Funbalancing - Mags & Spidey

Unknown

Toxicadam wrote:

Well, if we are going to pretend like defense matters (and rewards players with points that make defenses), we need an AI that makes better choices than it currently does.

Putting precedent on a three match that will destroy 2-3 'special' tiles of the opponent over a simple 3 match that only collects a few AP for an ability isn't game-breaking. It's the way the game should already be.

Characters like Psylocke or Punisher would be borderline useless if the AI actually tried to destroy your special tiles.

Unknown

That's also why I said '2-3' tiles in a 3 way match should get precedent and not just any strike/web tile. If an opponent is laying down that many strike/web tiles to where they are bunching up, it's in the AI's best interest to get rid of them at that point.

You can always slightly increase the damage of the strike tiles to off-set the increased fragility.


It would also increase the utility of 'tile randomizers' like Loki and Moonstone who can protect/endanger strike tiles even more.

Unknown

Toxicadam wrote:

That's also why I said '2-3' tiles in a 3 way match should get precedent and not just any strike/web tile. If an opponent is laying down that many strike/web tiles to where they are bunching up, it's in the AI's best interest to get rid of them.

You can always slightly increase the damage of the strike tiles to off-set the increased fragility.

The special tiles are set up in a way where it's generally in your interest to destroy them immediately. The AI could figure this out, but if it did, most of your special tiles would simply get destroyed immediately. Let's take some obvious cases like Psychic Knife. You can collect some historical data and figure the average number of attacks per turn is say 1.5, and then multiply that by strike tile damage and then you'll figure that if it doesn't get destroyed in 4 turns it'll end up doing more damage than a comparable red ability, so you want to destroy that as soon as you can. But if the AI did that you'll just have people saying I used Psychic Knife and it got destroyed next turn, and if it happened often enough (which is what the AI should do if it's trying to beat you) you'll just have people switch to lazy Thor instead.

For special tiles to be viable you're basically saying the AI needs to feign a certain level of ignorance and only destroy them when they exceed a certain threshold even though it really should destroy them right away. It's okay for the AI to throw a game as long as they throw the game too much? But who decides how much is too much?

Unknown

The AI already does that for 5 way matches. It makes some of them, but not all. The AI also makes 4 way matches of off-colors when it is only 3 AP (or less) away from launching an ability. Those are just two examples of the AI behaving with ignorance or 'impaired logic'.

Unknown

Phantron wrote:

Toxicadam wrote:

Well, if we are going to pretend like defense matters (and rewards players with points that make defenses), we need an AI that makes better choices than it currently does.

Putting precedent on a three match that will destroy 2-3 'special' tiles of the opponent over a simple 3 match that only collects a few AP for an ability isn't game-breaking. It's the way the game should already be.

Characters like Psylocke or Punisher would be borderline useless if the AI actually tried to destroy your special tiles.

Not really though, Half the time they are not in a position to be matched right away or even for a long time. Why should the AI be stupid just to let us win more? If you were playing a human they would 100% get rid of the specials / go for 5 of a kinds everytime. I have wondered if it would be good to have abilities where they can put tiles down like black widow's purple just in one long row so they get a 4 or 5 of a kind. It would be so much more frustrating because of many many more losses but winning because the AI is dumb seems, well, dumb.

Unknown

sms4002 wrote:

Phantron wrote:

Toxicadam wrote:

Well, if we are going to pretend like defense matters (and rewards players with points that make defenses), we need an AI that makes better choices than it currently does.

Putting precedent on a three match that will destroy 2-3 'special' tiles of the opponent over a simple 3 match that only collects a few AP for an ability isn't game-breaking. It's the way the game should already be.

Characters like Psylocke or Punisher would be borderline useless if the AI actually tried to destroy your special tiles.

Not really though, Half the time they are not in a position to be matched right away or even for a long time. Why should the AI be stupid just to let us win more? If you were playing a human they would 100% get rid of the specials / go for 5 of a kinds everytime. I have wondered if it would be good to have abilities where they can put tiles down like black widow's purple just in one long row so they get a 4 or 5 of a kind. It would be so much more frustrating because of many many more losses but winning because the AI is dumb seems, well, dumb.

+1

If they ever do this, I'd also appreciate it if they awarded the 70/140 ISO prize upon successful defenses to make up for the fact that games would be longer (and that there would be more losses now and then).

Unknown

Phantron wrote:

The AI plays Spiderman well enough given its 'one ability per turn' restriction. Try to play Spiderman with the same restriction and he's suddenly not as powerful because it'd be almost impossible to keep 3 guys stunned even with a very large amount of blue AP because your web tiles will be exposed.

The whole point of the AI is that it doesn't assign any value to any tiles/abilities because otherwise you might as well just tell it how to play, and there's nothing hard to tell the AI to make a match 5 with Magnetic Field every single time. It doesn't do it because it doesn't know what Magnetic Field actually does, and that's by design. It's not supposed to have knowledge of how the game works because if it did, you'd just lose most of the time, especially if it has the 'one ability per turn' restriction lifted as well. It might as well pop up a box like 'I got the next 50 moves figured out already, you should hit retreat and save yourself the embarassment' if it actually was trying to win.

Why would you lose most of the time? Attackers always go first, can choose what characters to target or attack with, and know what the opponent's characters are capable of. They could let the computer go after your special tiles and try to deny your strong colors, use abilities "correctly" when they activate them, and use multiple abilities on one turn, and yet, ESPECIALLY if they don't let the computer look ahead of its current turn or make smarter, complex choices of which abilities to use, you'd still have a big advantage.

Unknown

I was lucky enough to pick up 5 blue spidey covers rather rapidly (like, by day 35-40) without using any HP to upgrade and quickly realized how essential the blue stun ability is and proceeded to use him in every battle. For a while i just thought "hey I got lucky with some cover pulls and placed in the top 5 of a spidey reward event, I should be able to stunlock teams to win matches" but slowly I started to realize that my high win rate using the perma stunlock was pushing up my MMR (or whatever you call it) and the scaling started to force me into battles with ridiculously strong opponents to the point where if I can't get any blue going in the first 4-5 turns and he gets downed, it's an auto-retreat EVERY TIME. There is a 0% chance of me beating the teams I have been facing without the 2AP stun ability.

So the fact that he is broken combined with the current scaling system seems to have me backed into a corner. I can't win without him and if they nerf him even slightly I will likely start losing more than 75% of my battles (which quickly makes the game no longer fun/playable). I just hope that if they are going to nerf him (and I've read that they have shelved the idea of nerfing anyone indefinitely after the Thorverine backlash) that they also change the scaling system to compensate.

Question: If the nerf happens and I start losing many matches, will that cause the scaling to go the other direction? Will my opponents get weaker? Or is the scaling based purely off of roster strength irregardless of win/loss record?

Nonce Equitaur 2

From This is how you FunBalance Spider-Man.

Web Interference - Blue 5 AP
Spider-Man puts a web on a chosen location, slowing the enemy plans. He chooses and webs a basic tile, and opponent powers using that color are unusable until the webbing is removed. The webbing remains for 2 turns, then splits and remains on two random environment tiles, dissolving in a few hours.
Level Upgrades
Level 2: Countdown tiles may be webbed, stopping enemy countdown tiles for the webbed color.
Level 3: Attack tiles may be webbed, nullifying enemy attack tiles for the webbed color. Webbing remains 3 rounds.
Level 4: Protect tiles may be webbed, nullifying enemy protect tiles for the webbed color.
Level 5: Strike and trap tiles may be webbed, nullifying enemy strike and trap tiles for the webbed colors. Webbing remains 4 rounds.

This removes Stun entirely, and allows the opponents to move, but can nullify enemy powers for awhile. Full power against Daken/Rags, webbing a red tile would prevent Thunderclap for 4 rounds. A green match would still generate red strike tiles, and they would still hurt, unless it was a red strike tile that was webbed. If a green tile was webbed, Daken would no longer make red strike tiles on a green match. If the webbed tile was cleared, they could start using their powers immediately.

The webs clinging to environment tiles are usable for healing.

Unknown

gamar wrote:

Phantron wrote:

The AI plays Spiderman well enough given its 'one ability per turn' restriction. Try to play Spiderman with the same restriction and he's suddenly not as powerful because it'd be almost impossible to keep 3 guys stunned even with a very large amount of blue AP because your web tiles will be exposed.

The whole point of the AI is that it doesn't assign any value to any tiles/abilities because otherwise you might as well just tell it how to play, and there's nothing hard to tell the AI to make a match 5 with Magnetic Field every single time. It doesn't do it because it doesn't know what Magnetic Field actually does, and that's by design. It's not supposed to have knowledge of how the game works because if it did, you'd just lose most of the time, especially if it has the 'one ability per turn' restriction lifted as well. It might as well pop up a box like 'I got the next 50 moves figured out already, you should hit retreat and save yourself the embarassment' if it actually was trying to win.

Why would you lose most of the time? Attackers always go first, can choose what characters to target or attack with, and know what the opponent's characters are capable of. They could let the computer go after your special tiles and try to deny your strong colors, use abilities "correctly" when they activate them, and use multiple abilities on one turn, and yet, ESPECIALLY if they don't let the computer look ahead of its current turn or make smarter, complex choices of which abilities to use, you'd still have a big advantage.

But you're still saying the AI should be handicapped in some way to make it more beatable and it's already handicapped pretty significantly, and deciding what handicap is fairly arbitrary. But handicapping it less shifts the focus toward playing characters like lazy Thor. Let's take Psylocke, who does 965 damage with 8 red AP and creates a 192 strength strike tile. To do 200 damage per red AP, you'd need 1600 - 965 / 182 = 3.5 hits. Assuming both players always destroy 3 tiles each round, then to get 3.5 hits means a total of 21 tiles have been destroyed, which is roughly 1/3 of the board. So you can say 33% of the time it is not going to do better than 200 damage per red AP (worse than Adamantium Slash) and 66% of the time it'll be better. This ignores that doing a large chunk of damage at once may have its own value (like downing someone), or that the computer can certainly be expected to destroy more tiles per turn than a human player if it's programmed to do so due to its superior analysis of the board. With these odds you can say it seems like Psychic Knife is probably worth using, because we know the AI currently places no value on destroying special tiles so calculating its chance of survival merely as a function of tiles destroyed is reasonable.

But what if the AI focuses on tile destruction? If it can bump up the chance to destroy your tile to 50%, now you got an ability that is 50% of the time better than 200 damage per red AP and 50% of the time it is not. Why would you use that over an ability that simply does 200 damage per red AP? In this case, you'd just use lazy Thor, who should be expected to do 200 damage per red AP 100% of the time. There's no uncertainty here. In fact pretty much all balanced 3* abilities are done this way. If the enemy focuses on nailing your strike tiles on Judgment, it'd likely end up doing less damage than something like Thunderous Clap that's just straight up damage. All you'd get is teams end up being something like Hulk/lazy Thor/whoever because no balanced ability that depends on special tiles can possibly have any use if the opponent is set out specifically to stop it (otherwise it'd be way overpowered if you can't even stop it).

Unknown

Phantron wrote:

But you're still saying the AI should be handicapped in some way to make it more beatable and it's already handicapped pretty significantly, and deciding what handicap is fairly arbitrary.

Well, no, I'm saying that even if they have the computer play significantly "not-dumber" in the way you described in your earlier post, it's still handicapped in numerous ways. I'm not saying those handicaps "should" be there or not.

Phantron wrote:

Assuming both players always destroy 3 tiles each round, then to get 3.5 hits means a total of 21 tiles have been destroyed, which is roughly 1/3 of the board. So you can say 33% of the time it is not going to do better than 200 damage per red AP (worse than Adamantium Slash) and 66% of the time it'll be better.

Actually, you can't say that, because (1)the choices aren't random and (2) destroyed tiles replace themselves

Phantron wrote:

the computer can certainly be expected to destroy more tiles per turn than a human player if it's programmed to do so due to its superior analysis of the board.

Speak for yourself, buddy!

Phantron wrote:

But what if the AI focuses on tile destruction? If it can bump up the chance to destroy your tile to 50%, now you got an ability that is 50% of the time better than 200 damage per red AP and 50% of the time it is not. Why would you use that over an ability that simply does 200 damage per red AP? In this case, you'd just use lazy Thor, who should be expected to do 200 damage per red AP 100% of the time. There's no uncertainty here. In fact pretty much all balanced 3* abilities are done this way. If the enemy focuses on nailing your strike tiles on Judgment, it'd likely end up doing less damage than something like Thunderous Clap that's just straight up damage. All you'd get is teams end up being something like Hulk/lazy Thor/whoever because no balanced ability that depends on special tiles can possibly have any use if the opponent is set out specifically to stop it (otherwise it'd be way overpowered if you can't even stop it).

If necessary, the devs can always adjust the strike tile damage until it matches what balance they want, or have strike abilities make more tiles of weaker value, or any of many other options.

And why on earth would you assume that the chance of destroying any given tile would be 50%? If this were true, nobody would ever lose to Dakens or Patches because they could just match all those tiles away!

Unknown

For ability that places special tiles at random locations, like Judgment or Psychic Knife, you can basically estimate its chance of survival as (number of tiles destroyed) / 64 (number of tiles on the board).

In the case of Psychic Knife it has to survive 21 tile destruction to get 3.5 hits in (3.5 hits from strike tiles = 7 total matches because the other side gets to move as well), and 21/64 ~= 33% chance, so it is going to survive about 66% of the time to cross the 200 damage per red threshold, which is has to in order to be a viable attack.

Note that 66% isn't very good odds, because the break even point is 50%. If an ability has 50% chance to do better than 200 damage/red, and 50% chance to do worse, you're almost certainly better off just using an ability that does 200 damage/red 100% of the time instead.

And the assumption that the computer only gets 3 tiles per turn is very low. It is certainly capable of doing far better than that. It can easily calculate the result of every cascade on the board it can see. Sure you can make the computer not to do that (and it probably already purposely skips a good cascade on purpose already for this exact reason) but why increase the intelligence of the computer if you've to tell it to act dumber elsewhere to make up for it?

Unknown

(1) The matches aren't random. You are likely going to choose to not destroy your own strike tile, so your strike tile doesn't have to "survive" YOUR matches unless you want it to. Also, you can choose to activate it when a red match isn't available and won't become available for multiple turns or a cascade

(2) Whenever tiles are destroyed they are replaced, so even if matches WERE random, the relevant denominator is the number of tiles on the board over time, not the number of tiles on the board when it was placed. That is, 21/82, not 21/64. And that's also overstating it because tiles don't "fall in" randomly - the board isn't reshuffled whenever a match is made

Unknown

I posted this a while ago in suggestions - perhaps a simple, self limiting adjustment for SPidey is to keep him exactly as is - however, make his web tiles blue.

It will do a few things:

1. stop the potentially endless stunlock; very quickly, if you want more blue, then you will be destroying your own webs - which means total "stun-turn" is self capping.
2. if you are burning up the blue to keep the stun going, y'all won't have many web tiles to heal with. stun vs heal becomes a strategic choice.

BTW, I have 5 blue spidey; and use him - when I am not using OBW

Unknown

If we're assuming 3 tiles get randomly removed from the board, then on each turn there's a 61/64 chance of it staying. Over 7 matches, that is
(61/64)^7 = .7145....

1 - 21/64 is not correct because you are not accounting for the tiles getting replaced, so you will see more than 64 tiles on the board

1 - 21/85 (you meant 85 and not 82, right?) is also not correct because you're assuming every tile seen has the same likelihood of dying. Tiles that only show up in later rounds have less chances of disappearing by the 7th match, since we're assuming that at any one turn, every tile has an equal chance of disappearing

Even my first "correct" calculation is incorrect because it makes the assumption that tile matching is totally random. It's correct given the assumption, but the assumption itself is really hard to accept

It's a good idea to look at the board and assess the situation. Often times, when I'm trying to place red strike tiles, I'll get rid of some red to make it a bit tougher to remove the remaining red strike tiles. Like... if there's 3 reds that can be matched easily, and another 2 reds in hard to reach corners of the board, I'll make that red match, THEN through strike tiles onto the board to make sure that those weird locations are the ones with the strike tiles.

This is the sort of thing where humans would have an advantage over the AI. The AI could theoretically be programmed to make really smart plays and change strategies on the fly based on the board, AP, health etc. but it's not quite there yet. People are simply arguing that with the AI as it is, slightly improved to be better at match-5s and better match priority, humans will still have a big enough advantage.

Kelbris

Vinny J wrote:

I posted this a while ago in suggestions - perhaps a simple, self limiting adjustment for SPidey is to keep him exactly as is - however, make his web tiles blue.

It will do a few things:

1. stop the potentially endless stunlock; very quickly, if you want more blue, then you will be destroying your own webs - which means total "stun-turn" is self capping.
2. if you are burning up the blue to keep the stun going, y'all won't have many web tiles to heal with. stun vs heal becomes a strategic choice.

BTW, I have 5 blue spidey; and use him - when I am not using OBW

With the minimum cost being 3, this might work.

Unknown

noknuckles wrote:

If we're assuming 3 tiles get randomly removed from the board, then on each turn there's a 61/64 chance of it staying. Over 7 matches, that is
(61/64)^7 = .7145....

1 - 21/64 is not correct because you are not accounting for the tiles getting replaced, so you will see more than 64 tiles on the board

1 - 21/85 (you meant 85 and not 82, right?) is also not correct because you're assuming every tile seen has the same likelihood of dying. Tiles that only show up in later rounds have less chances of disappearing by the 7th match, since we're assuming that at any one turn, every tile has an equal chance of disappearing

Bah! But you see my point - and even these numbers are lowballing it because most matches, and even cascades, leave large amounts of the board untouched

Unknown

1 - (21/64) = 67%
(61/64) ^ 7 = 71%

For an estimation the two are close enough and saves you the trouble of getting out a calculator.

If you never destroy your own special tiles that's a very good way to handicap yourself because just because you're not touching a red match 3 doesn't mean the AI will leave it alone, and since red is valuable commodity, it's best for you to destroy your own strike tiles than letting the AI have it. Sure, you can try to move the red away somehow, but you usually can't and there's no guaranteed this won't lead to a bad move. About all you can do is make sure you didn't drop a red special tile when a red match is imminent, but even that doesn't really mean much due to cascades not to mention any tile destruction specific ability. If the AI is supposed to actually play for real, it'd mean whenever you go against Magneto with 5 blue, whatever special tiles you put up gets erased the next turn. Somehow I don't see that being anyone's definition of fun. For that matter, it'd mean an AI with Magneto basically takes an extra turn every turn, assuming they're still subject to the same 'one ability per turn' rule.

Simply put, maximizing AP gained is an overwhelmingly better strategy than any strategic refinement. There may be cases where it's okay to veer off the optimal plan for an especially powerful strike tile (for example, if players can get Threaten strike tiles) but the AI is programmed assuming the game is balanced so that you won't suddenly have 2 500 strength strike tiles show up on the board. Based on the dev threads there's good reason to believe that when the AI isn't maximizing the AP gain it's already doing that on purpose to give you an easier time. If it doesn't even need to maximize AP gain, there's no reason for it to bother with any lesser refinement to its strategy. Having the most AP possible is overwhelmingly a better strategy than any other strategy.

Unknown

Please stop nerfing, stop even discussing nerfing. Just don't do it. Improve weak and underused heroes. Nerfing creates an atmosphere of fear when using your hard earned ISO-8 and cash to develop your roster. Nerfing is almost never the answer. Even the Black Panther nerf was bogus. They just didn't get a huge complaint wave because people didn't have time to invest in him. Black panther has 2 bad abilities (just compare him to Punisher's green for strike tiles or Mag's blue for shields) and one great one (black). Why nerf him??? He wasn't dominating the PVP scene, which was the excuse we heard last time on the Thorverine nerf. If you want the game to take off, you need stability, that means no nerfs. I'm about to spend money and over 100,000 iso-8 in a champ threatened with a nerf. How can I be made whole after he is ruined? What did people with a useless Ragnarock get after the nerf?

Also, make Spider-man Bag Man useful or redeem him fully covered for 50,000 iso-8. He is a bad joke and that bad joke has run its course. Enough is enough.

I love this game and I've played over 600 hours. I will continue to play, but devs need to understand that 400 of those 600 hours were spent improving just 3 heroes in my roster, 1 was already nerfed (Thor) and now one more is the threatened by nerfs. How are you going to reimburse me for 400 hours of play? If you go minimum wage, they'll should pay me 7 x 400 = 1600 dollars, way too much. How about 1 dollar an hour? That's more than I've ever spent in this game. Devs need to look at nerfing in this light. If you want to see more strategies, different teams, etc... don't nerf, improve characters that people don't use enough. You have all the usage statistics, you know that there are at least a dozen heroes that barely ever see play. Why mess with the few who are great and fun to play with?

Cheers,
NazgulPrime

IceIX

Argh, you dragged up a two week old topic when there was already one discussing the same basic thing AND I'm consolidating conversation into one thread. This thread was made for lockin' and that's just what I'll do. viewtopic.php?f=7&t=5307