I've been watching some Starcraft 2 games lately and thinking about the design compromises made when designing RTS games.

On one hand you want skill to be a determining factor for victory, especially for multiplayer.
On the other you want newbies to start up easily and be able to have fun.
On the third, because apparently we're an alien in this metaphor, you want some forgiveness. Making one small mistake and thereby losing any chance at victory is not very fun, for the loser or the winner.

SC2 for instance does very well for 1 and rather decently for 2 but seems have got worse and worse for 3.

How do you think when balancing these factors, and any others, in your designs? Relevant videos, articles or other info you find suitable would also be very welcome.

My opinion will probably come across as rather negative on this front, but I don't think SC2 is a good example of 'skill' in video games.

As a background, I was ranked in WC3:TFT 16th and 23rd on two different years in NA on the ladder. I wasn't good enough to go 'pro' but I did regularly play against/with them. I've played a lot of RTSs, and know them inside and out.

The question is, what 'skills' does SC2 actually require? And for the most part its simply clicking speed. Higher APM = higher ladder ranking in SC2. There's very little strategy. Everyone is building the same units in the same order, the same buildings is the same place. The 'optimal' build orders for any given strategy are well known by anyone in gold+, and there are only a handful of viable strategies, deviate from the standard builds and you get rolled.

A metric I like to use is: strategy diversity. A well designed game, will require more complex and more diverse strategies (both in attacking and defending) as you increase the difficulty. A poorly designed game will do the opposite.

Consider chess, playing against your friend you only have to think 1-2 turns in advance and you wipe the floor with them. Grand Masters on the other hand are thinking many more turns in advance, with many more units being utilized (ie. beginners focus on their queen, masters are using every single piece, etc...). There's a real complexity there.

Same goes with real sports. Watch a minor/little league soccer match or hockey game and compare that to the pros, and the complexity and fluidity will be night and day, there's no mistaking one for the other.

Compare that to SC2. Watch a few gold/plat level players, and then watch some pro games. It'll be very hard to spot the difference. The top guys aren't utilizing more complex strategies, they're simply clicking faster. They're still going MMM, they just have 5 more marines at the 10min mark than say the gold/plat player. The gold/plat zerg player will have just as many hives but will have a few less larva and fewer creep tumors.

This is something that the average player does not understand, is not immediately intuitive, and is something that a game designer has to design for. Strategic diversity is a difficult, but rewarding, path. I find (in particularly lately) most game companies stick to having 2 or 3 optimal builds, and simply focus around balancing that.

So that said to your questions, there are two ways I see designers commonly approaching the problem of skill levels and newbie retention:

1st is to take the high road. Generally speaking players don't mind losing if they don't feel cheated, and there is clear and obvious paths for progression. If its clear to the player what and why things went wrong such that they can attempt to correct/improve, losing rarely becomes rage inducing (Atlas Reactor is a good example of this).

2nd and sadly (at least to me) the easier way to ensure noobie retention is to have a high enough RNG factor that noobies will accidentally win from time to time regardless of their skill level (LOL is a perfect example of this). A good-ole skinners box attached to a ladder is more than enough to masquerade as skill for most players; and I can't entirely fault the designers for taking the easier route given how much easier it is to do.

<snip>

So that said to your questions, there are two ways I see designers commonly approaching the problem of skill levels and newbie retention:

1st is to take the high road. Generally speaking players don't mind losing if they don't feel cheated, and there is clear and obvious paths for progression. If its clear to the player what and why things went wrong such that they can attempt to correct/improve, losing rarely becomes rage inducing (Atlas Reactor is a good example of this).

2nd and sadly (at least to me) the easier way to ensure noobie retention is to have a high enough RNG factor that noobies will accidentally win from time to time regardless of their skill level (LOL is a perfect example of this). A good-ole skinners box attached to a ladder is more than enough to masquerade as skill for most players; and I can't entirely fault the designers for taking the easier route given how much easier it is to do.

Thanks for a great reply, have an upvote :)

I'm not that familiar with high level SC2, since I haven't played it myself, so I can't argue your points on that but I do see your point on strategic diversity. Hm, maybe a design that scales something like this?

Level 1: Attack and defense
Level 2: Ranged attack, ranged defense, healing
Level 3: lvl2 + Buffs, debuffs
Level 4: lvl3 + symbiotic abilities

The higher the level the better you can utilize the different segments the higher your resources -> damage conversion.

To dig into the 1st choice a little, how do you make it clear why the player lost?

Well Extra Credits done one really good video about this topic:

https://www.youtube.com/embed/EitZRLt2G3w

I've been watching some Starcraft 2 games lately and thinking about the design compromises made when designing RTS games.

On one hand you want skill to be a determining factor for victory, especially for multiplayer.
On the other you want newbies to start up easily and be able to have fun.
On the third, because apparently we're an alien in this metaphor, you want some forgiveness. Making one small mistake and thereby losing any chance at victory is not very fun, for the loser or the winner.

Skill as the determining factor is usually always going to be the case unless you have large, dominant random elements. The question (as Ryan_001 gets at in his answer) is what KIND(s) of skill you want to be important. Do you want high strategic skill to be paramount? Then you are going to have to design for longer games, larger numbers of options/possibility spaces, and a heavy emphasis on balancing to do away with dominant strategies. Do you prefer tactical skill? Engagement outcome high dependent on terrain, positioning/movement/flanking, troop composition, and ability usage. Something else? Determine what elements reward that kind of skill and design to bring them up in importance.

Learning curve is a trickier one. For one thing, many games are easy to pick up in their basics, but I would argue that this does not necessarily make an easy learning curve. Sure, I may be able to beat simple AI's in starcraft using only marines and siege tanks, but I haven't learned the game by only using those two units. Learning difficulty is usually inverse to game complexity because it entails how much time I need to spend to understand any given game state at a glance. Learning curve is reduced when all necessary information is included inside the game, which is relative rarity in the age of wikis. I remember that the original Starcraft had a miss change for ranged units attacking other units on high ground - but I don't think this was every explained in the manual or tutorials. This significantly raises the learning curve for players who do not know it, but who instead might notice a discrepancy between how they understand a situation (attacking an enemy on higher ground, but full vision and with a superior force, thinking they will win) and what happens (somehow they lose! Why weren't they doing any damage?).

Your final case is highly dependent on that opponent. When i play game I absolutely do want to punish every enemy mistake. Most games have important swing cases that allow for reversals of small (or even large) leads under the right conditions to try and make up for this.

I've been watching some Starcraft 2 games lately and thinking about the design compromises made when designing RTS games.

On one hand you want skill to be a determining factor for victory, especially for multiplayer.
On the other you want newbies to start up easily and be able to have fun.
On the third, because apparently we're an alien in this metaphor, you want some forgiveness. Making one small mistake and thereby losing any chance at victory is not very fun, for the loser or the winner.

Skill as the determining factor is usually always going to be the case unless you have large, dominant random elements. The question (as Ryan_001 gets at in his answer) is what KIND(s) of skill you want to be important. Do you want high strategic skill to be paramount? Then you are going to have to design for longer games, larger numbers of options/possibility spaces, and a heavy emphasis on balancing to do away with dominant strategies. Do you prefer tactical skill? Engagement outcome high dependent on terrain, positioning/movement/flanking, troop composition, and ability usage. Something else? Determine what elements reward that kind of skill and design to bring them up in importance.

Learning curve is a trickier one. For one thing, many games are easy to pick up in their basics, but I would argue that this does not necessarily make an easy learning curve. Sure, I may be able to beat simple AI's in starcraft using only marines and siege tanks, but I haven't learned the game by only using those two units. Learning difficulty is usually inverse to game complexity because it entails how much time I need to spend to understand any given game state at a glance. Learning curve is reduced when all necessary information is included inside the game, which is relative rarity in the age of wikis. I remember that the original Starcraft had a miss change for ranged units attacking other units on high ground - but I don't think this was every explained in the manual or tutorials. This significantly raises the learning curve for players who do not know it, but who instead might notice a discrepancy between how they understand a situation (attacking an enemy on higher ground, but full vision and with a superior force, thinking they will win) and what happens (somehow they lose! Why weren't they doing any damage?).

Your final case is highly dependent on that opponent. When i play game I absolutely do want to punish every enemy mistake. Most games have important swing cases that allow for reversals of small (or even large) leads under the right conditions to try and make up for this.

Interesting points about learning curve. It seems to me some games have "handled" the information overflow by building a wiki/dictionary/atlas/thingamajig into their game too. Where do you think the limit lies for how much information can be crammed into a game without resorting to an in game wiki. My first impression is that that would rather limit the amount, or maybe it would more be an UX challenge, how do you present all the information in a concise and informative way.

Good last point, but I also think it depends on the mistake. Winning by "He went Bag so I went Scissor but he never switched to Rocks so I won" feels aweome while winning by "He happened to take the wrong path once so he lost his army" might feel a bit less awesome. What do you mean by "swing cases"?

I've been watching some Starcraft 2 games lately and thinking about the design compromises made when designing RTS games.

On one hand you want skill to be a determining factor for victory, especially for multiplayer.
On the other you want newbies to start up easily and be able to have fun.
On the third, because apparently we're an alien in this metaphor, you want some forgiveness. Making one small mistake and thereby losing any chance at victory is not very fun, for the loser or the winner.

SC2 for instance does very well for 1 and rather decently for 2 but seems have got worse and worse for 3.

How do you think when balancing these factors, and any others, in your designs? Relevant videos, articles or other info you find suitable would also be very welcome.

Perhaps having a difficulty level control that allows the player a wide range to adjust as they move up in ability ??

Best play for the most players is a major goal, and that includes wanting them to do alot or replay, instead of giving up quickly due to too steep a learning curve and always facing too hard an NPC opponent.

Trying to auto-determine a players performance to give 'Forgiveness' can be alot of difficult work if its to be more than just some simple progress metric system. Making the NPC play less optimal is usually not hard (while making optimal NPC play with multiple strategies or adaptive strategy is the much harder task ).

Flexibility for play style also is good - some players (at times) just want to have fun and not a challenge and cranking down the difficulty to survive to GET to the endgame can be a feature to please more players.

Interesting points about learning curve. It seems to me some games have "handled" the information overflow by building a wiki/dictionary/atlas/thingamajig into their game too. Where do you think the limit lies for how much information can be crammed into a game without resorting to an in game wiki. My first impression is that that would rather limit the amount, or maybe it would more be an UX challenge, how do you present all the information in a concise and informative way.

Good last point, but I also think it depends on the mistake. Winning by "He went Bag so I went Scissor but he never switched to Rocks so I won" feels aweome while winning by "He happened to take the wrong path once so he lost his army" might feel a bit less awesome. What do you mean by "swing cases"?

Yeah, probably mostly a UX challenge. The best way I've run into is just to make lots of tooltips for in-game text. And by swing cases I mean ways for a player who is behind to catch up rapidly. Think powerful board clears or legendaries in Hearthstone or area-of-effect damage in strategy games. Often, they are designed to take advantage of the exact behaviors the leading player will most often use to capitalize on their lead.

Interesting points about learning curve. It seems to me some games have "handled" the information overflow by building a wiki/dictionary/atlas/thingamajig into their game too. Where do you think the limit lies for how much information can be crammed into a game without resorting to an in game wiki. My first impression is that that would rather limit the amount, or maybe it would more be an UX challenge, how do you present all the information in a concise and informative way.

Good last point, but I also think it depends on the mistake. Winning by "He went Bag so I went Scissor but he never switched to Rocks so I won" feels aweome while winning by "He happened to take the wrong path once so he lost his army" might feel a bit less awesome. What do you mean by "swing cases"?

Yeah, probably mostly a UX challenge. The best way I've run into is just to make lots of tooltips for in-game text. And by swing cases I mean ways for a player who is behind to catch up rapidly. Think powerful board clears or legendaries in Hearthstone or area-of-effect damage in strategy games. Often, they are designed to take advantage of the exact behaviors the leading player will most often use to capitalize on their lead.

Tooltips sounds good. Then one can get a bit of information in at a lot of places, instead of massive blobs in a in-game-wiki. Or a wiki for all the lore, if any, for those that's interested.

Ah, alright. I know what you mean, didn't know it had a name but seems reasonable.