Map ballance

You can build a rock if I play paper, you can then build scissors to counter my paper.

If the map is inherently imbalanced you can't build something to counter that imbalance.

So you just assume that a computer is incapable of building a sufficiently asymmetrical balanced map?
I think it's possible, given enough time to refine the rules and algorithms.

A compromise would be to have maps that are gradually authored by people:
People could vote down unbalanced maps at the end of a match.
Matches at certain level would only play on maps that have been played X time and without unbalanced token.

The starting location in PA may end up being one you can choose (although I have yet to get my alpha key, what I've read from those so far is that it currently is choosable). Yes, you are locked in to it at game start, but if you can select it during the lobby, then it is a player's choice to start in a "weaker" vs a "stronger" location. Since we would have to assume that a player will choose what they perceive to be "the best" starting location available (choosing otherwise should be considered a bad player decision rather than a map imbalance), then I would suggest that anyone carrying on to the game proper is comfortable with their starting location. Anyone that cannot see any viable starting location would likely ask for a map reroll, or possibly even quit the lobby and look for another game, if all other players are comfortable with theirs.

"Fairness, in the context I’m using it here, refers to each player having an equal chance of winning even though they might start the game with different options."

Assuming each player has the option to select from an area, once everyone has selected you can claim that even a completely random map is "fair", because players have now agreed to start from those locations.

It does put the onus on the player to recognise a good vs a bad starting position, rather than on the map designer to create fair and balanced starts. For new players this will not necessarily be optimal, however anyone with experience of this type of game should be able to recognise particularly good or bad starts without too much effort.

I would also put up the assumption that Uber are intending to have algorithmic tweaks in the procedural generation; ensuring resource points have an relatively even distribution, setting starting selection areas based upon minimum criteria (e.g. amount of resource points, amount of land mass etc.) and other things you can do in the background to ensure a relatively level playing field for each player, even if the major details are the result of random functions. This is something anyone in the alpha is going to end up having a lot of suggestions for, as we get to a more stable build and start to look more at the big picture over "will the damn thing start up this time"

If there is one spawn location that has cover from artillery, the rest are out in the open, I can pick the spawn location with cover, you can't, how is that fair? We don't have an equal chance of winning so it's not. Choosing spawn locations does not make it a fair process unless all players have access to equal start locations, and on random asymmetrical planets that is almost never going to be the case. Also selecting start locations should probably not be available until everyone is ready to play otherwise the people who joined the lobby first have an unfair advantage in having more time able to determine the best start location. And if people are unhappy with their locations compared to others they will quit which means the other players time has been wasted and they have to try to create a new game where the same thing could happen all over again (unless it's an 8 player FFA or something where a dropped player or two isn't going to ruin the game).

I agree with torrasque that with a good enough algorithm you could create a balanced map.

I'm not sure how much player rating would work as a losing player would have a tendency to blame the map instead of himself. If many players play the same map you can get statistics based on the likelihood of 1 side winning.

Anyway, back to local imbalances. Local imbalances can be both having different strengths right now but it can also separated by time.
Timing attacks in Starcraft between different races is a perfect example. The timing attack tries to exploit timing windows when you are stronger than your opponent based on the inherent race you are facing. Starcraft 1 example. The Terran player can be able to bunker rush the Zerg at the start of the game which is a strategy that prevent or harass early Hatcheries. The Zerg player can then contain the Terran player with Zerglings and Mutalisk. The player terran can then push out when he got critical amounts of medics and marines or a mech army of goliaths, tanks and vultures before the Zerg players got dark swarm.

So for example if 1 starting position got earlygame advantage how does that compare to a player that got lategame advantage?

numptyscrub also added an important point. When players are able to chose starting positions it comes down to the skill of choosing starting position. If 1 player simply chose a bad spot. What is a good spot versus a bad spot in PA is yet to be discovered.

That is just one of many factors. For example you might have trouble expanding if you place behind huge cliffs. Artillery might only be relevant lategame so you are actually trading an early game advantage versus a lategame advantage.
That depends on how artillery work. For example. Maybe I can get close to your base hiding behind the cliffs and using mortars to hit you. Maybe I use mortars when you use regular artillery that can't hit behind the cliffs. Maybe I drop ontop off the cliffs with an early airrush and bombard you and prevent you from making air units.

True, the exact benefits of different terrain setups are yet to be determined, but the point was that it's possible for just one starting location to have a big advantage over the others so choosing doesn't improve the imbalance.

Except every starting position might have advantages and disadvantages and it is up to the players to make use of it and chose where they start.
You need to define what advantages there are, what disadvantages there are and how they span over the entire game before you can say it is imbalanced.

In regards to ranked matches here are a couple ideas.

1. Best of series instead of single match with Starting locations selected based on success in previous matches. So a higher ranked player would choose a starting point after a lower ranked player. Or if it were a series the starting locations for each consecutive round would be based on how long each player lasted in the previous round. IE those who are eliminated first would pick first.

2. Another option would be all players select what they feel is the best starting location. If more than one player selects the same location then that start point is eliminated and all players would then choose again. While this has the possibility of a map eventually running out of enough unique starting locations, the fact that a new one can be generated on the fly makes it relatively painless to start over with a new map.

For casual play these things can be configured in the match making interface. You could have options like: Randomized start location, Competitive, like option one, or eliminating, like option two, or some other self balancing mechanic that I have not thought of yet.

Not at all.
There are 2 approaches:
1) The starting locations are areas that are permutations of the same set of features - just assembled differently. The egg drops in part of that map and its up to the player to decide where they will start out their build. Taking a few mins to find a location that one particular player decides as optimum (from their own experience) might be more preferable to some who are thinking mid-long term. Others might take any location and start build immediately - sacrificing location for timely action.

2) If you can quantify the various advantages of one location over another and feel strongly about it - you can submit a bug report and the devs or map designer can adjust the map. By agreeing to play one map over another - users vote on how suitable a map is. Knowing a map is an advantage - experience you bring in before the game starts - an advantage over another player that doesn't have the same experience.

As I mentioned before - playing a rated game on a mirrored map isn't any measure of skill - just memory.

Some good ideas - You could iterate on this - you pick your top 3 choices and the system tries to make everybody happy - applying (1) the player rank for weighting those choices.

It does not matter if random maps could be generated as balanced (or at least close to), it only matters if random maps could be generated as imbalanced, so your point is not relevant (though true).

No you don't, not for this argument, there only needs to be any advantage to any type of terrain and since terrain can block some kinds of projectiles we already already know that to be the case. Also since we only need to be concerned with if a map can be imbalanced we only need to look at the worst case of that advantage to determine if it is something that is imbalanced, which would be one location having it and all others not, which would be imbalanced. And because it is impossible to say that with random maps worst case (or similar) scenarios will not happen, and given the sheer number of games that can reasonably be expected to be played it stands as a statistical probability that players would be playing on imbalanced random maps. Symmetrical maps fixes the balance problem quick and easy allowing for ranked play on random maps which is something I've always wanted in RTS games.

I do agree with torrasque that it would be possible to improve the random map generator to ensure balance even on asymmetrical maps, but as he hinted at it is something that could take a long time (months, and potentially years) to get even close to the same level of balance as symmetrical maps would provide (and is required for reliable ranking), but implementing symmetrical maps would be a much more trivial task.

That would probably work reasonably well except in the case of similarly ranked players where the worse ranked player may get too much of an advantage, and there's also an issue with the increased time commitment to play a series it's not unthinkable to require at least 2-3 hours of spare time to be able to play a single series.

Another good idea but the downside to this is that it also could be very time consuming especially for large maps with many players.

Try playing any RTS game against a decent player and not building for a few mins, you'll lose, it's simply not an option. And if the starting location is assembled differently then the balance is different, having a wall blocking your opponent's shortest route to your base is usually better than having a wall on the other side of your base blocking expansion away from your opponent.

As I've said I'm only concerned about the balance of asymmetrical maps when randomly generated, ones made or modified by a human are likely to be relatively balanced and not an issue.

I'm glad you went from calling the rotation of a tree an imbalance, to saying:

Which is the same thing as saying that a map can be how asymmetrical you want as long as it is balanced in which case I agree with you.

If I choose a location out in the open, and am happy to start the game with that choice, how is it not fair? How are you ascertaining that we do not have an equal chance of winning?

You are obviously making a lot of assumptions here, and it might be beneficial to look at some of those. For instance the point about cover from artillery; if I go for aircraft spam and never build a single artillery piece, the "value" of your cover can be drastically altered. If your covered location also had less easily reachable resource points, what you have assumed to be a superior start location could actually ended up being inferior over the course of gameplay, and by extension you could then be considered less likely to win. This is a function of the gameplay though, rather than initial conditions; any perceived initial geographic superiority can be overcome by decisions during gameplay, and a skilled player will be able to capitalise on that.

To requote the previously linked article: "Fairness, in the context I’m using it here, refers to each player having an equal chance of winning even though they might start the game with different options."

Actually I agree with godde, if you are arguing that a map is imbalanced because terrain has starting advantages, then I think you do indeed need to list what those advantages are, and in fact why you believe them to be an advantage. Because I'm not sure I follow your base reasoning at this point; sure you can hide from artillery behind a mountain, but that is not a starting condition because nobody is going to start with pre-built artillery.

If the terrain features in a theoretical start point are in the "wrong" place (i.e. I start on an open side), I can still use artillery effectively. If I plane rush, I can sneak them in from over the mountain and reduce the effectiveness of any ground-based AA, due to it being blocked by that same terrain.

To quote the previously linked article: "I want to stress that I am only talking about options that you’re locked into as the game starts. That’s a very important distinction. Options that open up after a game starts do not necessarily have to be fair against each other at all."

Since using artillery opens up after the game starts, having a natural counter on one or more sides is not necessarily unbalanced. It can be planned for and worked around by anyone of sufficient skill who does a little scouting (build artillery aiming at the open sides, or forgo artillery entirely for less terrain dependant options like aircraft). Adjusting the procedural algorithm to have terrain dense areas less resource dense would (IMO) automatically compensate for the "advantage" of terrain, as noted by my example above.

Still, it's early days, and I haven't even had a chance to look at some of the footage from peope who already have the alpha, never mind got my own key yet. I may well change my mind about a lot of this once I've had a chance to try it myself

Yeah I'd thought I might not have been clear enough about that in my long winded posts .

Given two players of equal skill who is more likely to win, the one with an advantage or the one without it? You do not evaluate balance using unbalanced variables, think of it like this, if you want to see if a set of scales is properly balanced you don't test it with a 1kg weight on one end and a 2kg on the other, you test it with two equal weights. So for map balance you test with you and yourself, if there is a single location that you would prefer then the map is imbalanced because one location is not enough for both you and yourself, you need to have two that you believe are almost perfectly equal to be able to say that the map is balanced. What really complicates matters though is that you are not the only one who will be evaluating the balance, all players will, which means that you need to be far more objective when balancing the maps preferably without any subjective influence, and the easiest and often best way to do that is with symmetry. With perfect symmetry it is impossible for any player to claim that they had a advantage or disadvantage which makes it the ultimate objective way to balance a map. However a really good map designer with excellent understanding of all game mechanics would be capable of producing asymmetrical maps that most players would consider relatively balanced, but that is something that would be extremely hard to program into a random map generator in a reliable way, and a random map generator does not have the advantage of being able to iterate upon the map to improve the balance over time the way authored maps can be improved.

Assuming units are properly balanced, being unable to meaningfully produce artillery is a significant disadvantage, I could simply focus on defending against air while you need to defend against anything I might decide to send at you. Also if PA is anything like SupCom then mountains will also provide some protection from bombs too.

Ahh, now I see why you are having trouble understanding the imbalance, you've misunderstood that section or at least how it applies to this situation. What it means is that the options that you’re locked into as the game starts need to be balanced, and the inherent advantages to the terrain of your starting location is one of those options, so if your starting location is more advantageous than others then the game is imbalanced. And it would be unbalanced if you had protection from artillery and others did not. The protection from artillery is not an option that opens up after the game starts, you have it from the very beginning regardless of if your opponents have artillery or not, so it is not exempt from needing to be balanced for the game to be fair.

Building artillery units is an option that opens up after the game starts, so the artillery units themselves do not need to be balanced for the game to be fair, they could kill a commander in one shot from anywhere on the map and still be fair because both players have an equal opportunity to do so (though it would not be fun as it would just be a race to build one). But as you can probably see, in this case a player with a start location that can protect them from the artillery has a massive advantage that is not fair because that protection was an option that was locked in at the start of the game and was not balanced because it was not an option for all players.

Yeah the importance of the imbalance is going to depend heavily on how the game plays, despite my current stubborn stance on this I wont rule out being the one that changes my mind either, though I think that's highly unlikely given past experiences with balance in RTS games .

The player who can read the map and is able to change his attack strategy basing it on the opponents units, layout and the terrain they are in.
If you're unable to adapt your strategy simply because you can't build out your ideal base because the terrain isn't optimal - then you deserve to lose regardless of how amazing your APM is or how well you're placed in the rankings.

If you want to test your scales - you test with both. Scales that only work when you have 2 1kg weights on either side aren't particularly effective

If you can qualify & quantify terrain characteristics then you can include the terrain in any value representation resulting from conflict.
Meaning that when it comes to the "score" you get from playing a map - you can factor in any perceived terrain benefit and the players ability to include that in their gameplay.

Which goes back to rankings in general - maybe we should figure out how to make them smarter and more representative. I posted some ideas earlier in the thread.

It's generally not done because it's complicated and in the facebook mentality - people are happy that having a high number of likes for their post means they're popular.

Developing a map generator that accounts for these issues isn't extremely hard - it's not a kiddie script - but it's doable. Writing any app well means understanding the problem and finding ways to issues creatively. Good developers tackle a problem by exploring the problem and allowing it to unfold as they develop a solution. Learning from what the problem itself is telling them - then incorporating that into the code.

Why can't a randomized system adapt & iterate the map over time?
As long as this characteristic is included in the problem definition - then there's no reason why game data can't be used to improve a map after every game that's played on it. If anything a app would be able to do it better than a person - because it's able to explore how the players adapted to the terrain features.

I said given two players of EQUAL SKILL. If they both possess equal skill then the player with the advantage will be able to correctly predict what options their opponent has attempt to overcome the disadvantage and can then use that to their advantage, the player with the advantage will always have more options available to them than the player with the disadvantage. It doesn't matter that there are options to minimize the advantage the other player has, the real disadvantage is that your options are more limited and a player of equal skill can take further advantage of that.

For example the obvious choice for the disadvantaged player is to go heavy air to fly over the mountain, so the advantaged player in knowing that can prioritize air defense and skimp on land defense, they can then use the resources that would normally have been used on land defense to pump out extra units while the disadvantaged player has to keep spending on both land and air defense because they don't have the same protection, eventually this leads to the advantaged player having a larger army and steamrolling the disadvantaged player. Assuming PA has similar units to SupCom this can get even worse at higher skill levels, in knowing that the disadvantaged player's best option for attack is to go air the advantaged player can then go air prevention with something like ASFs forcing the disadvantaged player to have to attack with ground based units where the advantaged player can really milk the advantage for everything it's worth.

As much as the disadvantaged player can adapt to minimize an advantage, the advantaged should be able to adapt to maximize the advantage.

That is certainly the ideal, but unfortunately it's just not possible to create such an algorithm within a reasonable time frame that would be consistent and accurate enough to deal with randomly generated maps. There are just too many variables that affect too many permutations of terrain features to be able to develop a system to reliably rate maps with a value that is able to influence rankings by the correct amount to ensure a fair rankings system.

Being able to use the terrain IS a tool experienced players have. Starting with the same terrain conditions for both players removed that skill from those applied to the game.
A players skill is their ability to use all the tools they have at their disposal.

Map A is designed with 2 landing spots. One spot is highly defensible (by any criteria you define) - the other isn't (by your own criteria).

100 games are played by the same 2 players - half the time Player 1 starts at one spot, the other half they start at the other.

The better, more skilled player - will win a majority of the time because they're able to change their strategy.

Maybe when enough games have been played on this map - the score a player gets would be a function of the map multiplier, the starting position and the combat/economy ratios. Maybe the map multiplier varies every n games played on it.

With your own definition - a player that recognises his opponent will predict his actions - will change or use that knowledge to his advantage. That also is a tool available - if you can do that and your opponent cannot then you have more "skill" than another player.

You fail to understand that skill isn't any single attribute and that any perceived advantage or disadvantage can be used by somebody with more experience or ability.

I understand what you're explaining wrt terrain and its value as a starting condition - but it's nieve.

Successful games find a balance of game play that naturally enhance creative problem solving without compromising the integrity of the overall environment.
Wouldn't you rather have a well balanced environment that allows you to properly deal with whatever situation you find yourself in? or a ranking system that properly reflects your skill?

This doesn't have to be done by Uber - from everything that's been said by them so far - they're open to community involvement - they just need to make the data accessible and let the community do their thing. We've already seen our first client mod here - many, myself included, are interested in pushing the boundaries a little to see what can be accomplished.

Sure that's a skill, but that has nothing to do with the balance of a map. It doesn't matter how good a player is, if one has a start location with a defensive bonus and the other does not it is unbalanced. Lets take a long cycling race as an example, if I was racing against a professional cyclist but he started with a rusted up BMX and I started with a proper bicycle specifically designed for the race then I have an advantage. It doesn't matter they are better than me at cycling races and would probably still be able to beat me (I'm not particularly fit) I still have an advantage over them because I have a better bicycle.

By playing both starting positions an equal number of times you cannot measure the imbalance of the map by the number of times the more skilled player wins. You would have to compare how many times they win when starting in one position compared to how many times they won in the other, and if the values do not match within a reasonable tolerance then the map is imbalanced, however using players of differing skill levels can cause discrepancies in the results that are a result of the difference in skill level as opposed to the imbalance of the map.

Perhaps you missed the several times that I have stated I'm only talking about the imbalance in asymmetrical maps when they are randomly generated for competitive play. You would be extremely unlikely to ever play the same random map twice, and even the number of times a random map was ever played would be unlikely to ever get to double digits.

As I keep trying to explain the skill doesn't matter for determining if a map is balanced or not, it doesn't matter if a particular player has the skill take advantage of a starting location or not, what matters for balance is if ANY player could gain an unfair advantage from a starting location because other players do not have the option for the same (or even an equivalent) advantage, and that is not something that cannot be guaranteed to never happen on a random map unless you have symmetry.

Map imbalances compromise the integrity of the overall environment for competitive play (which again is all I have been talking about).

In multiplayer if you are able to properly deal with whatever situation you find yourself in then the game is fundamentally flawed, it means that two skilled players can never defeat each other because they would always be able to deal with the situation resulting in a stalemate. You need to have situations that can't be dealt with to allow players to win, the objective of a game is to build such a situation for the opponent so that you can win, what you need is the tools to do so and a balanced environment for players to start from, and random symmetrical maps in PA would surely fit that bill but random asymmetrical maps would create imbalances that would not.
Also in a competitive environment a ranking system that properly reflects your skill is essential for things like ladders, matchmaking and for feedback on how skilled you are compared to other players, without such a ranking system it's pointless having competitive matches.
And that makes the choice more than a little obvious, if you can even call it a choice.

I'll have to partially disagree with that. Skill might not completely determine whether a map is balanced or not, but it does have an impact. If player A only knows how to turtle, and that's their only play style for whatever reason - maybe they've played to many tactical RTS games, where everything takes like a year to build - they would have a different opinion on what "balanced" means, compared to player B, who only knows how to rush, and has played way to much as the Zerg.

Plus, speaking about PA in particular, we have the ability to look at the various starting points, and choose which ever one we like the most, looking at that cyclist example combined with choosing starting points, the professional decided to start with the rusted up BMX, so it's his own damn fault.

Now lets look at that example as PA starting point selection. You're a moderately skilled gamer, against a top-tier skilled gamer. (To go along with the whole, being against a professional cyclist thing.)

You are both given a selection of starting points, and you decided to take the one that you view as best, lets say a starting point that's equally dependable from all directions, and only has two or three entry points due to impassable cracks in the ground or really large mountains or something like that, and is land-locked, and far from any ocean. The top-tier player decides to choose a place that's out in the open, like way out in the middle of a flatland that's on a coast. No natural defenses in any which way, and he can be attacked from any direction in any form (Naval, Land Or Air). He could have chosen a starting point that was a little more defend-able, or wasn't out in the middle of nothing, but he didn't. He made the conscious decision to start there.

Looking at that, as if those two starting points where the only two, you could say that the map was unbalanced as all bloody hell, but those weren't the only two places to start, and even though you might view yourself has having an advantage against your opponent, it wasn't because of the map, it was because of your opponents decision that caused you to have an advantage.

We'd have to find a medium, a certain meaning for balanced to apply to a spherical map, and not a flat map with boundaries. How do we determine balance for a map where you can go in basically ANY direction, and still reach your opponent?

You can't really ever have a true stalemate inside an RTS game though? Being able to deal with a situation at hand does not always mean victory, or a stalemate. If you were talking in one single situation at a time, then maybe.

IE: You send one tank, I destroy it. I then send one tank in response to destroying one of yours, since I now know that you have one less unit, but you end up destroying that one. Then you send one tank because you know that I have one less unit, but I destroy that one, and so forth. If games played like that, then yes, always being able to deal with a situation would cause a stalemate, but games don't work like that all the time, and sometimes even the most skilled player can't deal with a situation.

No you can't use player skill at all when considering if a MAP is balanced. Maps need to be balanced for ANY players which means and that means you can't factor in player skill because it is an unknown variable. You could randomly generate maps specifically for players and balance it for player skill so the better player gets worse starting locations, but then the outcome of the match is not an indication of relative player skill because the player skill had already been balanced out in the map generation making so it does not work for competitive matches where you are playing to improve your rank.

True but as I've said multiple times it's possible for there to be one spawn point that is significantly better than all other options, so in the cycling example it would be the equivalent of the professional cyclist choosing between several poorly maintained BMX bikes, and me getting to choose between several other poorly maintained BMX bikes and the proper bicycle for long races. The problem with random maps is preventing such an occurrence is exceedingly difficult (to the point that it is not worth doing) unless you use symmetry.

Just create symmetrical maps (or at least symmetrical starting planets), and it's all taken care of, the map is perfectly balanced and any imbalance from starting locations would be entirely due to the player.

You can if you design the game so that players "are able to properly deal with whatever situation you find yourself in", but it's very poor game design and as I said it would be fundamentally flawed. That's why you don't see it happen in RTS games, it's a really bad thing to do so developers don't do it, which was kind of my point.