Mixed difficulty balance

[holepercent]

(coming soon)

Is allowing mixed difficulty supposed to balance players of unequal skill playing in the same game? or to handicap the stronger player?

[Twilice]

If they implent the starcraft 2 map sorting feature they were supposed to be, it could look like this in the list:

*Element TD (easy)

*Element TD (normal)

*Element TD (hard)

But then again, we maybe want those "mode" sorting for things like:

*Element TD (all pick)

*Element TD (all random)

*Element TD (tower wars)

etc.

[Karawasa]

One of the last changes (4.3) of WC3 was the bounty balance. Basically, the gold base used in calculation of bounty is lower for lower difficulties. To put it in perspective, the total gold received from bounty on VH is about 90,000 while VE is now 60,000. Thoughts?

[Cisz]

If the money is less on lower difficulties, doesn't that make them harder?

What we want here is nerfing the ability of a lower diff player to put pressure on a high diff player, right? The game on easy should be easy, but the player playing on easy should not easily race a player on, say, hard.

-One way could be to prevent a lower diff player from finishing the wave. So the first player playing on the highest diff of the match finishing spawns the next wave. (Example: Players are playing on ve, ve, e, e, nm, nm, vh, vh. Only the 2 players on vh have the ability to start the next wave. If they both die, the right to start a wave early is handed down to the players on nm, and so on.) This could also be changed to "the players of the highest 2 difficulties" (vh and nm in my example), or even to "the players of highest diff or one diff below that" (only the guys on vh in my example, because they are two diffs ahead of the others).

-Or we could use variable timers, by measuring how long a player on vh actually needs for each wave in a real game, and force that onto players on lower diffs as a minimum wavelength. (Example: We play testgames and figure that a player on vh can finish wave 15 in 45s, and wave 32 in 75s. A player on nm finishes wave 15 first after 20s, and he will get a timer of 25s, or maybe a bit more. A player on ve finishes wave 32 in 15s and gets a timer of 60s, or maybe a bit more.)

From my experience, the time difference between a player on vh and ve increases over the game, so a fixed time difference would only be ballanced for a part of the game.

There is probably more possible, but I have to sleep first.

[Twilice]

-One way could be to prevent a lower diff player from finishing the wave. So the first player playing on the highest diff of the match finishing spawns the next wave. (Example: Players are playing on ve, ve, e, e, nm, nm, vh, vh. Only the 2 players on vh have the ability to start the next wave. If they both die, the right to start a wave early is handed down to the players on nm, and so on.) This could also be changed to "the players of the highest 2 difficulties" (vh and nm in my example), or even to "the players of highest diff or one diff below that" (only the guys on vh in my example, because they are two diffs ahead of the others).

Hey, that's a great idea. But maybe normal and hard should have a slight influence over the timer, while very easy and easy doesn't.

[echinodermata]

If the money is less on lower difficulties, doesn't that make them harder?

What we want here is nerfing the ability of a lower diff player to put pressure on a high diff player, right? The game on easy should be easy, but the player playing on easy should not easily race a player on, say, hard.

Agree.

-One way could be to prevent a lower diff player from finishing the wave. So the first player playing on the highest diff of the match finishing spawns the next wave. (Example: Players are playing on ve, ve, e, e, nm, nm, vh, vh. Only the 2 players on vh have the ability to start the next wave. If they both die, the right to start a wave early is handed down to the players on nm, and so on.) This could also be changed to "the players of the highest 2 difficulties" (vh and nm in my example), or even to "the players of highest diff or one diff below that" (only the guys on vh in my example, because they are two diffs ahead of the others).

The problem with that is the middle difficulty players. For example, if there's players on all the difficulties, the players on normal are getting gipped compared to the easys because they're playing at harder difficulty but can't influence the timer any more than the easys.

-Or we could use variable timers, by measuring how long a player on vh actually needs for each wave in a real game, and force that onto players on lower diffs as a minimum wavelength. (Example: We play testgames and figure that a player on vh can finish wave 15 in 45s, and wave 32 in 75s. A player on nm finishes wave 15 first after 20s, and he will get a timer of 25s, or maybe a bit more. A player on ve finishes wave 32 in 15s and gets a timer of 60s, or maybe a bit more.)

From my experience, the time difference between a player on vh and ve increases over the game, so a fixed time difference would only be ballanced for a part of the game.

Good idea.

[Karawasa]

Allow me to defend the variable bounty. Saying it just makes lower difficulties harder is quite a simplification. The bounty is determined by an exponential function. Thus, the difference in gold is felt more and more as the game progresses. It doesn't really become significant until mid game at the earliest. So, this isn't just about nerfing the difficulties. It's about nerfing their mid to late game, which is when they usually amass a million gold. I hope the difference is now clear.

As stated in a bunch of places, the problem of mixed difficulty is not present early to mid game. By the way, proposed countdown time for SC2 is;

VE - 15, E - 12, N - 9, H - 6, VH - 3

As you can see I made things a little more fast paced.

Update!

Taking into account Cisz's great idea of variable timers;

Level 1-20 = 15/12/9/6/3

Level 21-40 = 20/16/12/8/4

Level 41-60 = 25/20/15/10/5

Thoughts?

[holepercent]

make it a key point to test this thoroughly in beta.. mixed difficulty balance was never really tested in beta.. with the assumption that if VH was balanced, the rest would be balanced, (which isn't the case)..

[gwho]

It seems that the difficulty setting is individual - i've seen some players have 30% damage reduction, and others have 10% damage reduction

I think the voting in the beginning should make clear which polls apply to all and which apply individually.

I would also suggest that the game length poll come before the elemental rationing poll. Whether I would want all pick, all random or same random would largely depend on what level we're going to start on, not the other way around.

wouldn't the simplest way be to choose one variable to adjust and keep everything else the same?

You can adjust that one chosen variable level by level if you wanted to. (it can be a function not necessarily linear with level or linear with respect to other parameters).

Adding in complicated new calculation methods, or adjusting multiple variables shouldn't be necessary, and it just makes it a bigger headache to balance out and keep track.

[gwho]

Update!

Taking into account Cisz's great idea of variable timers;

Level 1-20 = 15/12/9/6/3

Level 21-40 = 20/16/12/8/4

Level 41-60 = 25/20/15/10/5

Thoughts?

what do these values mean?

what does a 15 mean?

the # of seconds after the player of concern has killed the last guy?

the # of seconds fixed to an entire wave?

which player? is it a difference or a value for a round?

i couldn't make out what the numbers meant in cisz's post either.

My first instinct is to respond that the player on lower difficulty setting requires less money to hold off the wave, therefore he has more principle to compound. this advantage is negated if we shorten the timer for a player on easier mode.... To have an advantage of being on easier, we should just keep everyone on the same timer, no? (timer defined to mean, identical wave starting times)

You can say the easier player will end up getting double benefit 1) easier creep; 2) more compounding. But this isn't a problem, it just means the the creep hp, (or whatever the primary parameter is) simply doesn't have to be changed as drastically to get the equivalent effect, compared to if there wasn't double benefit via compounding)

[holepercent]

only difficulty is selected by the player. and any player who chooses to -random in a pick game.

the timer refers to the countdown (based on difficulty) once any player finishes the wave.

[WindStrike]

I've done a lot of "host picks options", and when it gets to difficulty settings, it gives it twice. First, it allows me to pick any of them. Second, it then allows me to pick whichever one I picked then the ones higher than that. However, other people were apparently able to select lower difficulties than the initial one I set, which I assumed is the minimum difficulty pick. Is this supposed to happen?

[gwho]

I've done a lot of "host picks options", and when it gets to difficulty settings, it gives it twice. First, it allows me to pick any of them. Second, it then allows me to pick whichever one I picked then the ones higher than that. However, other people were apparently able to select lower difficulties than the initial one I set, which I assumed is the minimum difficulty pick. Is this supposed to happen?

i noticed being asked twice for difficulty twice as well. the second time, it sometimes doesnt have the easiest mode as an option. kind of messy

[dg86]

IMO the best way to address this would be this:

The people with the highest difficulty set the wave countdown.

Everyone else gets a variable amount of interest, adding .25% per difficulty level.

So ve gets 2%, e gets 2.25%, n gets 2.5%, h gets 2.75%, and vh gets 3% base interest.

Therefore, even though normal doesn't count down the timer if there's a hard or a very hard in the game, they still get additional benefits compared to very easy.

That or you can shift it so normal has 2% interest and punish easy and very easy, so you'd have 1.5%/1.75%/2%/2.25%/2.5% base interest.

[holepercent]

i noticed being asked twice for difficulty twice as well. the second time, it sometimes doesnt have the easiest mode as an option. kind of messy

only for host. the first one sets the minimum difficulty for the entire game. the second one sets your own difficulty. so for the first one, you can set normal for the game, then very hard for yourself in the second dialog.

[Karawasa]

How does it feel in 0.69? I tweaked the countdown timers.

[gwho]

i noticed being asked twice for difficulty twice as well. the second time, it sometimes doesnt have the easiest mode as an option. kind of messy

only for host. the first one sets the minimum difficulty for the entire game. the second one sets your own difficulty. so for the first one, you can set normal for the game, then very hard for yourself in the second dialog.

ooooo, lets have that labeled!

[Karawasa]

From another thread...

So, I propose building off of dg86's idea. Players get a weight from 1-5. But instead of the threshold being half of the total, the threshold is simply the average. Therefore in a non-mixed difficulty game, the countdown still starts upon first person finishing. In the extreme example of 7 VE and 1 VH, you would have 12/8 = 1.5. Meaning two VE people would be needed to start the countdown. Is that optimal in the extreme case? Perhaps not. But it's better than it is now and we maintain the competitive aspect.

Another question is how to convey this information to the player if we do go this route. Some people might scratch their heads as to when the timer will start...

[dg86]

IMO that still punishes people from choosing difficulty levels above the normal, which would more or less defeat the purpose.

What does it say when to test your strategy, someone does a single player game in very hard, rather than go to multiplayer?

[Karawasa]

IMO that still punishes people from choosing difficulty levels above the normal, which would more or less defeat the purpose.

The purpose is to find a compromise that helps alleviate this problem yet maintains the competitive aspect of the game. What if we did this instead;

We set the countdown timers back to 1,2,3,4,5 multipliers (currently 1,2,4,6,7). We implement the "weight idea" of dg86. We then calculate the average. The timer still starts after the first person clears the wave. However, it is multiplied by the average divided by the persons weight. It can't go below 1 for the "bonus" multiplier. In a non-mixed difficulty game, nothing would change. In the extreme of 7 VE and 1 VH, that gives us 12/8 = 1.5 average. A VE person finishes first, the countdown therefore has a total multiplier of 7.5. In practice this would turn out to be even higher as the average will be higher (yet VE person still has only the same weight). If the VH guy finishes first in that example, it would still be a 1*1 multiplier (no change).

[dg86]

As I said in other threads, there's nothing competitive about people playing on easier difficulty levels. Those who want to compete should be playing at the highest difficulty level of players in the game. I see no reason that a VE player or two should be able to start the countdown for a VH player under almost any circumstances.

[Karawasa]

I agree with that sentiment completely. However, short of giving the host no choice but to force non-mixed...this is the best we can do (imo). I'm confident that it will make mixed games feel a lot better. Will it be perfect? Perhaps not. But then again, there is no true way of perfecting mixed games. They are after all playing an easier game than you. But at the very least, we can minimize the impact they have.

PS: In other words, let's give this a shot and see how it goes.

[Twilice]

Can you make a scheme or table out of it? I have a little problem reading/understanding it

[Aranneas]

Why not just switch to a non-mixed mode? If host sets difficulty, that's what everyone plays at. If not, average the votes.

[Karawasa]

How does it feel now?