Game Designers Answer your Hardcore Questions

Game Designers Answer your Hardcore Questions

February 13, 2020 62 By Sebastian Fry


Due to a technical error the introductions for this talk were not recorded, but here are the speakers: At the computer here he’s running a 3.9.0
build of Path of Exile. Mark. Yeah? We have some new skills and supports coming
in 3.9.0. We do? Yes. One of them I believe is called Volley Arrow. No. Yes. Wrong. Nova Arrow. Arrow Nova. Arrow Nova. We go through some names. Let’s see what that looks like. Yep. In the meantime, what do you guys think of
Arrow Nova as a name? I think it’s pretty good. It’s alright. It’s descriptive, I guess. It does do a nova. It does do a nova. Okay so the way this works, it supports any
bow skill that fires an arrow. Nice. Gratz on the ding. Okay good. Good name. If you fire an arrow at a location, it will
fall to the ground and shoot a nova of projectiles. Pretty straight forward. Hey Mark. Yep? What’s the upper limit on the number of projectiles
can have? That’s an interesting question. There’s sort of three answers to that. Obviously there’s a soft limit in that there’s
only so many projectile modifiers in the game you can accumulate. Yeah. There is also a hard upper limit, in that at
around 2.2 billion. That’s more than we can store in an integer. Okay, yep. So we’re between one and 2.2 billion. But the medium limit would be somewhere around
here. Someone want to throw out a number? Keep it reasonable. Who’s got a number? Sixty nine. Sixty nine, that’s a funny number. Sixty nine, let’s go with that. So we got sixty nine extra. So it’s a total of seventy four arrows from
my Frenzy. What’s that look like? I mean. That’s pretty good. That seems good. Hey, what if we double it? Do some math, real quick. Yeah, we’re going to test Mark. Can we add a zero? I think we can do better. Yeah lets add a zero there. Let’s just, go all the way. Wait. Wait. We’ll get there. Wait. Just gotta believe. I don’t believe. It’s not happening. No, it’s a disaster. Sorry guys, talk’s over. This would be the effective limit. Alright alright. While Mark figures out what just happened. Let’s actually ask one of these questions. We’ve been getting some questions in. There have been a lot of changes to skills
and gems in general. We have progress. Oh that’s good. Oh that’s tasty. I don’t want to hear any more complaints about
slow down by the way. We had someone ask to make it crash. We might just have that happen. We’ll see what happens. Alright there have been a lot of changes to
skills and gems in general. In these changes will unarmed get any love? It’s more of a balance question. But I’m actually kind of curious. It’s definitely something every time we make
a new melee skill we consider. Should this work unarmed? Can it work unarmed? And then how broken will it be with unarmed
with Facebreaker’s damage multiplier specifically. We’re often thinking about it, but it is a
little like, “This is borderline too powerful. Let’s put it in. That’s fine Facebreaker needs some love”. And other times it’s like, “Yeah they’re doing okay at the moment. Let’s make this added physical damage weapon
only”. Or something like that. That’s reasonable. That seems very reasonable. Okay here’s an actual mechanical question. Dread Banner gives 2% fortify effect per stack,
testing showed the banner needed to be placed. Does it buff fortify effect if fortify cannot
be applied because you already have fortify? So I guess the question is. Does this snapshot fortify? Hey Mark. Does this snapshot fortify? Fortify’s effect is determined at the time
you gain the actual fortify buff. Fortify is effectively always snap shotted. It just rarely matters. There you go. Snap shotting confirmed still in the game. It lasts four seconds Nick. Here’s a great one. So metamorph is coming up. This might be a little spoilery but we’ll
find out. Are the bonuses to life, damage, item quantity
etc from the metamorphs you create in addition to the stats a monster gains from
being a rare? Are they all factored in? And how do they interact with map mods? The damage, life, rarity, quantity and there’s an extra stat there
that is multiplicative with the damage. So it will be X percent more rarity, quantity,
damage, life. Entirely multiplicative yeah. So run your hard maps. Fight hard metamorphs. Get mad loot. That’s a good tag line, let’s go with that. This is more about the internals of GGG as
a company but hey we’re hurting for questions. Path of Exile is one of the most complicated
games. It feels like about seven plus games in one. Does one game designer work on a single section
like skill trees or bosses? Or is it all designed together? There’s usually sort of one person coming
up with ideas, they take it to the team. Or three or four balance experts and we build
on it and it grows. Especially with uniques. It gets run through QA, and stuff gets changed
and we find a million bugs. Then we have to redesign something. Or we come up with a perfect design and then
we take it to Mark1 and he’s like, “No”. It can’t be done. Hey Mark. What’s an example of something recently, probably something I’ve suggested, that we couldn’t do for technical reasons? One of the ascendancies we revealed today. Which one? Stuff to do with accuracy. Aaah. Spell accuracy. Not sure if you heard it. Sorry. That’s new to me. For background. Designers come up with insane ideas. And then programmers have to implement them. And sometimes it can’t be done. Sometimes the ideas conflict with other crazy
ideas which we’ve already put in the game last time. And you can’t do both because they
require things to happen in different orders. The implementation concept of spell accuracy
is what we’re talking about here. Mark possibly doesn’t know about that at all. I haven’t heard of that at all. Obviously quite. There is a little bit of fear when even bringing
stuff because it’s just there’s no way. We have to, there is almost like a skill in
being able to sell it. Get everyone excited, it’s what you should
do anyway right? You should get the team hyped for it. And
it helps for them to know, programmers especially. What it’s for, where it’s going to be used,
what kind of players will be using it. If it’s several hundred, however many hours
of implementation. For something that’s going to be used by one player ever. It’s hard to convince a programmer to give
up the next several weeks of their life implementing this thing. So you know, in the case of an ascendancy
class it’s obviously going to be seen and used by a lot of people. So it’s worth going the extra mile. But that is a good example of a system we
have yet to implement. But we obviously believe we will. Hey Mark, good luck with that. Varunastra was an example, I mean it was obviously
quite unique. It wasn’t unused of course. But I think it touched something like thirteen
thousand lines of code. It took a programmer a very long time to get
through all that. It still causes headaches. Yeah and that’s not including the testing
time of someone checking all of that. To go through and check every single thing
was… It was hard to, even though its a cool and
popular unique it’s really hard to justify doing something like that. So yea again, really gotta believe in it. It’s also really easy to just say no to a
good idea because it’s going to be a nightmare. Well not even necessarily be a guaranteed
point of actually successfully implementing it. Also sometimes it can be very constructive. Like we created instant skills for Steelskin
specifically but then it started spreading through so many other skills and it’s improved
the game a lot because of it. And we can do entirely new designs because
of that single addition. There are often ideas where there needs just
to be a critical amount of use that we have to hit before we make it work. Otherwise it’s just not worth working on. I think instant is probably a wonderful example
of that. Here’s a good question. Damage conversion order. Is this something hard coded or is it an assigned
limit to prevent looping conversion and would this be looked at being reworked? So basically. What up with damage conversion? We have talked about reworking it and we’ve
come up with a few different patterns and orders and ways it can work. Every single time we look at it and been like,
“This breaks. Every one of these examples breaks so many
things”. It’s like, “Throw away everything you know
about how you build characters at the moment And make something worse in other ways”. It’s an order that we’ve picked but the implementation
very hard coded. The fundamental rules we currently have for
our damage conversion, well for our conversion system. It’s not limited to damage. All conversion of the game follows the same
rules. We came up with these through trial and error. We’ve been through a couple of different systems. First that when you convert thing X to thing
Y. That happens first and then it can be affected
by modifiers which would apply to X or which would apply to Y. So you get both sets. If you convert your physical damage to fire,
it’s still affected by physical modifiers as well as fire modifiers. And the second thing is you can if you have
two conversions. From like X to Y and Y to Z, it goes in sequence. And because of those we need to avoid loops
because otherwise it would continue to try and go in sequence. And depending on how much you’re converting
each stage you can get into having to solve infinite series and limits. And that is not something we can do in reasonable
time to keep the game not chugging like it was before. When we implemented the current system where
we changed… Originally you didn’t get the modifiers from
the original type. Infernal Blow, the original version, the converted
damage wasn’t affected by physical modifiers. And that it turns out was quite bad at the
time. Because the tree was just covered in physical
only modifiers for all your melee stuff. It wasn’t good. And so when we added the new system we had
to define that there would be an order. Just because if there is the capability for loops,
the system has to handle those loops and those calculations And everything becomes a thousand
times more complicated. Yep. We’ve also got so many different damage stats
now that just doing a single damage hit is still expensive. Because it’s got a lot of
things to look at. So adding more steps is not something we want
to do lightly. The basic damage calculation is now at twenty
two thousand lines of code. I measured this within the last month. So it’s probably only a couple of thousand
higher than that now. And that’s every time you do any damage, any
hit to a target, right? Not quite. That’s actually just when you change your
stats. Like if you look at your skills, like my frenzy
here says it deals 4-10 physical damage. Because the calculation has happened. It’s not being calculated every time I hit
a thing. It’s calculated when something happens which
would change that stat. And now when I hit things it just looks up the calculated values. But those can change quite frequently. And we still have to be able to do those calculations when we need to, very efficiently and fast. Here’s a completely different question. This is about Metamorph. Are there plans to make Metamorph combinations
like oils? Or do we have to rely on data miners again? Also general thoughts on making numbers like
monster hp/damage public. This is interesting because Metamorph doesn’t
have combinations. Why don’t we go into a little bit about how
Metamorph works. So within a given zone, say The Coast. You can only combine the monsters within that
instance. So could be Fire Fury plus Cannibal plus a
zombie plus a Sand Spitter and another cannibal or whatever the case is. Can we just get the big screen taken down
for now while I deal with this. It’s fine. Why don’t we get the url for submitting questions
up. There it is. So within that single instance you can put
together a boss that is of those monsters. And ultimately it’s going to turn into a zombie. Do a melee attack because you can’t… In the case of monsters that just have a melee
or a basic spammable. It’s often the case that is going to be used in a particular slot. So this is the eye, I guess. It will be able to use that skill with no
cooldown. And then you can put, say a cannibal with
Throw Fire in a different slot. And it will give it a modified version of the Torch Throw mortar skill. Could be you know some sort of carpeting the
area in a shape like that. It could be just a tracking throw several
times. It could be one that modifies the burning
ground to last a lot longer is it’s more of an area denial. And that can go in another slot. And Fire Fury can go in a third slot. And she will even have modified versions of
Fire Storm. So again it could be that you put it… like
the slot is generally associated with the type of mechanic. So for example the brain slot is for area
denial. So it will modify one of Fire Fury’s skills
to be an area denial like boss ability. And in some cases with bosses it’s just a
one to one copy. Because some more advanced bosses already have area denials built in. So you’ve got an area denial Fire Fury or
cannibal. Let’s say you’ve got a long cooldown kind
of epic skill for a cannibal. You’ve got the spammable is the melee zombie. You’ve got a short cooldown Sand Spitter
snipe maybe. And you’ve got another tracking Barrage-like
skill that’s Sand Spitter as well. So you put those five slots in and you can
see the names of the skills. Just for clarification each slot is basically
masked as a body part. So all you actually need is an eye, a lung,
a brain etc etc. The game handles the rest. Yes. So you’ve got your five slots there. You see the power bar. You see how much of a modifier they have. Of course going back a little bit we don’t
have the base life exposed here. Because we don’t and we’ll get to that soon. But you put your five in. You see how much of a modifier it is. You know if it’s going to be rare. As you’re inserting them you know if it’s
a rare monster, you if it’s a magic monster. And you know what mods it has. You also know what rewards it grants. And this is the thing we’ve done with a lot
of leagues lately where you just a kind of weapon type reward. Or a currency type reward. Or a Breach type reward. So those are also bound to the monsters. It is a really a build-your-own-boss and
sort of build-your-own-reward experience. And then of course you summon it and it’s
gonna rotate between those skills. You will start to get familiarity of the skill
names and what they kind of mean. But you know what you’re inputting. You know what skills they do and they’re only
just doing advanced version of that. So it shouldn’t be anything unfamiliar. If you die, it’s your own fault basically. Yeah a little, yeah. Unless that happens. The crash. In which case, sorry. Blame Mark1. So, as for endgame. It doesn’t matter what specific composition
you put together. There’s not some secret hidden thing. You know the rewards you’re getting. You know
the skills they do. You can see the power bar. You can see the stat modifiers. It is a, what you see is what you get. Except for what you will learn is how the
skills are modified. You don’t know if it’s gonna be a barrage
or tracking barrage until you’ve experienced it once. Hopefully the name helps indicate to some
degree what the behavior will be. But to truly get to the crux of this question. No, hopefully you won’t need a wiki. We’ve learned our lesson. We want things to be clear and predictable
to a degree. It will say what kind of reward you get. You’ll have some indication of the difficulty
etc etc. There are certain mechanics you want to hide,
you want to have a discovery to. This is not one of them. Not on that level at least. Life for example, we don’t expose that because
we don’t want you to focus on the number that’s visible. Right? Yeah. That’s a tricky one, I mean you could ask. Again no good answer for this. But why not expose it I mean if you ask me
that question. Okay Mark. Why not expose it? Yea, that’s me asking myself. I don’t have a great answer. But then you could also say Why do I expose it? I also don’t feel I need it. Maybe that’s not the case with everyone. Yeah of course the more information the more
helpful. So again, that’s a non-answer. We do have to deal with party play. Where life can be dynamic. You can approach a monster and it would say
one value. And then you would hit it and you’ve got a
party member that’s just entered the instance and it would be a different value. Only in the cases where we don’t snapshot
it. Which is quite rare but we’re gonna more
into making that not rare. Well, we’ve done with with Shaper for example. I think we did it with Shaper. Elder, Elder Guardians, if a party member
zones in it does actively update their life. Where as a lot of existing bosses, well most
existing bosses in content. The life is fixed unless you know, put it
to sleep or run far enough away or leave the instance and come back. Then it will update to the number of players. Alright, I’ve got another mechanical question. Does PoE have a tick rate? Does attack/cast time round to the nearest
tick? Mark1. Let’s talk about server ticks. So both the server and the client are processing
game frames. Time is effectively split into little frame
time chunks. Right. So frames per second on your client is the
number of times visuals refresh. Right? So the client actually has two different concepts
in that there are rendering frames. Which are when it’s putting stuff to your
screen. And the gameplay frames can actually be separate
from that. Which leads to some complicated stuff where
certain things need to be done on the timing of rendering frames so that you actually see
them moving smoothly. Whereas other stuff needs to happen in the
general gameplay stuff. Which is the one which is inline more with the server. Now I’m pretty confident from memory that
the server is doing 30 game frames per second. I think that’s right. It’s really not something which matters to
sort of ninety nine percent of players. Time appears continuous. But what happens when you get too much attack
or cast speed? Now there technically, I don’t know if you
can reach it. There might be a thing. If you are casting a spell. And in processing the frame it says this much
time has passed. And that brings you to the end of your animation
within that time. If you are, you know you’re holding down a
button to do a spell. The game does understand you have finished
this animation. It knows you’re going to start the new one. It applies the remaining amount of that frame’s
processing to the new animation. So that you’re not chugging along. Like if you’ve got the very start of a server
frame. You finish your animation. It will start the new one immediately and
spend the rest of the frame advancing through that animation. So while to a certain extent events within
a frame are happening the same time. For a lot of game purposes. If finishing one animation and starting the
new one does smoothly go through without wasting the rest of the frame. If you had a cast speed where you were casting
the spell more than thirty times in a second. Like casting more than one spell per server
frame. I suspect that would just not work. I haven’t tried. We probably shouldn’t try it on the big screen. So let’s avoid that one. Last I tested, if you did exceed… As you start getting into those extreme
numbers of attacks and casts per second. You do start losing it. It seemed to not have so much of a hard cap. As much as it starts to kind of diminish away. That was quite a while ago though. Just improving server performance and all
that will generally improve how much of that is lost, I guess. This might be a jump down a rabbit hole. Are there any plans to change how percentage
gained as extra chaos damage interacts with elemental conversions? Will it continue to gain chaos damage from
each conversion step for the foreseeable future? Before we get to the answer to that balance
question. Let’s explore what that actually means. So Mark1, how does percentage gained as extra
chaos damage interact with conversion? It is conversion. Okay. Fundamentally what that stat is doing is it’s
giving you… a percentage of thing X, gained as thing Y
is exactly the same mechanic as a percentage of X converted to Y. Except that you don’t lose the X. They’re the same thing except that you change it into the new stuff and have the old stuff. You can have your cake and have eaten the
cake. Or I guess turned the cake into a muffin. How would you ever go in that direction? Who wants..? Stick with the cake. That’s the question we’re asking now is it? Fundamentally, the stats which give you
elemental damage or non chaos damage as extra chaos damage. That’s a one stat label for individual
stats saying for each damage type, of which there are five because darkness doesn’t
really count. Darkness of the Azurite mine variety, correct? Yeah. That’s a secret, not really damage type damage
type by the way. Only exists for damage over time. You can’t deal hits of darkness damage. There you go. So don’t even try. For each of the applicable damage types,
you have a stat which converts some of that damage type to chaos. And as we established if conversions can chain
they will do so. So at the point you’re converting from physical, it says, well, that’s not chaos so some of it
you get as chaos. And then maybe you also convert some physical
to fire and then if you’re converting some of the fire to cold. No you can’t go that way. Physical to cold, cold to fire. Yep. Then at each of those steps it’s saying, well,
the thing we are converting from is non chaos. And this thing says that some percentage of
that goes to the chaos damage. And because of those rules which I mentioned
before where they have to happen. They’re independent and in sequence. And because of that and the way they’re calculated
you can’t… There isn’t a difference between the ones
from the non chaos stat. Which is sort of labeling multiple individual
stats which really exist there and other conversions. So Rory. What’s the end result? It’s too strong. It allows you if you’re doing conversion then
conversion then conversion through multiple elements and then have this non chaos added as chaos, then that 20% could be applied three or four
times from the original physical damage amount. And the end result ends up being a lot bigger
than 20%. Which is very unintuitive. Yes. It’s an interesting mechanic. It is a very interesting mechanic. But you have to know multiple layers of other
mechanics to really take advantage of it. Yes and there’s a very specific, few specific
ways to take advantage of it. And if those are the way to get the most damage out
of certain weapons and certain mechanics, then it gets boring after a bit if
you have to keep doing that. We’ve talked about ways to change how this
works. But nothing that’s made Mark1 happy with our
suggestions. He is the gatekeeper of conversions. There have been certain plans which were definitely
going to fix that and were going to work. Which you didn’t like as well. Oh yea, I mean obviously everything has some
balance consequences as well. Like, “Oh this opens up a new horrible worm
hole”. Remember that question earlier about whether
or not one person works on a thing. There’s your answer. Just a bit more real quick. We’ve discussed several times about changing
that. As was kind of explained. We have had in fact several… I cannot remember any I was trying to think
of it for that whole time. We’ve had several things where it’s like,
“We can do this thing. But only if we rework or change the conversion
system”. It’s just come up and come up and come up. And this has been part of it. I don’t think we’ve had a three month cycle
without this coming up as a balance concern internally, obviously externally. We’ll get there. We will get there. I can’t say when again, I can’t even say how
based on again this discussion. Argue about it now. No, no I’m kidding. Yeah we’ll figure something out. I’ve actually had multiple questions. About “nearby”. We’ll all give you different answers. I’ll tell you that. Is “nearby” going to be standardized? No. Maybe. Yes. No, no, no, no. Okay, now why? The reason for your answer of why is “nearby” fundamentally is modifiable. It’s affected by, “nearby” just indicates that
something has a range. It’s used to say. Because we can’t just say, this affects all
enemies. When there is a range on it. “Nearby” is a thing to say “This is affected
by range”. And in huge numbers of those cases that range
is entirely modifiable by the player in huge ways. Like, it can’t imply a specific range because… It isn’t a specific range. Yes. It is fundamentally the term was added
to the game to describe a thing which is wildly variable. I guess that second point to that is why,
if we don’t already and in every case show what the value is then. That’s probably something we… That one comes back to what we were discussing
before, of judging the worth against time. Because that is something we would like to
do. There has been discussion of how we would
be able to do it. And what is necessary. And it comes down to quite a lot of work. And at some point maybe we’ll find something that
really makes that feel justified. But at the moment the seasaw hasn’t quite
tilted the other way on that yet. We haven’t hit that breaking point yet. Yeah. Are there any plans to streamline descriptors
such as “You” to properly reflect how they function? Currently some usages of e.g. the “You” descriptor
includes totems and minions, while others do not. I guess the question is Sometimes the you is your totem. Sometimes it isn’t. Which one is the true you? So, this is a thing which Rory and I have discussed. Although I don’t know how well he’ll remember it, because it was at least two years ago when
we last had a big discussion. We had big plans to do a full review of this. And there still has not been time for that. So, they’re still looking for time to figure
out who “you” are. We are trying to incrementally change things
as we find them to make them a bit more standard. Unfortunately, there are multiple pressures
on how we describe things in Path of Exile. Obviously, clarity is one and I always try
to emphasize clarity. Not the aura. No. The aura is bad. Don’t use it. Just to be clear. Yeah, just to have clarity. Yeah. But there is also, like, we do try to keep them
short. Particularly if they’re going to break into
multiple lines. That makes them much more confusing for people
to read. It causes problems. And so, some things which perhaps should say
like, “Your skills” just say “you” because we need that to
fit into a line. Mark1 and I have regularly had arguments about
how to describe mods that get heated. Many people in the Grinding Gear office are
very passionate about how we communicate these effects to our players. We care a lot. No one has died in the ensuing fights yet. But it’s only a matter of time. I think we all acknowledge it’s not perfect,
but I could be a lot worse. Yes. It goes through a lot of rounds of figuring
out. I can say that recently I did manage to successfully get a push the two effects which just described “You”, but also included minions which aren’t even using your skills like totems are. Now explicitly say “You” or “Your minions”. That’s Headhunter and I don’t remember
the other one. Well. Maybe we’ll find out. Mark2? I feel like it was a Blight unique. Oh, yes, it was one which creates the blighted
spore thing on kill. I don’t remember what it’s called. The Blight totem looking thing. It works for minions? Okay cool. Cool. Oh the mushroom, the spore? Yea the mushroomy thing. This is a big problem in development. Where nothing is named until the last thirty
minutes. And by the way. It’s not a spore. That’s the fruiting body. I agreed with you when you pointed that out
but the name was set in stone by that point. The amount of times I’ve had to explain fungal
biology! Ugh, okay. What are some of your favorite builds that
were unintended? This is not really mechanical but it kind
of gets to some interesting mechanical interactions that sometimes we don’t predict. Oh, I would have to say the first version of
the full autocaster. It used Cast When Damage Taken to cast a bunch
of spells including Summon skeletons. It used a unique jewel to make them have a
very low duration so they basically immediately died. Heartbound Loop caused you to take damage
when they died. And that triggered cast when damage taken
again it was all… They had just enough duration for the cooldowns. And it would keep autocasting things and
you could just walk around. He says it was unintended. It was intended. I remember seeing the Reddit thread about
that and all the people posting in that, saying “Oh god don’t let GGG see it, it will be nerfed.” “We must protect this. This is precious.” QA was sending me GIFs of endless Raging Spirits
and going “Is this fine?” And I was like, “Yea, this is really good”. We love seeing people coming up with crazy,
ridiculous, weird ways to combine the little pieces we’ve given them. And sometimes they
do things we just could not have guessed. I think my answer is every Writhing Jar build. That flask is my baby. Wormblaster needs to die. The number of times that these guys have tried
to kill Writhing Jar, and I’m like, “No. It’s delicious, you’ll never take it away”. How many servers have died at the hands of
Wormblaster? I do think that unique added the singular
most amount of capacity to crash instances. Especially in your hideout for a good time
there. I feel like every day you’d wake up and just
check your phone or get in and there it’s just like, “There’s some new Writhing Jar exploit or crash”. You’re just jealous because you didn’t design
it. Maybe. I’m pretty sure. I think there have been at least fifteen different
things I’ve been told to nerf because they get broken by Wormblaster. And every time I turn to Rory, he sits near
me at work. And I say, “When are we accepting that the flask is the
actual problem?” The answer is never. Not on my watch. How much further can you take monster AI? And as players have become better at the game, do you feel that reworking the behavior of monsters is a way to maintain the difficulty
and intrigue of combat? I’m just going to jump in here. I think Path of Exile 2 is a perfect example
of the new wave of monster AI. And taking advantage of interesting skills
and positioning and all sorts of interesting stuff. There’s definitely a lot of room for improvement. We did an experiment to be fair with Warbands. I know that was a long time ago. It was kind of a different era. We tested a lot more complex AI with that,
that just didn’t work. I felt it just wasn’t observable. I felt that just most people didn’t see what
we had done. There was behavior there that, like, the higher
ranks behave differently if there are lower ranks with them in certain proximities. They have different behaviors of when to advance
and when not. Maybe more people observed it than I thought
but I was pretty sure they just had no idea. And then we were kind of like, “So okay over
complicating it in that regard, just, what’s the point?” It didn’t work”. I mean if everyone wants to tell me that I
am wrong and it was a far more interesting AI experience. So be it. But I don’t believe so. We kind of learnt that lesson at the very
least. And now, I’d say the one thing probably holding
us back, and we can fix this, is the AI is actually quite expensive. Or the capacity to… On a server processing level. Yes. Not just like… I charge extra for all AI skill sets. We’re not selling monster behaviors. Sometimes we have on accident. There was a few cases. The Celestial Raise Zombie I believe had as
an example. It’s not so much AI, but I believe it’s slam
had a different cooldown than the non-Celestial. It was worse. You were paying to lose at that point. Pay to lose. I think also we did a big AI change, not the
most recent one but before that to Summon Raging Spirits AI. The Ghostflame Raging Spirit never got the
change. So if you were using that one, you had the
old AI. So pay to be dumber? That’s arguable I guess. It may have been a downgrade as far as some
people were concerned. It may have been an upgrade. It’s very easy to make AI of monsters just… I don’t know if anyone remembers the bug with
if you were ever in a map with the regurgitated monsters. And it would just start performing
terribly. Yep, you know it. We got someone nodding. That was just the way it works is we have
these conditions which the monster performs certain actions. That was just two conditions in the wrong
order and it just caused it to become exponentially more expensive per object near it. When I say objects I mean monsters, players,
items, skills, like just any form of entity really. And it became so expensive that the moment
that you got within about one hundred and fifty units of them you could just see. If you were on lockstep you would just start
lagging. The instance isn’t handling it at all. It’s a bit risky. It’s a bit dangerous. We need to make a lot of it more efficient. We have plans to do that. And in the process of course and again with Path of
Exile 2. We’ll be expanding on that and then we’ll
apply a lot of it to what we can with any new content adding to Path of Exile 1 between those points as well. There is a point at which you don’t really
get much out of AI. Like, AI is best spent on boss encounters, right? A normal monster that dies in a second and a half. It doesn’t matter if it’s got tons of great AI. Because you’re probably not gonna see it. It’s about considering where we use it. Which it always come back in GGG to how best
to spend the resources. Which is man hours, QA time. Yea I feel a bit bad. A lot of these things is like, “Yea we know
about it, we’re gonna do it. I can’t give a time”. It feels bad to give these answers but the
reality is. That is the reality. And then it’s like, “Okay that would get
in the way of”… I mean it’s not like we’re not doing new
stuff as hopefully was you know, shown today. But there is an actual, we have to decide
whether or not we’re going to do A or B. We feel like we made the right decision. So sometimes it does seem like we’re neglecting
a certain thing. It’s not… We’re not happy about it. And you know the whole… It could be as simple as ways as mechanics
are described to players. Or the ways the mechanics work. Yea, we’ll get around to it all again. Which isn’t the best answer. But…yeah. Is there a concern? Actually, no, no, this is a good one for Rory
specifically. Oh. Did you intend Herald of Agony to be played
as a stand alone skill? As in the six-link Herald of Agony Guardian? When I originally built it the whole idea
was that… Rory designed the skill. Yea. Take it away. Originally it was supposed to be alongside
this. It’s powerful enough to support you. It’s powerful enough to be worth using unlike
a golem. Which you’ve got to like fully focus on if
you want to make the most of it. This is supposed to just something you can
just build on the thing. That was the plan. It was a great plan. However… What happened? The very first time I just passed it to someone
from QA who’s got some balance expertise. And I’m like, “Okay. Make a build of this and do what you can with it”. “Okay I’ve got this Marauder with 9k life.” And I’m like, “Oh, how are you doing damage”? And he’s like, “I’m not, but I’ve got this
amazing minion doing all the work for me”. And I’m like, “Or you could just lose a
couple of thousand life. And you could be doing as much damage as the minion. And your clear speed would double”. And hes like, “But I can’t die”. QA plays hardcore. So at that point I realised what a mistake
I’d made. No, it was, “What a great design I’d accidentally created”. But it was honestly, everyone was having so
much fun with it. It’s a cool skill. I like it. We nerfed it a bit before release. We were like, “Na this is too good to give
stuff for free”. And then we nerfed it a bit after release. Just for good measure. Yea, yea. Here’s one. It just says: ‘Porcupines. On death mechanics that bypass defensive layers
like curses, blind etc. Thoughts?’ Yeah. That’s pretty cool. Yea neat. How else are we going to kill players? No, it’s an interesting one. Because this also comes to signalling. Some defenses obviously work against some
types of damage. Armour will work against physical damage. So a skill has to look like it’s physical
otherwise. In order to make sense to be mitigated. But sometimes, we goof a little bit in the
signalling. Or at least early on. For example, use a monster skill that looks
like an attack. But is a spell. Or vice versa. And as a result you would get weird interactions. I think, correct me if I’m wrong. But the Porcupine on death thing. Which looks like an attack. It is not an attack. It is. It is an attack yes. It’s more just the… okay, sure, the curse is
obviously the monster’s dead. The curses may apply? I believe there is a unique item now that
allows your curses to persist on dead things. In which case it would still be affecting
the Porcupine. So it comes down to the monster being dead
at the point with the damage. Because it dies and then it releases projectiles. So by the time it hits you, its definitely
finished dying. It’s a slightly outdated design now, I would say. But if you, the unique is not intended for
having curses apply to on death effects. But it will in theory do that. It is one way. I mean personally yes. It’s not the way I would design a mechanic
nowadays. You see a lot more of this. It’s kind of the pack version of the former
volatile. Which I have an experience with. During that freeze mine league. That was like six years ago? Yea. But it’s just a zero telegraph kinda of… You can’t dodge the projectiles reasonably,
right? I mean they’re so rapid. So fast. You can’t predict the angles. Yea, you’re going to take a bunch of damage. Sometimes that’s okay. It’s okay to force damage on the player. These probably would now with the amount of
danger they have. Would probably be designed in such a way there
is a little bit more of a telegraphing notice. Before the damage occurs. It’s very hard to take a monster that is successfully killing players. And I say successfully. I measure success, my success. On the amount of players I can kill. In game, right? And the amount of ruckus I can produce. Those are my two. Kind of, not really. It’s really hard to take something that’s… succeeding in terms of its job as a monster
and to then you know, nerf it. I mean a change to make them do something then omit
it would effectively render them quite harmless. To most players at least, quite harmless. We’ll move onto another question. What mechanic have you always wanted to implement,
that you were unable to due to engine limitations? At least two people on this stage have worked on supporter unique items over the years. They often come up with the most outlandish mechanics. No offence to any supporters. Often in a good way. Sometimes in a less good way. One I remember in particular was… I admit it’s a cool idea but also completely nuts. A totem that was chained to your character
would follow you around as you ran. It was a pair of boots. Pair of boots and the totem was chained to
your ankle. And it would follow you around. And I remember there was like something if you whirling blades, the chain wraps around you and you gain the suffocating debuff. Yeah, I do remember this. It’s quite a funny image. It’s just not trivial. I think there are engine limitations there. Have you guys got any mechanics that you’ve
craved? Original Blade Vortex, where each of the blades
had their own physical contact. Until of course it destroyed. Flashbacks. I can’t think of anything specific on my front. There’s definitely stuff that comes to mind
but I just… There’s a skill that will probably never see
the light of day. It was called Vortex Mine. And what did that do Rory? I don’t remember. It was a slowing effect. It was around the same time we were experimenting
with the push flask. And then we had, it just felt terrible. It didn’t feel good to use. It sucked minions into it, right? It had actual reverse knockback. We had this kind of pulling and pushing
stuff going around. So the mine would pull monsters to the
centre, and the closer to the centre the slower they were. It just, I think the main reason honestly
it would just induce ridiculous amounts of desync. I can’t even recall if it was pre-lockstep. I’m not sure. I think that was pre-lockstep. Or at least it was pre- us releasing lockstep. But maybe while it was being worked on internally
in secret. Even now, just recalculating patching for
every monster as they’re slowly pulled towards a point and so overlapping with each other. Related to this to a degree is also related
to a previous talk. Was totem reflect at all related to an earlier
discussion, we try not to remove builds we might put mechanics back as uniques. Why does totem reflect make almost no sound
now? That one might just be a bug. But the moving old ascendancies onto uniques
is a thing we have done. Also reverse knockback totem. The rat totem builds. Awesome. I love ’em. That is one unintended mechanic I think that
made a fantastic build. I certainly didn’t see it coming. And I designed the gloves in that build. I believe, before we did that change we did
know we were destroying it. And we did know we were going to bring it
back. We were at a point where we were deciding
whether or not to undo what we had done. We’re looking at the pros and cons
of what we’ve changed. And we believe that it was better to stick with what we had. And release it on the unique. Which we did. I don’t know what the sound thing was. Sounds like a bug. Probably a bug. There is some sound bugs that we’ve found
in all the PoE 2 stuff we’ll fix up on live. I mean there’s hundreds of gems and items. There’s gonna be bugs. That’s just how software works. One last question because we’re almost out
of time. Will Metamorphs that are using map boss abilities
still include the invulnerability phases that some have? For example, Daresso. No. No, we select what skills they have. In particular, it’s those kind of immunity
phases aren’t really a skill. They’re a phase. The Metamorph isn’t inheriting the phase as
much as it is the skills that it does during some phases. Maybe on some extreme rare occasions for minuscule
amounts of time. But Metamorph bosses should have no immunity
phases at all. We’ve got one last question. I love this one. When you flicker, does “nearby” update before
or after the hit? Before. Before. In fact, we had to add extra code to Flicker
Strike so that it… It has to both defer its damage prediction
until after the teleport. Then it also manually tells any of your auras to update which things they’re currently affecting before it starts processing the damage. Fantastic. So that was not true for quite a while. But I think early in the 3.7.0 patch cycle
we added that. Cool. There you go. So before. We’re pretty much out of time. This is the last panel of day one of ExileCon. And I want to thank you all for coming here. I hope you’ve had a good time. I hope you’ve had a chance to play Path of
Exile 2. I want to thank Mark, Mark and Rory. Thank you for the great questions Nick. Thank you guys for the great questions. I didn’t realise they asked them. Thanks those guys. We’ll see you back here tomorrow. Thanks guys. Thank you. Thanks.