it depends on why you're punishing them. players aren't pigeons to be trained through your game-maze, they're an audience, so you want to pick the kind of punishment that conveys the right feel. roguelike style punishment works well because it means that the life of each character is a defined adventure that lends itself to being retold to other players. super meat boy style punishment works because it's brief and doesn't break the flow of play. it's all context and intent.

I'll tell you once I actually do it. I can only tell you what I want to do in the meantime.

I do want to justify player punishment as a integral design element of a game I have been thinking about making for a long time. Like start off with the standard shooter paradigm of giving a few lives at the start, and if you lose all those lives you lose your progress on that stage and your score resets. Each time you lose your three lives, you start back at the stage where you lost the last one, with three new lives, so the punishment seems trivial since you can always start back on the previous stage.

This only undercuts the true nature of the game, which involves the last series of stages being furiously difficult, combined with the aspect of purchasing or collecting extra lives throughout the areas of the game prior to facing the final series of stages. It is at this point the game will shift from the redundant lives being trivial, to the redundant lives being critical currency for advancing through these furiously difficult stages, and the punishment for losing them all is having to beat each future stage with a maximum of only 3 lives to sustain yourself.

Thus players aware of the situation that will eventually befall them will place value on not losing any of the extra lives they obtain throughout the early portions of the game, since the player who reaches the last part of the game with the largest surplus of lives has the largest advantage when it comes to overcoming the incredible challenges that lie within.

But I also want to experiment with any if not every form of player punishment imaginable. For instance, I could start this game off by introducing the players to a stage where the correct response initially is to shoot away anything that moves. Just as the player is just getting accustomed to being able to aptly shoot away anything that moves, they encounter a group of npc's that come right out in front of the player unexpectedly. I would do this under the specific expectation that the player will shoot them dead before considering that it could be a mistake, but the player will always have the choice of what he ends up actually doing, and that choice will influence how people view your character in the very first portion of the game. (I do it very deliberately in the beginning, so that the player is tipped off to his ability to manipulate the game around him, and the concept of there being consequences tied to the actions performed within that game's world.)

I'm making an open-ended RPG, free-roaming world with no saving and if they die, they have to start the game over.

Not.

They can save anytime except in any type of battle mode. Sooo, anytime they are killed they get to try again. This type of saving has been quite effective in RPG's, especially newer more expansive ones. The player is often their own biggest obstacle when they forget to save. It also gives the player more confidence in exploring new areas in a free roam world, because they can save before entering an area to find out that it is too difficult for their character.

I believe "punishing" a player for playing a game is actually the wrong way to look at it. Rather then punishing for making a mistake, I'd rather reward for doing well. In the end, players play games for rewards. Its the same reason people work, get education, commit crimes, go through labor, etc etc... it's all about the reward.

That's true. I don't mean to take all the challenge out of it; I just don't believe in taking away previous rewards for screwing up. If that's the case, they might as well go gamble.

I think to some extent you can convey through graphics that something SEEMED difficult when it's not really that bad. I remember for some reason Goemon's Great Adventure had this effect on me because of the 2 1/2d perspective they could show all this shit happening on different layers that really doesn't affect you but it makes the situation seem more hectic. Another example: games where a building is on fire or a bomb is going to go off but there isn't actually a timer until you die. But sometimes you still rush through the scene like an idiot/sense of urgency

also bad guys with guns that are like BANG BANG instead of pewpew. increases the ohshit factor. Also laser sights fuckin red dots on your chest shitshit

I think I have a point anyway. Something as detailed as Pong can't have much effect when you lose/avoid losing. But maybe add in a bunch of stupid little details like an effect on the ball to make it SEEM like it's going faster, little player graphics that you can project yourself onto

also with Dragon Quest it's been like scientifically proven that all the numbers on display subconsciously make it seem like a lot of stuff's going on, as opposed to like a meter, also probably Final Fantasy they have one EXTRA number for each guy ho shit

And bullet hell shooters. Things that move in fractal patterns are at least four times as threatening as things that don't.

And the guy's face in doom. I never have a shit about the health meter until it was nearly death mode, but the GRAAH face makes shit hurt more.

I never really thought of that either. I have used the technique of making a player believe they should hurry subconsciously by sticking them in a burning building and having a woman scream for help. That being said, taking your time in the building has no adverse affect and allows you to find an awesome secret, but players will often think they should rush. If they DO decide to take their time and find the secret/that they aren't obliterated, they feel more accomplished then had it been the obvious thing to do.

I also use the "confined space" approach, where as a player has to fight a moderately difficult enemy in an enclosed space where the enemy can more easily harm the player. I add to this effect by making the walls or other obstacles in the area dangerous. It's made more difficult because they player has to focus on more then just the enemy they are fighting. This is VERY common in games, and goes right along with Biggles comment on fractal patterns - I couldn't agree more. I usually avoid those games, the scare the shit out of me.

This again has the possibility of just irritating the player, making them want to just reload to a far off save so they can get a better response. I'm going to admit, I do that all the time as a player (mass effect 1 and 2, Deus Ex are 3 particular games that come to mind). but it is an interesting concept. The problem is that players will enter your game with certain rightfully placed video-game-logic assumptions and video game tropes and if you utilize those specifically to trick them instead of clearly letting them know that all of their actions can and probably will have so and so consequences like in real life, rather than like how it is in video game land will will just drive them away.

So, basically, I'd say its important to let the player know before any consequences from their video game trope addled mind can be caused.

That is why I would only use it as a trope in the very first instance. I want to present the concept at the player in a way that they are used to seeing. I actually count on the fact of being able to re-load a save if they feel like re-playing up to that point is worth whatever change in response they wish to see. I don't see it as a punishment as much as it is a consequence of making that decision to go back and consciously re-play that portion of the game. By the same token, I would make the ability to save very open-ended so that the player has the most control over how far they need to go back. For short sighted decisions, this will work out in their favor, but I plan to also have decisions available with longer reaching consequences that don't show up until later. The effects won't ever be game breaking, but I would make certain it was absolutely clear how what was happening now is directly related to what decision the player made back then, rather than springing it on them as an unexpected punishment or simply a wrong type of choice. (heck, just to mock players that go back way too far with their saves to re-make a decision, I could easily implement a no-win type decision that requires a couple of hours of backtracking in order to realize that both decisions are bad)

Basically, I want to respect the player enough to not have to hold their hand through every decision they make throughout. Any type game is going to involve some sequence of decisions made in the heat of the moment, most of the time with only marginal expectation at what the actual outcome of that series of decisions entail. Video games often involve tools that operate outside the standard bounds of cause and effect, but they are all still played to fulfill a desire to go through and witness how your own decisions play out in hypothetical virtual settings.

As far as effectively punishing/rewarding players go in the context of the potential game I was discussing earlier, there are two things I want to prioritize above all else:

First, giving a game setting that seems entirely open ended initially, but actually has a short and relatively obvious path that steadily ramps in difficulty as it delivers the player to the option of entering the endgame, while not forcing them through a point of no return without clearly stating it as such. (this way, it would be possible to get all the way to the endgame with only the 3 free lives without even purchasing a single bonus life, then backtracking and getting a bunch of bonus lives all at the same time just for the endgame) Maybe even with options to advance through those stages even more quickly and having consequences for doing things that way instead. (Any areas outside of that path will be entirely optional, have clearly defined rewards, and ramp in difficulty based only on the player's current progress.) Or on the flip side, I can also provide a victory condition that can be achieved without playing the endgame levels, making them practically an optional affair just for players seeking that level of challenge from their game.

While the other being giving as much motivation as possible to play the game all over again with the possibility of exploring divergent paths where new decisions can lead to unseen options occurring. (which save-games could be used to exploit to an extent, but my game would be heavily class-based, so class based options would require starting the game entirely over as that class in order to actually observe and explore)