Hello dear FAF player,
today i want to ask two questions:
Question 1:
Do you want to see FAF putting work into making FAF awesome for the average casual RTS player?
According to the video "The Next Major RTS Will Fail. This Is Why." (highly recommended), around 77-80% (!!!) of Starcraft players are "casual" players (players that only played the campaign and never touched multiplayer). Starcraft is one of the most competitive RTS out there, so we might assume that the ratio is even more imbalanced for Supcom FA.
So my personal answer is YES!
More people means more ideas and creativity, more work put into FAF, more fun, more donations, more everything! And some of those casuals will naturally become salty, competitive tryhards by the nature of this community, providing more spice to the flow!
Question 2:
If yes, how could that be done?
Well, to know that, we have to forget ourselves and put ourselves into the shoes of a casual player. Luckily i am somewhat casual. You want proof? I only play bot matches in Dota 2 and I tried playing that game with a controller. I still like to play the FAF campaign from time to time even though it is way too easy for me at 1100 global rating. Hopefully that qualifies me as casual enough!
However, if you are even more casual than me, i want to hear your opinion!
Now lets talk about what causal players want. The key difference is quite easy to understand:
Casual players do not want to compete, casual players want to progress!
Chapter 1: Difficulty
The idea of creating an auto-balanced queue for casual players, where players win/loose 50% of the time thanks to Trueskill like in TMM would be stupid. Why?
Because casual players to not really want to be challenged by a competitve opponent. Casual players want to instead have a fun and satisfying gaming experience, which for them means having a feeling of PROGRESS, and while this can include personally improving at the game, it does not need to.
However, how easy exactly should it be? And the answer is different for every player. Some players can stand loosing against opponents more often than others. This is why in gaming there is the very popular concept of "game difficulty". Difficulty allows players to decide what the "skill gap" between them and their AI opponent should be.
This skill gap preference is different for every casual player! There are games that do not have difficulty settings, which are usually explicitely marketed at "hardcore" players like Dark Souls. However even Dark Souls has built in "difficulty" by allowing players to farm stats and play in Coop. I know it because it had to farm against the O&S boss, because I would have been too bad to beat it otherwise.
(difficulties in AoE2 DE)
Chapter 2: Progression
So our casual player selects a difficulty, the game should not stay that difficult the entire time. There should be a diffiulty progression that the player can keep up with. This is straighforward to handcraft in a RTS campaign, but how does difficulty work in RTS skirmish matches against AI?
Well, the progression in a skirmish against AI is to select a harder difficulty at some point. Lets say i played against Dota 2 bots at mid difficulty, at some point i might feel like im skilled enough to turn it up a notch and try playing against hard difficulty. The turning of that nob, if i win, is what is satisfying as a casual! This is why RPGs with auto-scaling difficulty are hated by quite a bunch of players, but thats a different story.
Chaper 3: FAF custom games vs. AI
Now that we know that the casual player desires to set the difficulty in a vs-AI game, why is this a problem in FAF? Don't we already have Sorian with various difficulty levels and a bunch of alternative AIs?
Well, the problem is that selecting a difficulty in FAF sucks. In fact, it is hardly possible for a casual player:
- The actual new good AIs are hidden in the mod vault and lengthy forums post, where a certified casul does not ever dare to look
- There is no readily accessible information about how good any of the AIs really are
- Even the AI authors do not always know how good their AI is
- AI performance changes as balance patches and AI patches hit, causing the performance of AIs to swing
- Some AIs are somewhat broken on some maps, the whole AI marker situation is a thing that casual players should not have to deal with - if possible, it should "just work".
- There seem to be no balanced steps in difficulty. How much better is Sorian Adapter compared to AI Easy? The casual player has no good way to find out if he should try Adaptive after Easy or if there is something in between
- The AI developers themselves have no incentive to create modifications of their AI for different skill brackets (nerfing the AI to create an easier AI variant, adding a cheating variant, finetuning the cheat multipliers etc.)
Creating a consistent concept of difficulty for FAF AIs
As a solution to these problems i propose the creation of "AI skill brackets". These are rating brackets that AIs are sorted into and that effectively define what "hard AI" etc. means on FAF. The numbers here can of course be changed, but something like this might work:
How AIs in each bracket should perform expressed in global rating:
- under 200: Very Easy
- 200-400: Easy
- 400-600: Normal
- 600-800: Hard
- over 800: Very Hard
This would allow us to simplify AI selection a lot. A casual user can just select one of these brackets and automatically get an AI implementation that plays acoording to that skill bracket. If AIs get better, they might change their bracket, allowing casuals to have consistent difficulty without thinking about how AI authors change their AIs.
If the casual user wants more control, they should still be able to select a specific AI in a specific bracket. If he doesn't, then just select a good default or a random AI from that bracket.
It should be noted that if an AI is exploitable with a very specific hard to find exploit, this should not be factored into the skill level of the AI. We can assume that a casual player never finds those exploits, or that they can quickly be fixed if found. The goal here is not to give an exact represenation of AI skill, but a consistent system that is usefull to the casual player who doesn't play very often.
The hard part
All of this until this point is just mostly client/lobby implementation work, nothing that i think would be too hard. Were it gets hard, is this:
How do we know which AI belongs to which skill level? How do we measure the skill of AIs? There already are multiple ideas, as expressed in the "AI in matchmaker queues". And there are other ideas like regular AI tournaments that could be used to manually asses AIs capabiliy. However first i want to know:
Should this be the next big project for FAF? Making playing against AIs easy, consistent, and a generrally pleasurable experience? Do you think creating well-defined skill brackets for AIs is a good idea? I think this would substantially improve FAF, but that is just my casual opinion.