Rethink the limit of special classes

mcompany
Autonomous Entity
Autonomous Entity
Posts: 872

Re: Rethink the limit of special classes

Post#21 » 24 Apr 2017, 23:28

First off, remembering back to 5.2, while snipers started to dominate towards the end, there was always counters (snipers are hard countered by shotguns, and I think that might be the only hard counter in the game besides maybe machine gun vs sniper). Second off, I'd like to say that if the rule changed like this, it would seem more limiting to other classes that aren't assaults, which still are extremely underpowered compared to special classes, with the only fair way to balance it for more assaults is to make larger maps with more starting bots

Anyways, I would agree with Pier. No matter how we change the system, the more optimized and more experienced players will always win in any deterministic game, so no rule change will change the amount of optimization you would need at the top level

Captain Spock (Castlevania)
Automaton
Automaton
Posts: 177

Re: Rethink the limit of special classes

Post#22 » 25 Apr 2017, 00:28

pier4r wrote:Well, I can tell you that since 5.3.1 the game is way more pleasant. If you think that more bots would not create dominant situation, you would be mistaken. Nullpointer can tell something about this.

As in every competitive game, at the top the games become tedious because the optimization is so advanced that to find the winning subtle change one has to spend a lot of effort (or has a lot of talent) still the database is full of upsets for stronger players that means that at the end there is always a corner case hat can be discovered.

My point is: whatever is the change that you make, one it is fixed and the game tends to be deterministic, at the end the top players would produce always similar optimal solutions that would require a lot of effort to be improved even slightly.


1. Whether more specific bots of the same kind creates dominate situation is what I'm wonder, but that the current rule creates dominate situation is a known fact.

2. I don't see any disagreement between our views about top level competitive games. However my point is that my suggested rule would introduce diversity, which entertains the medieval level players and stimulate top players to create all-around AIs.

BTW, I shown some "interesting" matches to my friend, but in his eyes, they looks like matches in a senior-group. Maybe the barrier between older players and newcomers is a problem.

Captain Spock (Castlevania)
Automaton
Automaton
Posts: 177

Re: Rethink the limit of special classes

Post#23 » 25 Apr 2017, 00:41

PEW wrote:
Castlevania wrote:
GFX47 wrote:All specialized bots should not compose more than half of your team

I like the current situation with only one specialized bot per class. Your idea would mean only one specialized bot in Kingmaker for example. I dont like that. Also I use recognising bot classes as a way of counting, which could not be done after that.


Of course everyone feels comfortable with the state quo. Actually I tried your way in retreat module too. It is really effective but takes efforts. I believe that once you would like to put efforts in, you can still do counting in the situation I described. It is still fair to everybody. Note that there still are a lot of people who don't have such a network we are using.

As to Kingmaker, at least it means 3 possible strategies: use shotgun to steal the resource once possible and intercept enemy transporters; use machine gun to establish a solid transporter team; use sniper to drive the enemy away from the resources. Now we have only one.

Captain Spock (Castlevania)
Automaton
Automaton
Posts: 177

Re: Rethink the limit of special classes

Post#24 » 25 Apr 2017, 00:46

mcompany wrote: Second off, I'd like to say that if the rule changed like this, it would seem more limiting to other classes that aren't assaults, which still are extremely underpowered compared to special classes, with the only fair way to balance it for more assaults is to make larger maps with more starting bots


That is not my intention to encourage using more assaults, since the only possible way--strength assault ability is unlikely to happen. My point is to use multiple special classes for tactical purpose and at the same time keep the match balanced.

Captain Spock (Castlevania)
Automaton
Automaton
Posts: 177

Re: Rethink the limit of special classes

Post#25 » 25 Apr 2017, 01:14

If most people like to have more kinds of special classes in a team, there is a more considerate rule:
The amount of any special class should not exceed the amount of assaults in a team.

TheKidPunisher
Autonomous Entity
Autonomous Entity
Posts: 565

Re: Rethink the limit of special classes

Post#26 » 25 Apr 2017, 03:21

Castlevania wrote:If most people like to have more kinds of special classes in a team, there is a more considerate rule:
The amount of any special class should not exceed the amount of assaults in a team.

How it is suppose to work on mind game?

It is just to be sure to understand what you want. I do not use special bots but i think that rules for special bots as to be the same on every map, it is more fair.

Captain Spock (Castlevania)
Automaton
Automaton
Posts: 177

Re: Rethink the limit of special classes

Post#27 » 25 Apr 2017, 03:37

TheKidPunisher wrote:
Castlevania wrote:If most people like to have more kinds of special classes in a team, there is a more considerate rule:
The amount of any special class should not exceed the amount of assaults in a team.

How it is suppose to work on mind game?


Assault Assault
Assault Shotgun
Assault Machinegun
Assault Sniper

User avatar
Kanishka
Skynet
Skynet
Posts: 1421
Contact:

Re: Rethink the limit of special classes

Post#28 » 25 Apr 2017, 06:41

Assault on Mind Game, no thanks.
Fixes break an AI more than bugs do. :ugeek:

Gladiabots Off-Topic Chat


My Stats: Kanishka_RN3;Kanishka_MiPad

harthag
Automaton
Automaton
Posts: 197

Re: Rethink the limit of special classes

Post#29 » 25 Apr 2017, 08:03

GFX47 wrote:I want to try limiting usage of special classes ... to refocus the game on AI programming ... I'm trying to lower the impact of classes on the combat system.
I see a number of possible solutions for this:

1. The original suggestion of this post, a simple change to the number of special bot classes allowed. This can take a few different forms:
a) further limitations, in addition to the ones already in place. this would force the player to have a more specific team composition on most maps, leading to a more specialized AI for that team composition. it would also mean the opponent would likely have a more specialized AI, that the player would have to prepare for in their AI instead if in their team composition.
b) lifting existing limitations to some extent. this allows for a wider range of possible team compositions on both sides, so the player has to adjust AI for their own composition as well as further plan for a wider variety of possible enemy compositions. this would probably have to be combined with some rebalancing of bot stats to avoid having any particular composition be OP.
2. Rebalancing the bot stats. Make the bot classes a true rock/paper/scissors of each other from a stats perspective. Not that any one class will always beat another, but that any class in it's preferred situation will always beat ALL classes that are not in their preferred situation. The two obvious of these would be sniper and shotgun. Make the sniper kill anything that stays at long range, make any bot doomed that gets in short range of an enemy shotgun. Assault and machinegun are harder to define, but I've seen it suggested that the assaults strength is in combining the bursts of multiple assault bots at once, so maybe something to make any 2 assaults (or maybe 3) against any combination of the same number of non-assault bots. I'm open to suggestion on machinegun, but that's the concept.
3. Increasing what all bots, regardless of class or stats, are capable of (this is my favorite). Increasing the programming options. Again, there are multiple possible variations to this, and the two most common "categories" I've seen are movement (like strafing and formations), and awareness (like directions and increased range options). I've seen, and understand, the main argument against this one, specifically that one of the key aspects of the game is to see what can be done with a limited set if tools to work with. I agree, to a point, but my counterargument is that adding to what a bot can do allows for completely new strategies and tactics that are not possible any other way. Other solutions lead to more sophisticated versions of the AIs that already exist. This one leads to completely new concepts.

pier4r
Skynet
Skynet
Posts: 3390

Re: Rethink the limit of special classes

Post#30 » 25 Apr 2017, 09:03

Castlevania wrote:If most people like to have more kinds of special classes in a team, there is a more considerate rule:
The amount of any special class should not exceed the amount of assaults in a team.


This I like more.

I like also the 2nd idea of harthag.

To me every bot should have its "sweet spot". Shotgun has speed and dominance short range. Mg is fatty but dominates medium range and resource collection. Sniper , snipes. Assaults is fast but not quite. Damaging but not quite.

So far the Damage per second is (considering average real battle cases)

short range:
SG > MG > AS > SN

medium range:
MG > AS >= SG > SN

long range
SN > MG >= AS > SG

Assault does not "stand" out. So either it gets faster, or with more shield (but then it substitutes the MG), or gets a sweet spot.

I would say, assault accuracy
short range: unchanged
medium range: unchanged
long range: 15% (instead of 10%)

That would make it quite different from the mg.
http://www.reddit.com/r/Gladiabots/wiki/players/pier4r_nvidia_shield_k1 -> Gladiabots CHAT, stats, insights and more ;

User avatar
PEW
Script
Script
Posts: 35
Contact:

Re: Rethink the limit of special classes

Post#31 » 25 Apr 2017, 11:55

pier4r wrote:I would say, assault accuracy
short range: unchanged
medium range: unchanged
long range: 15% (instead of 10%)

That would make it quite different from the mg.


Actually the MG and AS are the same on long range at the beginning, but after some time the MG gets a bit more powerful. Giving the AS a bit higher accuracy at long range, the differences here get even smaller.

I like the way it is. Every bot has nearly the same strength if you don't optimize them per class but you can only optimize the specialized bots. That's perfect.

Captain Spock (Castlevania)
Automaton
Automaton
Posts: 177

Re: Rethink the limit of special classes

Post#32 » 25 Apr 2017, 14:50

pier4r wrote:
Castlevania wrote:If most people like to have more kinds of special classes in a team, there is a more considerate rule:
The amount of any special class should not exceed the amount of assaults in a team.


This I like more.

I like also the 2nd idea of harthag.

To me every bot should have its "sweet spot". Shotgun has speed and dominance short range. Mg is fatty but dominates medium range and resource collection. Sniper , snipes. Assaults is fast but not quite. Damaging but not quite.

So far the Damage per second is (considering average real battle cases)

short range:
SG > MG > AS > SN

medium range:
MG > AS >= SG > SN

long range
SN > MG >= AS > SG

Assult does not "stand" out. So either it gets faster, or with more shield (but then it substitutes the MG), or gets a sweet spot.

I would say, assault accuracy
short range: unchanged
medium range: unchanged
long range: 15% (instead of 10%)

That would make it quite different from the mg.


Your idea does balance the stats, but we will see more "tower defensive" style matches.

In terms of balancing, I prefer to make the assault a lit bit faster. Unable to run away from a long range sniper undamaged even with full shield makes it embarrassed. And so, we distinguish the role of assault from machinegun's:
assault: aggressive medium range dominant, as what its name indicate, good at n vs 1.
machinegun: defensive medium range dominant, good at holding the ground, prevails in 1 vs 1 situations.

pier4r
Skynet
Skynet
Posts: 3390

Re: Rethink the limit of special classes

Post#33 » 25 Apr 2017, 16:28

Castlevania, you have a point. So a tad faster, not a tad more accurate.
http://www.reddit.com/r/Gladiabots/wiki/players/pier4r_nvidia_shield_k1 -> Gladiabots CHAT, stats, insights and more ;

TheKidPunisher
Autonomous Entity
Autonomous Entity
Posts: 565

Re: Rethink the limit of special classes

Post#34 » 25 Apr 2017, 17:22

Castlevania wrote:
TheKidPunisher wrote:
Castlevania wrote:If most people like to have more kinds of special classes in a team, there is a more considerate rule:
The amount of any special class should not exceed the amount of assaults in a team.

How it is suppose to work on mind game?


Assault Assault
Assault Shotgun
Assault Machinegun
Assault Sniper


A lot of people will say that is not a good idea (in grand master i never see an assault in ennemy team for this map).

pier4r
Skynet
Skynet
Posts: 3390

Re: Rethink the limit of special classes

Post#35 » 25 Apr 2017, 18:08

@The kid punisher. This is true until you are forced to use an assault bot.
http://www.reddit.com/r/Gladiabots/wiki/players/pier4r_nvidia_shield_k1 -> Gladiabots CHAT, stats, insights and more ;

mcompany
Autonomous Entity
Autonomous Entity
Posts: 872

Re: Rethink the limit of special classes

Post#36 » 25 Apr 2017, 18:41

Castlevania wrote:That is not my intention to encourage using more assaults, since the only possible way--strength assault ability is unlikely to happen. My point is to use multiple special classes for tactical purpose and at the same time keep the match balanced.

However, at the same time, the end result of such a plan on most maps would end up adding a bunch of assaults at the cost of other boys anyways and would most likely be more limiting on the amount of strategies available
harthag wrote:Rebalancing the bot stats...

Currently, I think most of the stats are pretty decent (besides the fact that it's incredibly hard to approach a sniper if it has the available cover, but I'm not sure there is a fix to that). The only real noteworthy thing is that assaults are weaker than machine guns in all ranges (including short range, where it actually gets a accuracy bonus compared to machine guns), has extremely flimsy armor (or rather, an extremely flimsy shield), and not enough speed carrying or not carrying to make up for these weaknesses. This would make a lot more sense in an RTS environment, where one can use sturdy, expensive machine guns or cheaper, weaker assaults, but this game isn't an RTS yet, so I don't think assaults should be forced on the player, or if they are, they actually get decent stats
harthag wrote:3. Increasing what all bots, regardless of class or stats, are capable of (this is my favorite). Increasing the programming options. Again, there are multiple possible variations to this, and the two most common "categories" I've seen are movement (like strafing and formations), and awareness (like directions and increased range options). I've seen, and understand, the main argument against this one, specifically that one of the key aspects of the game is to see what can be done with a limited set if tools to work with. I agree, to a point, but my counterargument is that adding to what a bot can do allows for completely new strategies and tactics that are not possible any other way. Other solutions lead to more sophisticated versions of the AIs that already exist. This one leads to completely new concepts.

Imo, there are so many situations with needed information that we simply don't have that this is almost guaranteed to be necessary

Captain Spock (Castlevania)
Automaton
Automaton
Posts: 177

Re: Rethink the limit of special classes

Post#37 » 25 Apr 2017, 20:09

mcompany wrote:
Castlevania wrote:That is not my intention to encourage using more assaults, since the only possible way--strength assault ability is unlikely to happen. My point is to use multiple special classes for tactical purpose and at the same time keep the match balanced.

However, at the same time, the end result of such a plan on most maps would end up adding a bunch of assaults at the cost of other boys anyways and would most likely be more limiting on the amount of strategies available

As far as I know, diversified strategies is boosted by diversified configuration.
In my suggestion, the total kind of special classes will decrease in very small maps, but could you please illustrate a "strategy" that requires coordination between different special classes? At least I don't have one. What I see is just that you cannot use "dancing shotgun" and "shooting at the edge" snipers at the same time in Mind Game. Ironically, Mind Game is the most notorious rock-paper-scissor in current rule.

User avatar
PEW
Script
Script
Posts: 35
Contact:

Re: Rethink the limit of special classes

Post#38 » 25 Apr 2017, 23:14

With all the parameters given, the assault is perfect in being so average. To make it more competitive, we should not change one parameter to make it more powerful, because it would destroy the nice balance. This game is designed to start with Assaults, built you AI for Assaults, then unlocking new bots and specialize them. That's good.

So instead, to make the assault more powerful, it should have a different, unique power, like planting a bomb once every minute or do a self destruct blowing everything in short range away. An ability that has to be unlocked even after Sniper class.

harthag
Automaton
Automaton
Posts: 197

Re: Rethink the limit of special classes

Post#39 » 26 Apr 2017, 00:46

mcompany wrote:
harthag wrote:Rebalancing the bot stats...
Currently, I think most of the stats are pretty decent
Oh, I agree. That suggestion was not because I think they're "broken". I just think adjusting their stats to emphasize their strengths and weaknesses even more will allow AIs to be built around the battle tactics of using them effectively. I'll use what you said here as an example(and touch on what you mentioned about decent stats for assaults):
mcompany wrote: besides the fact that it's incredibly hard to approach a sniper if it has the available cover
start with a board that allows 5 bots per team. One team is set up with 2 snipers and 3 machineguns to cover them. As stats stand now, that's "incredibly hard to approach", especially without knowing the enemy setup in advance. Let's assume the player takes a more "well rounded" approach, how about 1sniper, an assault for cover, another assault and 2 shotguns for aggressive offense. Unless one player seriously outclasses the other, the well covered snipers are going to pick apart the assaults while the machineguns hold off the shotguns. Once assaults are gone, focus fire will very quickly finish off the rest. But if we change the stats a bit, the lack of short range on the machinegun/sniper setup might become a problem. 2 shotguns at close range, 2 assaults at mid, and a sniper at long, all in their sweet spot, could make short work of 5 bots with only 1 target at optimal range, if the AI handles them all correctly for their strengths and weaknesses. Of course the assaults would have to have a speed tweek, and maybe damage, and the lone sniper and/or its guarding assault would probably not survive, but by then the shotguns could be in close, and the other assault on their heels, punching through the reloading machineguns before the snipers can get more than a shot off.

Thats just one possibility as an example. The point being that stats adjustments would make both setups be valid tactics, but the outcome is more likely to be determined by the AI that more effectively executes its tactic, instead if the setup itself determining the outcome.

pier4r
Skynet
Skynet
Posts: 3390

Re: Rethink the limit of special classes

Post#40 » 26 Apr 2017, 15:25

harthag I'm not sure if you are talking about setups with unlimited classes, but with unlimited classes you will likely see snipers all over, maybe sometimes shotguns for shotgun rush.

I did not like 5.2. much. I would prefer the actual situation way more.
http://www.reddit.com/r/Gladiabots/wiki/players/pier4r_nvidia_shield_k1 -> Gladiabots CHAT, stats, insights and more ;

Return to “Feature Requests”

Who is online

Users browsing this forum: No registered users and 1 guest