Thinking Outside the Box: 1 turn is between 1/4 and 1/2 a second

Thinking Outside the Box is my article series for concepts unlikely (if not outright impossible) to include in this game, but are worth considering for potential PoE sequels and ARPGs in general. The previous article in this series, regarding an alternative to socketing, can be found here.

Live-action games seem turnless. This is an illusion, not reality; it is similar to the illusion of video, which is in fact a series of discrete images. Live-action games simply use turns which are of the shortest length possible.

How long is this, precisely?
Average human reaction time is 264ms. Therefore, regardless of a game's design, a turn can't be less than, say, 250ms (aimed at those with slightly-better-than-average reaction time or better).

For online games or those aimed at people with slower reaction times, etimated ping is 122ms roundtrip. 264ms + 122ms = 386ms. Add processing time on both the server and client end, give a buffer for people with slower reactions, and we're probably looking at about 500ms between responses to stimuli.

What does this mean?

1. Any combination of things which kill a player from full health in 250-500ms (or less) can make players feel helpless, unless it was telegraphed in advance.

When designing a live-action online game, you need to be particularly careful about creating full-to-dead kills in under 500ms. It's okay to kill players in one hit, but only when that one hit has been "charging up" for multiple turns, and there is some realistic way out which the player can pursue in the turns before the OHKO.

Or, the player can be backed into a corner over several turns, where enemies can then burst the player down. It's vital that such situations have very clear warnings well in advance, and are avoidable if players choose to avoid them.

2. Player healing faster than 200% max health per second should not exist, and anything over 50% max health per second should be closely monitored.

Let's say monsters are designed according to Principle 1 above. If so, players will not be one-shot in 500ms or less, unless they ignore clearly telegraphed signals. This means: at most, they will be killed over two turns (roughly 1 second). Which in turn means that healing 100% max health per second would trivialize all of the content... unless the game fought back by reneging on Principle 1.

This is something we've seen occur in PoE. Principle 2 has been violated by Vaal Pact and by items such as Atziri's Acuity, which in turn has led to monster designs which attempt to challenge such builds with heavy, nontelegraphed damage.

Instant-healing flasks also play a role. An endgame player can easily have access to up to 3600 Life in instant healing using only two flask slots. Although these are restricted by a cooldown, this mechanic ensures that even in situations which kill players within a duration between 600 and 1000ms, players will usually survive, unless another such attempt occurs before flasks recharge. This also pushes GGG to use more spike damage, or at least throw multiple under-1-second kill shots at players within remarkably short timeframes.

2b. Players should not be able to log out or town portal in less than half a second, and any delay less than 2 seconds is iffy.

This is related to instant healing. If logging out is essentially healing back to full health as far as avoiding death goes, then players should not be capable of doing so at an instant speed.

The skipping of two "turns" is the bare minimum in creating a sufficient deterrent. It would mean players trying to use the method to cheat death would need to predict at least one turn ahead, and the method would fail against an obviously emminent demise.

The same penalty should apply to those using Alt+F4 or who suffer internet or power outages. At least a portion of this time would occur naturally (waiting on connection to timeout due to lack of graceful logout procedure).

Note that this assumes a relatively fast pace of play. For even slower paces, a longer penalty may be considered. (But why make the game slower paced?)

3. In a live-action game, FPS and desync issues skip multiple turns.

Let's say your game suffers a 2.5-second FPS delay in the middle of combat. An analgous experience in the turn-based digital CCG Hearthstone would be: you get disconnected and cannot reconnect until 5 to 10 turns later. Needless to say, such a huge penalty would no doubt be utterly devastating - the game would be thrown to your opponent.

The prevention of performance issues, most importantly graphics processing delays (more prominent than desync) is something the GGG staff don't seem to believe in. They should. Without a client players can count on to run smooth, GGG's intent of a fast-paced ARPG cannot truly be realized. For all but highest-end systems, the current performance levels are outrageous.

Btw I should credit SUPERSEXYSATAN for the original idea on the instant healing flasks, although I've modified the concept significantly.

There are probably additional applications of the half-second turn concept which I haven't yet touched upon.

I've always been in favor of more attrition based damage instead of spike and you are dead. Unfortunately, that is really not going to happen unless GGG are willing to undo quite a few aspects of the game. I can't see them doing it at this point.

I think the balancing that revolves around you either being at full hp or dead will continue. A shame.

You also have to keep in mind that if you remove instant healing and scale back on burst. People will just log out to avoid death. You'd need to address that as well.

"

CliveHowlitzer wrote:

I've always been in favor of more attrition based damage instead of spike and you are dead. Unfortunately, that is really not going to happen unless GGG are willing to undo quite a few aspects of the game. I can't see them doing it at this point.

I think the balancing that revolves around you either being at full hp or dead will continue. A shame.

This is why I was (and still am) considering adding this to my "Thinking Outside the Box" series of posts, which deal with ARPG design ideas this game (but not necessarily this franchise) have probably missed the boat on.

"

CliveHowlitzer wrote:

You also have to keep in mind that if you remove instant healing and scale back on burst. People will just log out to avoid death. You'd need to address that as well.

Updated OP to address this. (I knew there were applications I was forgetting.)

A move towards more attrition based damage systems would be a big improvement to the combat. That was one of the things I actually liked a lot about gauntlet dark legacy, if you played poorly, you died, but you died slowly over the course of minutes not instantly to a wtfdamagespike.

To really achieve attrition style combat, the changes would need to be pretty sweeping. I think they'd have to look something like this:
* Health flask effectiveness and/or recharge dramatically reduced.
* Visiting towns/hideouts no longer refills flasks or heals you to full.
* Damage from monsters substantially reduced or player life/ES just substantially increased.
* Leech rate cap reduced, maybe the leech base numbers too.
* Vaal pact nerfed further or outright removed.
* Atziri's acuity nerfed further or substantially redesigned.
* Life gain on hit stats dramatically reduced(or again, just player life substantially increased).
* Logging out no longer heals you to full, your last life/ES level is preserved.
* Reflect dramatically reduced(and/or player life/ES substantially increased.
* Life regen potential dramatically reduced across the board(it should be more like the flat regen on rare gear is now).
* RF burning dramatically reduced.
* ES recharge substantially reduced.
* Any other regen/degen affects adjusted(blood rage, death's oath, etc)

And then you'd have to deal with some design issues such as:
* Differentiating lgoh from life leech.
* Making sure a max regen/leech/lgoh build isn't invincible while still keeping more normal levels of investment worthwhile.
* Differentiating ES from life.

I think it would be quite interesting if done correctly, but it's probably not going to happen since they'd have to rebalance the entire progression of the game and changing the combat so fundamentally is a risky move from a business perspective.

As an aside, your point about only using one skill is a little bit of a stretch. The linking system and dps potential are the main reasons people focus on a single skill, getting locked into the previous action for an extra quarter second isn't much of a factor in comparison to those two.

"

Vesuvius079 wrote:

A move towards more attrition based damage systems would be a big improvement to the combat. That was one of the things I actually liked a lot about gauntlet dark legacy, if you played poorly, you died, but you died slowly over the course of minutes not instantly to a wtfdamagespike.

Minutes? No. No no no no no.

Seconds. As in more than 1 but less than 60.

How many turns is fair for a properly difficult, fun encounter in a more obviously turn-based RPG? Perhaps 5, 10, 20 or even 30 turns (consider how a it's almost impossible for a Hearthstone game to go past 30 turns). But hundreds? Noop. Way too many.

If a turn is half a second, then one minute is a long time. Anything past 15 seconds, without any breaks, is questionably long. Anything shorter than 3 seconds feels remakably short.

To make things perfectly clear, I have no problem with "three turn kills" in PoE. By which I mean, if a situation can kill you in 1.5 seconds, but you can deal with it in under 1 second (but hopefully longer than half a second), things are fine. Such situations shouldn't be overused, but they're not problematic until then. (That is what spike damage should look like.)

I think fast-paced combat is a good thing. Actually, it's a crucial element to a good ARPG, because turn length is very short! The point of the OP is not that a fast pace is bad, only that there is such a thing as too fast a pace, one which human reaction time fails to keep up with.

Flask refills in town, life/ES regen/recharge, and (noninstant) leech rates and logouts are not problems under a fast-but-not-too-fast model. Leech rates up to 50% max life per second are safe (full heal over 4 turns). LGOH should be "leeches x life on hit," and thus covered by leech rate.

Reflected damage follows the same rules as other monster damage. Is it clearly telegraphed well in advance? If so, it can instakill you without being unfair, provided it doesn't do so if you switch to a less multi-target skill. If not, it needs to be carefully balanced to avoid instakills (or just better telegraphed).

Lastly, my comment about skill speed and spam only applies at the fastest paces. Up to 2 APS (or arguably even 4 APS if game is not online, or arguably 2.5 APS if you make the argument that a real-time turn in an online game is 400ms), no issues with human reaction time. There are issues past that point, however. But this doesn't mean 2APS is less friendly than 1APS; they're the same in the multiple-skills regard, except for people with abnormally slow reaction times.

As long as you aren't going to ask me to make 1 frame links online, I am fine.

Scrotie, you are doing calculations in Lab Conditions.
who's to say how much "reaction time" including ping, processing etc' a real situation has, and I assure you my calculated real-world "reaction time" in Path Of Exile, is different than yours, which is different than Clive's, which is different than BMBI's...
heck, my "reaction time" right now, is different than my "reaction time" 10 seconds ago.

one-shots aren't good design, and while this "not good design" can still be tolerable (heck, even fun depending on the implementation) in Single-Player and locally-based games, it's just terrible in online/remote games.
anyone would tell you, there's a whole lot of shit that can happen between a client and a server, depending on a million different factors - so even if the one-shot is incredibly well-telegraphed, it really shouldn't be there.

and this is the point where I agree with Clive: given no one-shots, damage should be attrition-based. then and only then, we can get rid of instant flasks.
though I prefer not to. instead, I prefer to change instant flasks to the "oh shit" buttons they are supposed to be: 1 use. takes quite a while to refill. players shouldn't be able to spam "the red button", and they should have a good strategy for when to use it.

"

BerMalBerIst1 wrote:

"

johnKeys wrote:

and I assure you my calculated real-world "reaction time" in Path Of Exile, is different than yours, which is different than Clive's, which is different than BMBI's...

My reaction time is 0.16 seconds.

You mirin brah? XD

connect to a server in Antarctica, using Dental Floss, and measure again :)

"

johnKeys wrote:

Scrotie, you are doing calculations in Lab Conditions.
who's to say how much "reaction time" including ping, processing etc' a real situation has, and I assure you my calculated real-world "reaction time" in Path Of Exile, is different than yours, which is different than Clive's, which is different than BMBI's...
heck, my "reaction time" right now, is different than my "reaction time" 10 seconds ago.

Your core question is "who's to say?"

The answer is: game designers. In making a live-action game you cannot passively leave this question unanswered, because at some point the game will, inevitably, present the player with a stimulus which they must react to to influence outcomes. If the question is not answered on the theoretical level, then it is answered on the practical level.

This means designers of live-action games must either come up with a "one-size-fits-all" approximation of reaction time, or provide multiple different "sizes" as part of a variable difficulty setting system. And like baseball hats, one-size-fits-all works rather well here. Out of 16 million samples, ~250ms worked very well for reaction time. Ping latency is probably much more variable, but still rather predictable.

We understand the one-size-fits-all hat will never fit you, john. But you gotta admit those hats fit most everyone else.

Btw, I don't consider a monster skill with a 1-second or longer windup to ever truly be a one-shot. It's like saying you one-shot something with Hyper Beam in Pokemon. I mean, yeah, kind of, but it wasn't a one-turn KO.