had see this re-evaluation happen in the IE games, I was sure it's just a matter of time for something equivalent or better to be implemented into PoE. I'm not a professional programmer, but I know that if the AI is capable of moving around characters, and of knowing engagement radius, it would certainly be able to move around characters along their engagement radius, preventing engagement when it doesn't want to get in engagement.

Edited December 5, 2014 by Gairnulf

[Void3dge]

So engagement is still in ... Thankyou Very much

[constantine]

You can abuse the Cautious Attack modal a bit, pause and toggle it on when you see an enemy making an attack animation against you, and toggle it off when the attack ends

Unacceptable for the game to be subject to abuse so easily.

[constantine]

In the next BB update, you should see the following changes:

 

* Defender no longer increases Engagement range. It was the only ability in the game that allowed an increased Engagement range, so with this change, all Engagement ranges will be the same.

 

* When a character breaks Engagement from a defender, they go on a list for that defender of "recently Engaged". The defender is prevented from re-Engaging that target for a constant amount of time. We will adjust this constant until it feels appropriate.

 

Not quite related, but connected to an issue that Sensuki highlighted: AIs were not updating their targeting preferences (of any sort) while moving, which made kiting extremely easy. In the next BB update, moving AIs will periodically re-evaluate targeting preferences within a narrow range of time. We will tune that time range as necessary and, if other common exploits become obvious, we will consider how to address them. However, a micro-oriented player with a party of high movement characters devoted to kiting presents, if not an unsolvable problem, one where investing serious programming time to it has diminishing returns.

Thank you for enlighting us on the team's work for that matter and I hope this eliminates the exploits.

 

On another matter, you didn't mention something about diseng. atks proccurence, so we can assume that engagement retains its current role- stationary melee ?

 

-Or when saying 'defender', you mean anyone in melee who has engaged a target ?

Edited December 5, 2014 by constantine

[Shevek]

@J.E. Sawyer

 

Thank you guys for all the hard work. This sounds great.

[Doppelschwert]

the content of patches seem to get better and better. so the main question now is probably when the next patch is going to arrive.

[Sensuki]

Adding a loop for reevaluation of targeting preferences (implying there will be a number of such preferences) is what I expected would happen. To think of all the hundreds of kilobytes collectively written on how exploitable Engagement is, addressed in a couple of sentences...

I don't think that completely addresses it. It should still technically be possible to get an opening round of disengagement attacks by luring melee foes to a clump of your characters, particularly with Fighters who have multiple engagements. I will try to achieve that in the next build and see how I go.

 

You saw that I was able to achieve at least one disengagement attack per-character with my Rogue and Barbarian examples, while smarter targeting may make that harder to achieve, should still be possible, and if so - will remain the best way to open combat against a group of melee opponents.

 

Good thing is, with the improvements to AI targeting, no engagement only needs some animation fixes and ability mods to be perfectly playable.

 

edit: Defender Mode will need a buff or something now, because outside of being able to exploit engagement range, Cautious Attack is way better.

Edited December 6, 2014 by Sensuki

[Hormalakh]

However, a micro-oriented player with a party of high movement characters devoted to kiting presents, if not an unsolvable problem, one where investing serious programming time to it has diminishing returns.

 

You should really look into Defense of the Ancients (dota2) and turn rate, Josh. I've mentioned it several times before, but in case it hasn't caught your attention, it should. http://dota2.gamepedia.com/Turn_rate

 

It's a very simple and elegant disincentive to high movement (at least in regards to turning) and solves the problem of "kiting" - though again not fully. It's simple and intuitive, yet covers quite a few cases of "high" movement.

Edited December 6, 2014 by Hormalakh

[Jotra]

the content of patches seem to get better and better. so the main question now is probably when the next patch is going to arrive.

 

Let's hope they delay this game again if needed. I'm not sure they got enough time to polish this in time for early 2015 release.

[Gairnulf]

I don't think that completely addresses it. It should still technically be possible to get an opening round of disengagement attacks by luring melee foes to a clump of your characters, particularly with Fighters who have multiple engagements. I will try to achieve that in the next build and see how I go.You saw that I was able to achieve at least one disengagement attack per-character with my Rogue and Barbarian examples, while smarter targeting may make that harder to achieve, should still be possible, and if so - will remain the best way to open combat against a group of melee opponents.Good thing is, with the improvements to AI targeting, no engagement only needs some animation fixes and ability mods to be perfectly playable.edit: Defender Mode will need a buff or something now, because outside of being able to exploit engagement range, Cautious Attack is way better.

The luring of melee enemies relies on them never reacquiring targets, which Josh says is a recognized problem. I think it's all down to how well the changes they make to targeting will be tested out. If loopholes in the logic remain, it will continue to be exploitable.

 

Ideally, I would prefer to see personality types for various enenmies, where different classes of enenmies have different behavior in battle, which affects how/if they reacquire targets, will they switch between weapon sets (ranged/melee or melee/melee), preferred type of enemy (melee, ranged, caster etc). If a group of beetles has the same targeting logic as a group of mercenaries that's less interesting and more predictable. Of course I realize this is quite some more work.

 

Let's hope they delay this game again if needed. I'm not sure they got enough time to polish this in time for early 2015 release.

In my first post after the last delay was announced, I said I expect it will not be ready by early 2015 and will be delayed one more time. Based on what I see now, I'd say I'm ready to call it complete if a few serious problems are fixed, but I'm sure there are many more problems which I don't see.

[Quadrone]

 

the content of patches seem to get better and better. so the main question now is probably when the next patch is going to arrive.

 

Let's hope they delay this game again if needed. I'm not sure they got enough time to polish this in time for early 2015 release.

 

 

Don't see that happening. v364 is already in a playable state, though still rough arround the edges.

With likely 4 more patches to go, if the pace remains about the same, I really can't see why you'd even call for another delay.

[GordonHalfman]

One thing I'm not sure about is why the distance criterion for disengagement is so small. Didn't it used to be the case that you had to actually disengage to proc a disengagement attack? I haven't been following the debate but is there a reason this was changed?

 

I find it annoying because it means I can't make small adjustments to my positioning in melee, for example say I want my fighter to shuffle slightly to the left to make more room in a choke-point for someone to join him on the front line, now I get a disengagement attack when I'm not really trying to disengage or kite anyone. 

 

I would also make two suggestions to tone down disengagement. One is just to make it so disengagement attacks can't crit, this would make the maximum penalty for disengaging lower and more predictable and perhaps reduce the feeling of "ok so just never move ever".

 

The other is make the dexterity stat give bonus deflection against disengagement attacks only. This seems like it would be an appropriate bonus for dexterity, like the tumble skill in DnD and would bolster an otherwise underwhelming stat a bit.