XCOM Inspired Fantasy Game

[Nikita_Sadkov]

Sure, let's find the worst possible implementation and make it a strawman against the feature.

I never did and never will play games with no randomness. At least not the kind which emulates some reality. No randomness means no uncertainty, and therefore no life or immersion. Such a game is dead, dead and dead. It has no appeal. I'd rather play tic tac toe, at least it doesn't fail miserably at pretending to portray some sort of setting.

It doesn't matter how good your game is if it's dead.

Then why people play Starcraft?

[Solarius Scorch]

Because it's a e-sport, I suppose. It's not meant to be immersive, or actually simulate anything.

It's basically chess with fireworks.

[Nikita_Sadkov]

Because it's a e-sport, I suppose. It's not meant to be immersive, or actually simulate anything.

It's basically chess with fireworks.

But a lot of randomness arises from player actions. Starcraft's deterministic rules coupled with unpredictable user input are as good as random, but without the nasty parts of the random. Lets not forget, that in reality there is no such thing as "random" - it is just a buzzword for describing complex things, we don't understand. As Einstein said "God doesn't play dice". What appears to be random to us is just some complex high frequency process, which our brain can't handle, so we see white/brown/pink noise.

[ohartenstein23]

As Einstein said "God doesn't play dice". What appears to be random to use is just some complex high frequency process, which our brain can't handle, so we see white/brown/pink noise.

Einstein said this in response to an interpretation of quantum mechanics because the math suggested the universe is not in fact deterministic, it just looks that way when you're in the realm of things that don't behave all that quantum mechanically. It's the opinion of a man who didn't like the interpretation of a new theory that is now taught as university chemistry and physics.

The blanket statement that the universe isn't random cannot be made definitively when the underlying math we're using to describe it is inherently about probabilities, not definite motions.

In terms of how this applies to game design, Starcraft can be interesting when you don't know what the other player is going to do. Without that knowledge, the inherent unpredictability could be modeled as a random process. It gets less interesting when "meta" strategies evolve and play becomes predictable again because the same input strategies will get the same counter-strategies, leaving the result down to hoping your opponent makes a mistake. Even then, you don't know if or when the other player will make a mistake, and an unpredictability will determine your outcome.

If you make a game without multiplayer, the interest is generated by the computer decision-making against you. To hold the player's interest, you either have to make something deterministic be suitably complex so that no player maps out the inputs to behavior and you get "one way to play," or you can make the computer opponent and situations unpredictable by adding complexity modeled by randomness. I personally wouldn't want to do the first method as that means I'm assuming my intellect is enough greater than my players' that they won't ever solve the deterministic problem I put before them. Not that I'd want to use the randomness as a cop-out either, but it does give a way to hold the interest of players longer.

[Solarius Scorch]

Lets not forget, that in reality there is no such thing as "random" - it is just a buzzword for describing complex things, we don't understand.

Sorry but when did you learn physics, in 1860s?
While some physicists today admit the Universe may be deterministic, even they unanimously agree it's meaningless, since other laws of physics prevent us from fully knowing the initial conditions. Therefore, for all intents and purposes, it is not deterministic.

A situation where two units fighting one another under the same conditions always lead to exactly the same result isn't even funny, it's a mockery of simulation.

You keep saying Starcraft this, Starcraft that, while constantly ignoring the elephant in the room: Starcraft is not a simulation game, it is - as I said before - glorified chess. In chess, both participants have access to 100% information and the game is about measuring your analytical penis against your opponent (who can predict the next steps better), which is a completely different kind of activity and has zero relation to real world, which is entirely chaotic.

But all this can be my misinterpretation. I was under the impression that you wanted to make a civilization game, since you keep referring to titles like Master of Magic which are such titles. If your goal instead is to make another chess variant, with closed information and only dependent on thinking X moves ahead, then you should have said so from the very beginning and I wouldn't bother getting excited about it.

[Yankes]

Another thing Starcraft have AGI component too, without decent APM you will be out maneuvered and lose more units than you should.
but

which is a completely different kind of activity and has zero relation to real world, which is entirely chaotic.

I disagree, if it would be entirely chaotic then you would not be able do do ANY thing, any planing, any strategy, any living. Big class of problems in real life can be solve by skill set similar to one that chess use, why? Because knowing EVERY thing about chess board state is impossible, plan 10 move ahead and you start risking running out of atoms on earth to store all possible states. You need only work on very small subset of possibles and you could miss one critical path.
Good players probably play knowing what they do not know, and place they figures in defense positions from some unexpected threads.
And this is skill that can be easy used in normal life. You can expects some groups of behaviors that can happens in real life, and knowing that some things are not know to you. With this "figures" you can plan your "moves".

[Bobit]

Solarius, hidden information is randomness. If two players play The Prisoner's Dilemma or  something like that, optimal play for them is to do something X% of the time and another thing X% of the time. Which may not seem like a big deal but in complex games (Stratego maybe for example, though as a mainstream board game it's not great) it leads to pretty much the same randomness as true randomness. Starcraft therefore has at least a small amount of randomness just due to fog of war.

I mean obviously no one's smart enough to solve Chess so it still has diversity. And even if Soccer/Football and Starcraft/Dota were "solvable", it would depend on a person's capabilities and skills at different aspects of the game, which is pretty realistic.

However, turn-based single-player games are much better off with "true" randomness, if only AI randomness unless they are handcrafted puzzle games, but I think those get boring after a few hours.

Also, physicists tend to take positions that are helpful to them and look pretty harmless, like Schrodinger's Cat being taken as serious evidence that two impossible things are true at the same time rather than simply not being able to know which one is true, but are philosophically very impactful. Though I don't want to dig further on that in this board.

[Nikita_Sadkov]

But all this can be my misinterpretation. I was under the impression that you wanted to make a civilization game, since you keep referring to titles like Master of Magic which are such titles. If your goal instead is to make another chess variant, with closed information and only dependent on thinking X moves ahead, then you should have said so from the very beginning and I wouldn't bother getting excited about it.

My design idea is closer to https://en.wikipedia.org/wiki/Dark_chess and https://en.wikipedia.org/wiki/Fischer_random_chess

I.e. there are random starting conditions (i.e. map is randomly generated), in addition to fog of war, acting like your typical pseudo-random number generator. AI is allowed to throw a dice in its decision processes. Then there are spells like "mist", which obscure vision. But the idea is that player can choice a strategy to reduce randomness. For example, player can build more scout units.

[Solarius Scorch]

butI disagree, if it would be entirely chaotic then you would not be able do do ANY thing, any planing, any strategy, any living.

Well, I should have said "entirely chaotic within certain parameters". Though technically it is as you said - anything can happen anytime. The chances are just astronomically low. But that's nitpicking and is not really related to the subject.

Big class of problems in real life can be solve by skill set similar to one that chess use, why? Because knowing EVERY thing about chess board state is impossible, plan 10 move ahead and you start risking running out of atoms on earth to store all possible states. You need only work on very small subset of possibles and you could miss one critical path.

Yes, of course. But I can't remember ever talking about skills, I am talking about modelling the game world.

Good players probably play knowing what they do not know, and place they figures in defense positions from some unexpected threads.

Sure, all players are limited in this regard, which is why playing against each other is possible. But I don't think it relates to the subject.

Solarius, hidden information is randomness. If two players play The Prisoner's Dilemma or  something like that, optimal play for them is to do something X% of the time and another thing X% of the time. Which may not seem like a big deal but in complex games (Stratego maybe for example, though as a mainstream board game it's not great) it leads to pretty much the same randomness as true randomness. Starcraft therefore has at least a small amount of randomness just due to fog of war.

I mean obviously no one's smart enough to solve Chess so it still has diversity. And even if Soccer/Football and Starcraft/Dota were "solvable", it would depend on a person's capabilities and skills at different aspects of the game, which is pretty realistic.

However, turn-based single-player games are much better off with "true" randomness, if only AI randomness unless they are handcrafted puzzle games, but I think those get boring after a few hours.

Sorry, but this completely not on subject, at least not what I was discussing. Why is everyone trying to explain chess to me now?

Also, physicists tend to take positions that are helpful to them and look pretty harmless, like Schrodinger's Cat being taken as serious evidence that two impossible things are true at the same time rather than simply not being able to know which one is true, but are philosophically very impactful. Though I don't want to dig further on that in this board.

Yeah, me neither, though I know why this thought experiment was conceived and that it was pretty much criticism of certain aspects of quantum theory rather than support of it.

My design idea is closer to https://en.wikipedia.org/wiki/Dark_chess and https://en.wikipedia.org/wiki/Fischer_random_chess

I.e. there are random starting conditions (i.e. map is randomly generated), in addition to fog of war, acting like your typical pseudo-random number generator. AI is allowed to throw a dice in its decision processes. Then there are spells like "mist", which obscure vision. But the idea is that player can choice a strategy to reduce randomness. For example, player can build more scout units.

Fair, and I certainly won't criticize you for that. Just saying it's disappointing for me personally, because I don't care about this kind of games, and I thought your project was something different. But that's entirely on me.

[Bobit]

Yeah I clearly jumped into a conversation without reading the context lol

[Nikita_Sadkov]

Fair, and I certainly won't criticize you for that. Just saying it's disappointing for me personally, because I don't care about this kind of games, and I thought your project was something different. But that's entirely on me.

Yeah. I've decided to take "no explicit PRNG" rule as a challenge, similar to "silence vow" monks take. Then again, there are many explicit PRNG strategy games, while almost no true deterministic ones (beside Chess and Go), and I'm trying to distance my game from the mainstream a bit, so I'm looking for obscure features and try to insert them into the game. Like I've adapted numerous ideas from the Lords of Magic.

As I mentioned, adding random chance is easy. It is a game design shortcut. But implementing actually believable world without any Deus ex Machina is hard. Because then everything have to be explained somehow using in-game rules. Randomness is similar to God. It is really simple explanation and works for most purposes. I.e. card games are completely chance driven, yet still believable for people with imagination. But then there are autistic people like me with poor imagination, who need everything detailed, who will ask what source of entropy God used for creating that exact star system, what clothes he wore, what hair style he had, and the magic just fades away...

[Solarius Scorch]

A very good point!

I totally get what you mean. For me a game as it is is only a part of what is going on in my experience, the rest is in my head. The game itself is a feeder which provides material for interpretation, this is simply what I like to do in games. So kudos for understanding that, even if you chose a different path.

[Nikita_Sadkov]

I think the game needs a better water shader. The current static blue water is really ugly. Something like the following would look orders of magnitude better.

[Nikita_Sadkov]

Further work on map generator. Now it creates cliffs and waterfalls. A lot more work is required to make it more general. I also need to implement proper city generation.

I've considered using something like wavefunction collapse map generator, but it is really slow, hard to control, and does basically the same I already do, albeit within a nice framework. Map generation is basically a pathfinding process inside some really complex space, but its goal is not getting to some precise point, but getting to a good enough point. Give that, you can apply all the genetic algorithms, from markov chains and neural nets to simulated annealing. Then again, when you generate your map from some predefined parts which have restriction on how they connect, you have a markov chain. The problem is that no generic algorithm would create a fun and non totally broken map, so a lot of manual workaround will be required in any case.

For example, I need a way to describe a branching river or a an island inside a river, or a river with a pond. Then there are roads and bridges interacting with them. All that stuff should be placed properly - no circular rivers or crazy roads: if the there is a cliff on the way, road should go around it. No idea how to easily describe that with say markov chains or WFC. ANNs are nice for classifiers and simple filtering, but not for complex structures, which should interact with game mechanics nicely.

[memmaker]

Just wanted to chime in with an interesting note:

I believe randomness can be used to varying degrees. The less randomness and the more deterministic a game is, the more it resembles a puzzle instead of a strategy/tactics game.

Taken to the extreme, this concept can be seen in "Into the Breach" (https://www.youtube.com/watch?v=SZg8GAnjgyc).

A very interesting little game, which I would definitely put in the realm of puzzle games, even though it looks like turn bases tactics. Every move is completely deterministic and you even get a preview of what the AI will do next turn. Interesting concept.