Addressing the OP:

What if you want to make a game, any game whatsoever, where combat is involved, and you really want for player to lose some battles. Not that the battles are scripted for player to lose, any given battle is winnable, but rather that winning or losing a battle will direct the story in different directions"

This is the most interesting aspect of the discussion for me: rather than how a save system should be implemented, instead what should happen in the case of a "failure".

Probably the obvious thing to do is just make sure you tell the player that loosing isn't a big deal. But if your game has some particular 'optimal ending' (assuming there is an ending) the a player is likely to do what he can to head straight for it. And if an event occurs that makes that optimal result no longer attainable, he's going to feel disappointed and want to change the outcome of the event even if there may still be an interesting experience ahead of him. If there isn't really an optimal result and the player knows it then the he probably won't be focusing on looking for a specific end but rather on exploring what possibilities are available.

Quoted for sound logic and bolded what I feel is the most insightful part of that post

For instance, game only saves on exit and reloading deletes the save, but you have combat ahead. How do you determine the risk ?
By gathering information about the enemy. Therefore we set a "scouting meter", which will fill as we get more details about the encounter ahead. Someone told you that there is an ambush on the road ahead, and there are likely to be 7 men, 4 of them are likely archers, etc. So the scouting meter is up to 80%, meaning that there is a 80% chance that the enemy will have exactly 7 men and 4 of them archers. So you decide to take your chances in combat, and get defeated. But the amount of struggle you produce gets added to the scouting meter, and fill the "failure meter" which is a bonus you can then spend on getting yourself back on track in some fashion you cant achieve any other way.

So what this does, in my mind, is encourages the player to earnestly try and win every encounter, through preparation, evaluation, and information gathering, and then applying as much of it as possible on the field.

What do you think ?

I've tried hard to "re-factor" that paragraph in my head since I feel that this is a good idea, but I don't quite understand the implementation. Can you think of another gameplay scenario where a similar system could be applied Karnot?

In my opinion, it's a hard problem to address in any conventional game I can think of, and considering how your average gamer these days seems to have a severe case of obsessive compulsive disorder (meaning they can't control themselves from reloading in the case of a less than optimum result) I'd say the easiest way to address this is to make a game where no reloading of prior saves is allowed at all, but there is no true "proper" ending, nor is there a "permadeath", and neither is the game very long.

In which case, I'd suggest a game which features some type of randomly generated scenarios is best, with the outcome of the next depending on the outcome of the previous, or seeded to some particular gameplay aspect, or possibly even totally random in a way that can't be predicted by the player.

Perhaps grand strategy games like Shogun Total War have an aspect of this, since if you are defeated in a particular battle you may loose some territory, but gain an advantage due to the other AI opponents no longer viewing you as the greatest threat, and instead targeting the winner of that battle rather than yourself. That would be an example of a less than optimal result which ends up yielding the player a (not immediately obvious) advantage. Then again that's more along the lines of “gameplay” rather than “story”.

Can anyone think of an example game with vastly branching storylines though? Specifically, rather than each encounter/mission/quest being “able” to be failed by the player, while the player is still able to continue with the “main” storyline, instead I mean that the outcome of each mission truly impacts the conclusion, and not in the sense that “if I did everything bad/good/perfectly then I get the bad/good/perfect ending”.

Or perhaps just examples of sub optimum play that yields a non obvious advantage, or examples of "carrying forward" some benefit in the case of failure – the "game-play" side is probably the easier problem to solve lol

rather than how a save system should be implemented, instead what should happen in the case of a "failure".[/quote]
True. Some people mistakenly thought that the issue with saving is that it breaks the so-called "immersion", which i have nothing against, but this topic isnt about that. Its about breaking the connection between failure and gameover, making failure a viable path to take and keep playing.

I'd say the easiest way to address this is to make a game where no reloading of prior saves is allowed at all, but there is no true "proper" ending, nor is there a "permadeath", and neither is the game very long.In which case, I'd suggest a game which features some type of randomly generated scenarios is best, with the outcome of the next depending on the outcome of the previous, or seeded to some particular gameplay aspect, or possibly even totally random in a way that can't be predicted by the player.[/quote]
As i have begun to actually work on the project, i guess i will go on and be more specific about an actual game i am making.
If anyone remembers the Battletech books, there was once a Gray Legion trilogy. What i want to make is basically "Gray Legion : The Game". (though obviously not using Battletech license or books storyline). A tactical game involving giant robots and other military forces. The game would work on two levels : there is the player - a leader of interplanetary mercenary unit, free agent who can pick his own battles; and then there is a global war - something not even remotely under player's direct control, and having a timer for conclusion. After a set amount of time has passed - the war (and the game) ends, one way or another, that is not in question. The question is in what position will the player turn out to be when the war ends. So there are "endings", but no "winning" for the game. And i want to stylize it all as a kind of future historical chronicles, so every game will tell the story of both the war and the player. Losing a battle will not mean losing a war.

I've tried hard to "re-factor" that paragraph in my head since I feel that this is a good idea, but I don't quite understand the implementation.[/quote]
I was just throwing a draft of an idea, it hasnt formed fully just yet, so i probably wont be able to be as clear as i want it to be, but i'll try.
Let's say that in the game, your primary resource is "money" (note the quotes). And you have two "accounts". If you win - you get "money" put into one account, the one which you can use in any way you wish, including the way to achieve your meta-goal. If you lose - you also get "money", maybe only a fraction of winning prize, but they are put into another account, which you can ONLY use to build back your team's strength, buy equipment and recruit men. The more glorious your victory - the more "money" you get, but also you get more "money" for more spectacular failures.

How i came to this idea was, i thought of player losing a battle. What do i do with him ? Make him start basically from scratch, and give him some access to low-tier equipment and let him grind the money to build up all over again ? Not good. What then ? I want to pull them back into regualr gameplay as soon as possible, but not simply "respawn" the player as if nothing happened. So i started inventing ways for player to do that, for example i thought of having to rent out your spaceship as space cargo truck, and make it all automated, so it will make money for each turnaround. I wanted it to be no more than a couple minutes of real time, but in game time it will take months as a trade-off. And you really have to treasure your time. But i wanted an even faster way. So i thought "why not use the same mechanics for losing, as i do for winning ?". No need for grinding, no need for additional mechanics, losing is exactly the same as winning, only with a minus put in front of it. The consequences are different, but the gameplay is the same.

Can anyone think of an example game with vastly branching storylines though? Specifically, rather than each encounter/mission/quest being “able” to be failed by the player, while the player is still able to continue with the “main” storyline, instead I mean that the outcome of each mission truly impacts the conclusion, and not in the sense that “if I did everything bad/good/perfectly then I get the bad/good/perfect ending”.Or perhaps just examples of sub optimum play that yields a non obvious advantage, or examples of "carrying forward" some benefit in the case of failure[/quote]
Well...
In Super Robot Taisen games, the story is often quite branched, and often the choice of a story branch is determined by amount of "skill points" awarded for special merits in previous levels. Amount of "skill points" is also proportionate to difficulty level, so the more of them you get - the harder versions of the level you get.
Panzer Corps has kind of the same deal, if you perform good - you advance your Nazi army on Moscow, if you are underachiever - you spend the second half of the game in Western Europe.

if a game doesn't let me stop playing when I want to stop playing. I just don't want the game to hold me hostage![/quote]
I have never advocated that. You are fighting a strawman, my friend.

If you quit the game and reopen it, finding yourself in exactly the same place, that is persistence (and pretty much every game should support persistence, though some still don't). Save/load on demand is an entirely orthogonal concept - save/load is one way of providing persistence, but a persistent game need not provide arbitrary save/load (only saving automatically when the player quits).

Moreover, persistence is (almost) entirely orthogonal to game design and game difficulty whereas saving tends to have a drastic effect on how the game is played.
It's not necessarily 100% wrong to make a single player game run in real time without a possibility to pause (and quit in a persistent game is effectively a pause), but I think it would be a good design decision only very rarely. Maybe in some kind of horror game where decisionmaking under time pressure is key.

I like the separation between save/load and persistence. I also like the idea of failure consequences which are neither grinding nor reloading. I think the failure=death mindset is what keeps people hovering over that save button. Perhaps you have two main paths; the hero story and the underdog story.

I'm of two minds over people wanting to replay to see all endings/best ending. Maybe you implicitly save at the major decision points they've encountered, but they can't go back up the tree of save points until they've clocked the current branch? Maybe glorious failure points earn you the choice to go back a certain distance in the tree?

I'm of two minds over people wanting to replay to see all endings/best ending.[/quote]
Well, how is this different from people wanting to replay the game to see the same ending ? For example, grand strategy games, like Civilization, are specifically created to be replayable, but the ending is always the same "your nation is the greatest blah blah blah". Oh, and there might be a short video if you won through space race. So i dont quite see the problem there. People play games not to see the ending, it is simply a way to give closure to the whole playthrough.

You are (i suppose) coming from the mindset where the game is 98% the same no matter how many times you play it, like Bioshock, and one decision in the end is all it takes to get a different ending. Am i right ?