If a game is to have loss of progress as a death penalty, then the devs should IMO also remember to follow certain preconditional rules:

1. Death is rarely, if ever, random. Players need to feel that the death is somewhat deserved and reasonable.

2. Loss of progress must not reduce overall experience. E.g. players often accept a loss of up to 10 or even 15 minutes worth of play, but not 1-2 hours.

3. The penalty of poor decisions is only death if the player has first gotten some preceding non-death penalty and then ignored them.

These are just things I thought about right now, I'm sure others can flesh these things out better than me. But IMO, I feel like "Save Game" is just scratching the surface of what a game designer can implement with regards to death penalties.

In my opinion loss of progress is the best death penalty in almost all genres. It gives the player also opportunity to reflect on the choices he made that made the death happen.

But I would like to shift the discussion from saving to the frequency of dying. The key to great gameplay experience is simply make the player very much afraid of dying but very rarely doing so. If the player dies too much and is forced to repeat however big or small portion of the game it kills off the atmosphere almost regardless the genre. Nothing is scary, cute or exciting anymore it's just "same". The developer has huge responsibility to give the player enough subtle hints in each part of the game to keep him on the edge of the seat yet dying almost never. If your players pound their head against the wall in every turn you're taking them through then you've done your job poorly.

Just a good example, can be skipped

I have to give the honorary mention to game I will remember for the rest of my life: Resident Evil 2. I know where I saved last, I know how much I've played since and that much is at stake. Even the amount of saves is limited. The game keeps throwing bigger and badder guys at me without throwing in a safe room with save opportunity. Ammo and health is limited, even inventory space is very limited and the player is facing unknown. Even when you think you've beated the game a new form of the final boss creeps on you as you're on the verge of escape. Beat it? It comes again, even more hideous and powerful. No saving. I was left with very little amount of ammo, minimal amount of health for the last couple of encounters and I felt like I was "this close" to dying the entire latter half of the game.

I finished the game perhaps dying 1 or 2 times in the beginning while I was still learning the controls and the genre but I didn't die in the end of the scenario B which was the most intense one. After I beat it I buried the game deep in my drawers and swore never to touch it again. It was way. Too. Scary.

I've seen so many horror games failing big time at this. Especially some of the indie titles can be hopeless with only 1 or 2 developers playtested the release during development. To a fresh player the situation is not all so apparent and every time the road branches there's no telling what the alternatives are. Regardless on if you just want to make quick progress or want to explore everything, you are forced to pick at random and have 50% chance of doing the exact opposite you wanted. New player could be stuck on some part not realizing something could be broken, moved or picked up because that is obvious to the guy who made that entity and did the playtesting. This is a serious problem also the commercial games suffer from due to lack of time to playtest and strict schedule to begin with. Even the scariest boss loses its appeal after you've watched his opening roar 50 times and tried everything you can imagine to the point it where the process begins to resemble glitch hunting.

When the character dies the immersion breaks right there. Repeated dying and/or if you throw him right back in the situation where he was at it makes the player not care about dying thus making any future immersion even harder to establish.

For that reason I also dislike autosave and quick saving. Why do you need autosave? Because you've made the next portion too hard for the player, ie he is likely to die. The player should have enough time within the encounter to learn the boss patterns and how to beat him. It shouldn't take the player X amount of deaths. For replay and experienced players there should be level difficulty options that tailor the experience accordingly.

Why does the player use quicksave? Because he expects to die. That is never a good thing to immersion either. Quicksaves are also used to do all kinds of trolling within the game and trying to play too perfect (will this hurt me? oh it does, reload save) all of which just breaks the experience.

TL;DR: Fix the problem of how to deal with death and saving by minimizing the amount of dying you'll put the player through.

I think that planning in general is here to produce a recipe that decreases the difficulty of executioning action.

Sometimes the execution is no challenge in all the cases and lies only in planning, for example chess- it is not challenging to pick a figurine and move. But to have a plan (recipe)...

Oposite this lets take spear throwing competition is it bad plan to throw 1 meter farther then the others? The challenge is in execution. But this is extremity that also dont have too that much to do with combat as we can only influence only ours training.

Lets move to the Ganryu Islad beach 13th April 1612. Being Musashi what recipe can we produce to ease the ultimate goal? Place is defined and even choice could not help us probably. What is Kojiro known for? Proudness, long reach of his sword, accuracy etc. What if we know how the certain emotion(pique from delay) affects behaviour (execution)? Maybe all we need than is to have a longer reach (that is not apparent(wow place can help us with that -water to waist), and being tiny bit slower than Kojiro with slice but thanks to different timing coming from different reach that should be enough.

To me it would be motivating to use planing to be less wounded and to able to go wider corridor that those who don´t plan. Players should be somehow warned before entering their relative death ground (how about pet pulling them back, having a vision or voice warning).

Regarding planing players should be able to get voluntarily information about their enemy or their goals. For example in Witcher book Geralt paid lots of money to guys that were searching information for him.

The point of my earlier post is that if the player knows that decisions are permanent and could result in at least a loss of time, then they will not be so eager to see if they can survive a 30ft drop off a cliff. Most games allow the player to save at any point in the game. This gives the player SOME invincibility, giving them the "go ahead" to do things a normal person would say is stupid. Even something as simple as trying to steal something without getting caught. Come on EVERYONE has saved the game just before stealing a cool sword just in case someone notices. This means that if you are caught, just reload and try it again when you think it's safe. This gives the player an "out" to the realism. If you take that away from them, then you can make the reward for the risk even more significant. I mean, when I played Oblivion, the first thing I would do is steal everything not nailed down. If I got caught, I would reload and try again. Eventually I had stolen enough stuff to get anything I wanted..... Kind of made it boring after a while.

Yes, but carma is kind of like "being caught" automatically-- not very realistic IMO.

I think of it more as consciousness. And it's realistic because stealing is wrong whether you are seen or not. :)

Players are also given very few options to handle the problems they are presented with (so imagination isnt required).

Yes, that's a problem. How do you make a game that isn't so "procedural"? One would have to almost create an entire universe with all its laws and let it go. "All its laws" means from gravity to love! This is almost impossible to do in reality and truly impossible with the computers we have today. We have to make compromises. Sure, do the common things that computers can easily do like gravity, but when it comes to interactions, there are no simple equations to help us out.

I think this is the biggest question a game designer has about a game (RPG): How do you make the player feel as though his game experience is unique?