Vtyjkyuk

In the end it all boils down to one axiom: Game design is hard!

And multiplayer game design is even harder. After years of development, you think you came up with a perfectly balanced game design which allows for multiple viable playing strategies. You did extensive playtesting with a dozen or so full-time inhouse testers and a hundred external spare-time beta testers. They all tried different strategies and they can't tell which one is the optimal one.

But the moment you publish your game, thousands of people (millions if you are one of the lucky few who land a hit like Dota2 or LoL) will jump on it and look for ways to optimize their gameplay. They will scrutinize your game mechanics in ways you never considered possible. And when it is a multiplayer game, knowledge will spread quickly. When one player figured out some new trick, everyone playing with them will learn that trick from them.That knowledge will then be used by all the players to build upon and do even deeper analysis of the game. They will write forum posts and wiki articles to share their research. This thorough overanalyzing will result in someone finding the one strategy which is just a tiny bit more optimal than the others. That knowledge will spread, and soon everyone will play that strategy.

But there is one game mechanic you can not optimize for, and that's human psychology. So if you still want to allow multiple viable strategies, then allow the players to play mind games.

1. Create a rock-paper-scissors system with different strategies. Strategy A beats strategy B and strategy B beats strategy C, but strategy C beats strategy A. A more complex game might have more than three strategies, and they might not necessarily be perfectly balanced against each other. It is OK when one strategy beats more strategies than the others or is much easier to play. All that matters is that you have at least one viable hard-counter for every possible strategy.


2. Design your match progression in a way that players need to commit to one of these strategies early on. You should still allow to change strategies mid-game. Otherwise the match is decided before it even started, which can be pretty frustrating. But changing strategy should be a costly and risky move.


3. Give the players the ability to hide from their opponent which strategy they pursue until the opponent fully committed to theirs. Also add ways for players to mislead the opponent about which strategy they actually pursue, so they can trick the opponent into wasting time and resources on preparing a counter for a strategy they won't use.

Now every strategy is viable as long as the player can bluff their opponent into assuming they pursue a different one.

You said it: Dota has a mixture of RPS balance and fixed, hidden character selection.

The characters that the enemy team chooses to play are hidden information, and thus you must evaluate your own choice of character against a probabilistic model of what the enemy is going to play. Because outcomes in Dota are more dependant on hero selection than in Lol, it is very important to both select heroes that are strong against the heroes that the enemy is likely to play and you want to avoid heroes that are likely to be countered by your enemy doing the exact same thing.

This means that if a hero becomes very popular, then people will choose to play its counters more often, even if such counters are not as strong in general as the popular hero. In other words, the dominant meta heroes in Dota are dependent on a priori expectations about enemy heroes. So hero usage fluctuates chaotically even in absence of balance patches, and at any moment even the theoretically strongest team composition will be beaten by a purpose-made counter team.

It's a cheap way to force diverse hero usage in my opinion. Matches that are decided before players even begin to play are not particularly engaging, but Dota's popularity is surely not entirely illegitimate (I don't play either Dota or Lol btw).

Note that if you remove one of the two components then the whole thing doesn't work anymore: in Lol hero choice is hidden but because counters aren't as strong, the theoretical meta team tends to not have a proper counter with enough of an advantage. In Overwatch you can switch heroes midgame and so even though some heroes are hard countered, you can just switch if you do get countered and generally you can recover from the loss of momentum. Interestingly, it means that in Overwatch some heroes don't get play time even when their counter is not on the field. This is merely because of the threat of potential switching of the enemy to that hard counter.

Player of both here. DOTA has a set amount of damage for most attributes. A Lina ultimate will always deal X amount of damage unless the enemy has a debuff on them that increases the damage they take. This means that Mage damage output from skills remains relatively the same throughout the game.

However, this is offset but each character having a "main" stat like strength, agility, intellect. which adds the the base auto attack damage. The scaling is a little more gradual with utility items and activation items making most of the changes.

This allows for just about any character to be viable as you can almost always defeat the enemy by changing your utility itemization. Get the BKB (Anti-Magic damage) to deal with a pesky mage. Get a Force Staff to give yourself the mobility to escape a jungler. These item changes can effectively shift the strategy unlike LoL.

LoL's take on the MOBA genre is that there will be a rotating list of strong characters to prevent burnout on playing a singular strong META list. This is accentuated by the itemizations and AP/AD affecting the damage of abilities and at different ratios.

To delve deeper into "Why do some games persistently have mostly one viable strategy, while others can have many?". It is the design philosophy. LoL as mentioned above, intentionally has a Flavor of the Month META list. While DOTA as mentioned above, intentionally goes for such a wide variety of counters in combination with slowly increasing attributes to keep the play more like Spy Vs. Spy where gathering knowledge and implementing it becomes more important than individual picks (usually).

Balancing is hard. Thus, changing which characters are stronger at certain patches leads to variety and is easier in terms of design. DOTA's approach is much more fragile but much more rewarding in terms of what characters you can play. Occasionally, certain skillsets have become meta and dominated it but, can always be countered with a certain playstyle or item set.

What other posts have described from a gaming perspective is simply called a stable mixed strategy equilibrium in game theory. Here are some requirements for this:

You generally want to have a game of imperfect information or simultaneous moves. Game theoretically, simultaneous moves are a special case of imperfect information. Without imperfect information, each choice by a player becomes an optimization problem with little flexibility for a player. If one choice is just slightly better than another, one should make that choice. With imperfect information, a player can make a choice that has value because it is unexpected, even if under perfect information there would be a strategy for the other player to make this choice suboptimal.

What is also helpful is to have a game with incomplete information about the other players. Not knowing the goals or ability of another player means that different strategies can be optimal depending on the beliefs the player has about the other player's goal or ability.

The question is what makes the mixed strategy stable? Suppose that a small deviation from the mixed strategy equilibrium (increasing the probability of one action by a small amount) incentivizes others to respond in such a way that the player wants to deviate even further from the mixed strategy equilibrium. Then this equilibrium is unstable and unlikely to be observed in reality. What makes an equilibrium stable is that the other player has an optimal response to my deviation, such that I want to move back towards the equilibrium. Hard counters provide such a mechanism. If a player takes a particular action with greater likelihood, the other players will respond by playing the hard counter to that action with greater likelihood, incentivizing the player to decrease the probability of that action.

Here is how some games achieve strategic diversity. Rock paper scissors is a simultaneous move game with hard counters. Thus, it has a stable mixed strategy with maximal diversity given the number of strategies. Chess is a game of perfect information in sequential moves. This is bad for strategic diversity in principle. However, given our limited computational abilities, it is actually a game of imperfect information as neither player can compute the optimal strategy to the end. Moreover, at the highest level it is a game of incomplete information because the players do not know which openings the opponent has prepared.

I'm not initmately familiar with LoL or Dota, and a definitive or comprehensive answer will be hard to come by in any case, but allow me to share some mechanisms I have observed.

Choice

From a naive viewpoint, the more unique options you offer to players, the greater the risk that one of them will turn out overpowered and/or break some core mechanic. In Dota/LOL, the number of playable characters, each with their own unique abilities, is quite high, so if only one of them turned out too powerful, that will shape the metagame. To make matters worse, many players tend to more or less blindly latch on to whatever strategy's deemed the most powerful, exaggerating the imbalance.

A common approach to mitigating this is to introduce specialization via classes or roles (e.g. tank, spank, healer in most MMOs). The idea is that if different options compete in different categories, there has to be at least some diversity (e.g. at minimium, one optimal tank, damage dealer and healer instead of just one optimal character).

In practice, this approach carries the danger of pigeonholing both individual options (a damage dealer might now be measured only by their dps, making it easier to find the "best" choice) and overall composition (a party needs to have a tank and a healer).

The above is a pretty typical attempt to "pre-empt" a specific metagame by, effectively, codifying a different one, thus limiting (viable) choice. This may well be intentional, as it can make a game more accessible and easier to balance towards the chosen playstyle (Blizzard likes to do this a lot), but whether desired or not, it will lead to a "stale" meta that swings heavily with balance patches. LoL would probably fall into that category.

On counters

I would argue that where an option falls on the "hard vs soft counter" spectrum matters far less than how options are chosen and what mechanisms a player has to react to an opponent's choice.

Consider Starcraft 2, for example. As a rule of thumb, you have to pick a build before even scouting the enemy. Once you know what they're up to, you can of course react, but that takes time. Philipp has outlined that mind game aspect pretty well.

Say you went for Roaches and your opponent went Phoenix. That's (at reasonable numbers) a hard counter, the Roaches can't ever shoot up. However, it doesn't mean you lost the game, you're just at a disadvantage until you transition into something else, say, Ling/Bane/Hydra.

Now, in LoL/Dota, you can't switch champions mid game, but you can pick items and change up your tactics to adapt to a limited degree. The relative influence of these mid-match decisions compared to your initial champion pick determines the "wiggle room" you have for making a strategy viable, affecting the range and diversity of viable strategies.

Synergy and interaction

As a rule of thumb, options that compete, but do not (meaningfully) interact lead to one superior option emerging and the rest being phased out.

Spellcasters in D&D 3.5 had, among many other tricks, two very similar methods of defeating a monster: reducing their hitpoints to zero, or reducing one of their ability scores (strength, intelligence etc.) to zero. The latter turned out to be slightly easier, so that's what character-optimizing players would pick.

In this case, changing the numbers will usually just lead to a few days of chaos until players have determined the best option again.

In the interest of a diverse and malleable metagame, options need to interact. Whether this is in the form of counters ("rock-paper-scissors"), synergies or something else entirely, the idea is that the presence of one element changes the strategic role or value of another, shifting in turn the value of the inciting element (the principle behind "perfect imbalance").

Again, there is a risk of severely limiting your metagame if you don't provide a wide range of interactions. If fire damage is unusually common or powerful, and Ice Armor is the only counter to that, guess which spell everyone will pack?

Now, if I were to hazard a guess, based on PSquall's description, it sounds like LoL with its rigid role and lane structure might have hit this nail as well.

What exactly makes a game having One Strategy To Rule Them All, and
what exactly makes a game having multiple viable strategies instead?

• The total number of interacting game mechanics which offer the player a variety of options which might lead to one or more victory conditions.


• The total number of simultaneous victory conditions, AND how loose tight or loose the constraints on achieving each victory condition are, at every step through play.


• Designers ability to recognise different strategies, and impose different pros and cons thereon.

More mature mods, and more recent games, tend to have more attention applied by the designer to (dynamic) balancing, as collective human knowledge of what makes good game design, grows.

Any early game prototype, in the absence of attempts to balance it, will be horribly unbalanced. The more dynamics we introduce, the more potential paths we give the player, which may offer opportunities to balance better; however this still doesn't imply the game is more balanced, just that there are now more choices through which the player might succeed over other strategies.

DoTA is the elder of the two games (second only to AoS) in the MOBA genre, and I would say that this is why it retains the more primitive "pick before you start" design.

As seen in AoE, Warlords Battlecry III, Rise of Nations, Total Annhilation and indeed many in the RTS genre, it's fairly essential to give players a range of different outcomes that can lead to victory. Any game that doesn't do this is, in the first place, narrowing its player base to those with a skillset optimal to achieving that One Metric of Success, for example, twitch-based first-person shooters.

Crucially, designers (should) want to support different play styles. This means allow something like the X-beats-Y-beats-Z-beats-X type cycles already mentioned here, although that graph can be a lot more complex than just a three part XYZ ring. Alternatives are the hallmark of good gameplay.

But it all depends on the designers' target market. If DoTA's target is older players who have been playing MOBAs for 15+ years happy, then they wouldn't change the dynamics, would they? The key for them might be (and probably is) keeping the older, loyal fanbase happy.

Given a non trivial game it takes time/work to find a winning strategy.

Once a winning strategy has been found other strategies can be adapted to counter it.

If adaptations are found the strategy is considered weak.
If no adaptations are found the strategy is strong.

Eventually strategies will be "stronger" as they have shown to be able to overcome "weaker" strategies.
The meta consist of every current winning strategy considered strong, but as time progresses more and more will be considered weak.

The meta can then become a single strategy that is strong vs everything or it becomes a set of weaker strategies that counter each other but are strong against all other known strategies.

The meta is just the players thinking there is a stable set of viable strategies.
It has no correlation to the possible strategies that could be.

In a "overbalanced" game new winning strategies are likely to be strong vs old strategies, because lack of time or choices to adapt after the strategy is deployed.

In a "underbalanced" game new strategies are likely to be weak vs old strategies, because time is given to adapt and overcome after the strategy is used.

Dota is "overbalanced" (compared to Lol), everything has a counter, and that counter has a counter etc.
This is largely because the game is hard, unforgiving, unpredictable and unintuitive. Once you are taken by surprise it is too late to recover. Unknown beats known.

Lol is "underbalanced" (compared to Dota) because the game is more intuitive, that is easy to reason about and apply once figured out.
Riot has long removed the design guidelines that makes explicitly it under balanced but it lives on in the game.
Notably (in the past) Lol enforced that it is never wrong to use skills, just not optimal.
Not wrong means that the gap between minimum and optimal skill usage smal. Failing to apply a strategy has less impact in Lol over Dota because of this skill floor.

There two steps to creating a game with many viable strategies:
1. Create a game where winning depends on both players choices
2. Cultivate a community that encourages creative approaches to playing the game

The first part by itself is not a huge issue, Rock Paper Scissors being a simple example, but the trick comes in how it interacts with the second part. For there to be a community that explores the game there needs to be something to explore. In the case of Dota or Starcraft this means lots of moving parts and continuous choices rather than discreet. For example, consider how positioning units in Starcraft can affect the outcome and how that might change if things snapped into a grid. (Not that grid based games can't have depth). All these factors give players a lot of points of control to manipulate.

However, it is also important to give community things that can be communicated in concrete terms. Items to buy in a MOBA or build times in Starcraft are good examples of that. This gives the community a shared language, but can also lead to rigidity if everything can be decided on ahead of time.

The most important thing to realize in all this is that balance in these types of games is mostly a social construct. If you let the community decided that there is only one way to win the game it doesn't matter if that is actually true or not.

This leads into the second step, cultivating a community.

Why is a community important for having many strategies? Because the community is who is actually using the strategies. Using League of Legends patches as an example, the meta changes at that point not solely because of the actual balance changes, but more because the community expects things to change. Suddenly, new ideas are allowed and the meta adapts. You can tell that this occurs when a community behaves as if a balance change was made when it was actually a mistake in the patch notes.

Different communities have different expectations around how the Meta should change, who can change it, how much to adhere to it, and how it should be enforced. The League community has, at times, enforced Meta adherence through toxic behavior, such as afking or verbal harassment, which leads to a very rigid Meta. If you can cultivate a community that doesn't self police so effectively than you can avoid this.

You need to make sure that the unique, the strange, and crazy strategies aren't squashed before they are discovered. Reducing cost of failure, encouraging positive player interactions, and proving places for players (new, old, skilled, unskilled) to practice and experiment can all be tools used to accomplish this. There are a bunch of ways to get players to feel free to explore the game space.

Even in games with ridged metas you can often see what the game could be in the Pro scene. Pro players have the social standing, by way of there skill to be afforded space to break the meta and often do.

So, I think the secret sauce to a shifting meta is a community that shifts the meta, the game just provides the space for it.

League player here, so everything i say about Dota has to be taken with a grain of salt.

League is designed to support several distinct roles/lanes in mind: Support, Carry, Mid, Jungle, Top. While in the recent past the game changed alot and the meta reflected that (No Marksman as carry, funneling a hyper carry etc.), those roles are pretty much consistens and Riot changes seem to aim to keep it this way (e. g. make funneling impossible, so no "jungle support" in mid). But most likely, you dont see support champs in toplane, or mages in jungle etc.
That said, everytime there are changes, that targets a specific type of champ, e.g. tank buffs or an assassin item gets nerfed, you feel those changes mostly in the roles, those type of champs get picked. For example Tanks in Top, Jungle and Support, assassins in mid, maybe top and recently in Support role, thus creating a swing in Meta.

The changes in Dota 2 are mostly soft, trying to balance out the meta, thus creating a plain field, where everything is possible, but because real hard counters exist, hero select is way more important and more strategic. You can't compensate that much with skill (Dota has way less skillshots than League for example), and you have to rely on, that everyone knows, what every other champ does.

This makes it very hard for a meta to establish, as the meta can get countered with maybe 2 or 3 champs. So Valves is interested in keeping champs as balanced as possible.

Riot keep those changes heavy handed, maybe in interest of strong meta swings and thus creating counter-meta and so on. And most likely, its their interest to keep this "Perfect Imbalance". Extra Credit had a great video about this.

You mention hard counter and soft counters. What distinguishes them? The only difference between them is if a counter can be overcome or not.

For example, Rock-Paper-Scissors is a hard-counter-game. Meaning that one option definitively counters another. To matter how much Paper tries, it will always lose to Sciccors.

As another example, if you play an ineffective Pokémon type against a Pokémon of a stronger type, you will be at a disadvantage, but that doesn't mean you can't win. Therefore, it is a soft counter.

Where do we draw the line between a hard counter and a soft counter? That's a really good question. I don't think we can create a definition that everyone agrees with.

Technically speaking, taking +100% damage from an enemy which counters you, and only doing 1% damage to that enemy is a soft counter. If the enemy never hits you and you consistently hit the enemy, you can technically still win.

But this may not be true in context (healers, potions, mana restrictions, health regen, ...)

The question is what is reasonably possible, and that mean you need to define what's reasonable, which is inherently subjective. How impossible must something be for it to actually be considered impossible?

Might I ask, what exactly makes a game having One Strategy To Rule Them All, and what exactly makes a game having multiple viable strategies instead?

Limiting myself to multiplayer versus games, there are two main game incentivizes:

• Getting the best result


• Beating the opponent

While they often coincide, changing the prioritization between the two options will often lead to different gameplay.

Getting the best result is like playing a racing game. It's all about who get across the finish first.

Contrast that to a racing game where the winner is decided by score, which can be increased in several ways (doing tricks, wrecking opponents, a high finish ranking).

Suddenly, there are multiple ways to win the game. You might hang back and therefore not try to cross the finish first, but instead have a clear shot at the opponents in front of you. You might go off-track completely in order to do tricks around the map to score points.

As long as your playstyle gives you a chance at getting a higher score than the others, you are using a valid strategy.

The key to having a varied-strategy-game is to allow for different win conditions (or different ways to achieve a single win condition).

The generalized approach to gameplay is a series of questions that the player asks themselves:

• How can I win the game?


• How can I prevent the others from making me lose the game?


• How can I positively impact my chance of victory?


• How can I negatively impact the opponent's chance of victory?

You may think that the first two questions are the same, but that is not always true. Taking the example of racing games, compare a race where you cannot bump into each other (you can only win by driving fast) and a race where you can bump into each other (you can win by driving fast, or you can make others lose by ramming them).

This is how a strategy is formed. It works in all types of games:

RPG-like games (DOTA, LoL)

• How can I win the game? => Get to the end objective.


• How can I prevent the others from making me lose the game? => Prevent them from getting to the end objective.


• How can I positively impact my chance of victory? => Get better stats, buff yourself. Avoid being temporarily disabled.


• How can I negatively impact the opponent's chance of victory? => Debuff the enemy or temporarily remove them from the board.

Sports games (Rocket League)

• How can I win the game? => Score goals.


• How can I prevent the others from making me lose the game? => Defend your own goal.


• How can I positively impact my chance of victory? => Learn to handle your car better, improve your teamwork, learn tactics.


• How can I negatively impact the opponent's chance of victory? => Run interference on the enemy's team work. Ram them, demolish them, read and counter their tactics.

Part of a strategy is to choose how to tip the odds in your favor. For example, in soccer/football, it's not uncommon for a team that's leading to start playing more defensively. The first strategy (scoring even more goals) is irrelevant as long as they can ensure the second strategy (preventing the opponent from scoring any goals).

If your team is better at defense than it is at offense, it makes sense to play defensively once you have the lead (not before). If you're better at offense instead of defense, it may make sense for you to play aggressively and assume to gain a bigger lead on your opponent that they cannot recover from.

This effectively boils down to a "when they zig, we zag" approach, which is the essence of multi-strategy games.

(Only just realised this is quite similar to @Arcane Engineer's answer, but I'll leave it up as it provides some examples form a different genre)

It's CCG rather than videogame based, but somewhat relevant: I read an interesting comment yesterday from Richard Garfield (who created Magic: the Gathering).

Context: he has a new game coming out in which every deck is unique, generated by an algorithm, so no two players will have the same deck.

Obviously, a lot of people are curious about how this is going to work. People are wondering how you can be sure an algorithmically generated deck will maintain balance between players.

He has a couple of anecdotes including one from an earlier competitive CCG, in which a particular deck building strategy emerged that was deemed to be unbeatable. There were repeated calls for elements of the game to be nerfed, since in competitive games anyone able to implement this strategy was guaranteed to win. They decided to leave it as it was. Then at the world championships some players suddenly unveiled a tactic which successfully countered the 'unbeatable' strategy.

Even the game designers didn't know whether a counter was possible. But they chose not to reduce the complexity of the game , and to leave it up to the players to find a solution.

To add to the discussion above as someone who plays both, league items and dota items are designed very differently. In leagues most items are bought for their passive stats while in dota items are bought for their active abilities.

In league this leads to less variation as the optimal item build can be built before hand while in dota it requires player execution and decision making to be effective. This allows the game designers to design differently. In league the designers have to plan the game around the items due to how static they are. In dota they can focus on designing around periods of action where item abilities are used.

Thus in league player decisions are set in stone from the get go if they execute the draft as they should but in dota the draft can be turned around by player decisions.

This is counterintuitive in that making the items more complex makes game balancing simpler, and that in general giving the players more hard decisions to make makes the game more fun ( in a game that isn't hyper casual)

Seems pretty simple...

Different game companies cater to different gamer demographics. Some like the grind, some like the strategy, and a newer thing is for some to get submersed in the game as part of their daily lives.

I also suspect it's easier to make money in a pay to win model in a more linear game with less options, but if all games were like that there would be an uncatered to demographic of gamers who want a bit more than to provide a credit card to win.

Something that might interest you is "Open Ai" for DotA2. Basically it's machine learning for bots. These bots managed to win against the best Human teams (except when they picked a bad hero lineup). So it isn't too unrealistic that even something this complex might be practically solved. Like chess.

Now, there are some concessions made for the bots (they have superhuman reflexes, map awareness and coordination. It's also not played with the full hero pool), but we already see players using some of the bots strategies that might seem unintuitive at first. Like suicidal strategies that generate more value then they cost, because the enemy has to respond and you buy time/space for the rest of the team. Or focusing more on buying regeneration while in lane.

Now we can say the exact opposite. While theoretically bots could solve the game, for players there's always more they could do. So there is a better strategy players just haven't found it yet.

So in theory there always is a perfect strategy, but in practise what matters is how fast your players find it, and how often they change their opinion. I think at the highest level this is increased by more complexity. Everyone can solve 3-wins after 20 minutes. Dota needs millions of ay hours.

Dota so has a interesting patch-strategy. There are some small patches every month, but really there are around 2 big patches per year. These patches shake up the very basics of the game. In the new environment other heroes and strategys shine, and you have to start solving again.

Obviously this has its own problems.