next »

[Alevice]

Honestly its usually betrter to just work on a total conversion for a blizzard game. Easy enough to mod and theorycraft

[s0]

I have dabbled about the possibility of a slower paced rts before. The truth is that the dune formula doesnt lend well to interesting slow-paced designs. There have been genre deviations/reinventions as of late that peek more closely to wargames that could float such a boat. Either case still relies on some havily complex systems that are still quite a challenge for a small group of developers, specialñly for a flourishing multiplayer ecosystem.

Real time is actually something quite different from turn based, even if speed isn't a factor. Getting continuous information and having the freedom to act at any time can be very interesting in a slow game. A few examples being Defcon, Eufloria and Firefight, which play out in real-time but where you are still able to take your time in making decisions (usually).

But limiting player actions doesn't really sound like a good solution, as new players won't be able to correct their mistakes.

ok i guess i was too hasty in my dismissal. defcon and eufloria are actually pretty good

[Siilk]

Some reasons an indie dev might not make an rts:

Rts games typically require either very sophisticated AI or netcode, neither of which are commonly within the scope of an indie game. Short of a very novel design, rts games require quite a bit of work on pathfinding, which is typically something indie devs avoid.

To be good/rewarding, rts games require a great deal of testing and balance -- which is typically outside of the scope of an indie game. To *feel* good and be engaging, you can't just depend on snappy fx and screenshake in an rts.

QFT. Decent AI and netcode are cornerstones of an RTS and this things are hard to do right even for a large teams of seasoned developers, not to mention a 2-man indie studio.

And I guess this

Honestly its usually betrter to just work on a total conversion for a blizzard game. Easy enough to mod and theorycraft

explains the rest. It's easier to make a TC or mod for an existing RTS than build your own engine from scratch. And with games like SC/SC2, CnC generals, Star wars empire at war, Total War, Homeworld etc etc being readily available for modding due to huge modding communities, there's almost no need for making an indie RTS engine in-house.

So in short, there *are* indie RTSs out there but they are made as mods, not something people make from the ground up.

[pelle]

Real time is actually something quite different from turn based, even if speed isn't a factor. Getting continuous information and having the freedom to act at any time can be very interesting in a slow game. A few examples being Defcon, Eufloria and Firefight, which play out in real-time but where you are still able to take your time in making decisions (usually).

But limiting player actions doesn't really sound like a good solution, as new players won't be able to correct their mistakes.

Exactly. Real-time isn't really about fast-time. Also adding delays when ordering units is just adding a bit of realism.

There are already some good slow (in comparison to typical rts) real-time tactics games like the Close Combat series or Flashpoint Campaigns: Red Storm.

Or you could try some of the real-time naval wargames like the old Harpoon games that you can play in real real-time, although most scenarios would take days or weeks (months?) to complete unless you hit the speed-up button a lot.

Was just thinking about mixing some ideas from those games with a more traditional rts-style, to make something that is still a silly mix of strategy/tactics (the scale in rts-games is so far off it is at best funny) and just fast enough to not be boring, but with focus on strategy rather than on clicking fast.

My point anyway is that the benefit of being indie is to be able to go outside of genres and experiment. If you don't like my exact idea (maybe I will not even like it if I try to implement it) but anyone here could surely come up with some other variation on the old worn out Dune 2 style of traditional rts and turn it into something that is both more interesting and more easy to implement. That is pretty much what everyone here do to all genres? So maybe the hardcore fans of rts-games will hate all every imaginable indie-rts, but I don't think they would be the target audience anyway?

[oodavid]

I love RTS games but think there are so many sub-genres it gets a bit difficult to get everyone on the same page, a recent bastardisation is the really real time strategy games that make you wait hours, days or even weeks to complete projects - the motivation behind this is well documented to be manipulative but there are useful things to learn from their approach.

You could always mess about with pacing by having a slow-motion powerup that allows you to make more decisions in a shorter space of time, probably a no-go with multiplayer but might be worth a tinker. Maybe in multiplayer you could have the players take turn to choose the gameplay speed. Who knows.

But yeah, I think the reason behind the lack of indie games is that to have a realistic shot at it you need to take a simplified sub-genre.

[s0]

I love RTS games but think there are so many sub-genres it gets a bit difficult to get everyone on the same page, a recent bastardisation is the really real time strategy games that make you wait hours, days or even weeks to complete projects - the motivation behind this is well documented to be manipulative but there are useful things to learn from their approach.

these games are actually ancient. browser mmo strategy games like ogame and travian have been around since the late 90s/early 00s. the timers exist to make these games fair in a massively multiplayer setting, otherwise they would be unplayable by anyone who doesnt have the time to play games 24/7. the idea to tie them to monetization schemes came later (these games were completely free for a long time).

[oodavid]

I love RTS games but think there are so many sub-genres it gets a bit difficult to get everyone on the same page, a recent bastardisation is the really real time strategy games that make you wait hours, days or even weeks to complete projects - the motivation behind this is well documented to be manipulative but there are useful things to learn from their approach.

these games are actually ancient. browser mmo strategy games like ogame and travian have been around since the late 90s/early 00s. the timers exist to make these games fair in a massively multiplayer setting, otherwise they would be unplayable by anyone who doesnt have the time to play games 24/7. the idea to tie them to monetization schemes came later (these games were completely free for a long time).

Valid point, well presented. My gaming knowledge isn't really that broad (nor deep for that matter).

[starsrift]

I often wondered this very same thing.

Then one day I tried to implement my own very basic chess AI.

Now I know why so many developers pass on these types of games. 

It sounds weird, but a chess AI is more complex to write than StarCraft AI. How could a 32 piece, 64 square game be harder? The element of turns make foresight and flexibility crucial.

Star/Warcraft / C&C is mostly an event polling system, with a task backbone.

[Netsu]

I often wondered this very same thing.

Then one day I tried to implement my own very basic chess AI.

Now I know why so many developers pass on these types of games. 

It sounds weird, but a chess AI is more complex to write than StarCraft AI. How could a 32 piece, 64 square game be harder? The element of turns make foresight and flexibility crucial.

Star/Warcraft / C&C is mostly an event polling system, with a task backbone.

I'd say it's easier to write AIs for RTS games specifically because they're real time. In chess you can carefully evaluate your every move and to beat a human player the computer must just be better at strategy. In let's say Startcraft on the other hand, the computer easily get's the advantage of being able to issue orders instantly and to multiple units in many places simultanously, and can have instant absolute knowledge of enything that happens within his units line of sight.

On the other hand, chess will be most likely 'solved' sooner or later (computer scientists will establish an unbeatable strategy), it already happened to checkers, while Starcraft has too many factors to be even approached in this way.

[Dacke]

I'd say that it's far easier to write a good Chess AI. Just learn some of the principles of tree searches, alpha-beta-pruning, state hashing, opening books and write a decent state-evaluator and you're set. You may need to read some theory about trees, but it isn't all that hard.

Good RTS AIs on the other hand don't really exist. There have been some competitions for SC1 AIs, but they are usually 1-trick-ponies and (last time I checked) don't stand a chance against a pro-level player. This despite being able to pump out an APM in the thousands.

Writing an AI that can beat a new/intermediate player in either game isn't really that hard, though.

On the other hand, chess will be most likely 'solved' sooner or later (computer scientists will establish an unbeatable strategy), it already happened to checkers, while Starcraft has too many factors to be even approached in this way.

No, as far as I can tell chess will most likely never be "solved". The search space is just too mind-bogglingly big. Unless someone figures out something cool with quantum computers or some really clever lemmas.
http://en.wikipedia.org/wiki/Solving_chess

[Netsu]

I'd say that it's far easier to write a good Chess AI. Just learn some of the principles of tree searches, alpha-beta-pruning, state hashing, opening books and write a decent state-evaluator and you're set. You may need to read some theory about trees, but it isn't all that hard.

Good RTS AIs on the other hand don't really exist. There have been some competitions for SC1 AIs, but they are usually 1-trick-ponies and (last time I checked) don't stand a chance against a pro-level player. This despite being able to pump out an APM in the thousands.

Writing an AI that can beat a new/intermediate player in either game isn't really that hard, though.

Interesting, I always assumed that RTS cpu players are only beaten by exploting the rules by which they react to various developments, but I am easily beaten by any basic AI in both chess and rts games, so how would I know

No, as far as I can tell chess will most likely never be "solved". The search space is just too mind-bogglingly big. Unless someone figures out something cool with quantum computers or some really clever lemmas.
http://en.wikipedia.org/wiki/Solving_chess

Never read much into it, but the search space for checkers also seems way too big to just go through in it's entirety in any human timeframe, so I assumed they somehow generalised and combined move dictonaries with heurestics proven to be optimal behaviours.

[Dacke]

Here is one SC1 AI competition:
http://webdocs.cs.ualberta.ca/~cdavid/starcraftaicomp/

RTS games pose a much greater challenge for AI research than chess because of hidden information, vast state and action spaces, and the requirement to act quickly. The best human players still have the upper hand in RTS games, but in the years to come this will likely change, thanks to competitions like this one.

(my emphasis)

A ton of research has already been done in the realm of perfect-knowledge games and tree searching. The techniques are well known at this point, making it "easy" to hack together a decent chess AI. On the other hand, RTS games can be played at a super-human speeds so we should see better-than-pro AIs once enough research has gone into it.

Checkers is estimated to have a search space that is the square root of chess, which is vastly smaller. If you'd need to search a million states in checkers, you'd need to search a million million states in chess. The person who solved checkers (using 2 decades and hundreds of computers) said that you'd probably need quantum computers to solve chess.

It's expected that in order to solve chess you'd need:
1,000,000,000,000,000,000,000,000,000,000,000,000,000,000 states in a dictionary
or 1,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000 variations tested

[Oskuro]

Writing an AI that can beat a new/intermediate player in either game isn't really that hard, though.

I think the problem with AI programming in indie games is not one of difficulty, but of focus. In order to develop a decent AI for a game, you have to focus on treating the player as another variable in your game, making your game "predictive", while more standard games are more "reactive", that is, the game plainly responds to player input.

I personally like AI programming, and still don't feel compelled to develop AI-dependent games.

[Dacke]

Indeed. A fun AI and an AI that can beat you are often two very different things.

edit: Sorry if I contradicted you by saying it's "easy". It requires a lot of studying and work to get it right. I was mostly talking in relative terms; by "easy" I simply meant "not hopelessly hard".

[ஒழுக்கின்மை (Paul Eres)]

another big reason is that rts games generally require multiplayer to be fun, and most indies shy away from multiplayer because they are rarely popular enough that players would find other people to play against

[Netsu]

So for checkers it would be half the zeroes, that's still billions of billions of states. Even just reading through them in a dictionary would take a 1000 1GHz CPUs over 20 years (assuming every operation takes only one CPU cycle). Actually computing perfect responses for each of them would take much, much longer.

Not that it matters much, but now that I took the time to calculate it I had to post or the time would feel totally wasted...

[s0]

another big reason is that rts games generally require multiplayer to be fun, and most indies shy away from multiplayer because they are rarely popular enough that players would find other people to play against

also obv server infrastructures are expensive and p2p is not exactly optimal. i mean look at how many issues online MP in nidhogg has.

[Netsu]

another big reason is that rts games generally require multiplayer to be fun, and most indies shy away from multiplayer because they are rarely popular enough that players would find other people to play against

also obv server infrastructures are expensive and p2p is not exactly optimal. i mean look at how many issues online MP in nidhogg has.

P2P is more than optimal actually, whatever issues nidhogg has is unrelated.

But networking RTSs is in itself more difficult than networking other games, because you can't really revert to syncing game states if client-side prediction goes wrong.

[1kW]

Indeed. A fun AI and an AI that can beat you are often two very different things.

Depends ... In terms of RTS a fun & strong AI is possible - just limit apm of the AI to the actual apm of the human player.

[s0]

another big reason is that rts games generally require multiplayer to be fun, and most indies shy away from multiplayer because they are rarely popular enough that players would find other people to play against

also obv server infrastructures are expensive and p2p is not exactly optimal. i mean look at how many issues online MP in nidhogg has.

P2P is more than optimal actually, whatever issues nidhogg has is unrelated.

But networking RTSs is in itself more difficult than networking other games, because you can't really revert to syncing game states if client-side prediction goes wrong.

i thought p2p becomes problematic/laggy when you're not playing against someone geographically close to you? (or at least thats my experience with p2p games)