A few years ago there was a mario competition to generate levels. When the contest was over I downloaded the entries and tried out some of them. Quickly I found that while they had successfully generated a lot of levels the levels were not very fun.
A few years ago I started playing a lot of mario style platformers writing down notes including what I found "fun" (for the day I would get around to making my own platformer).
Skip down to the "What is fun" section and you will find the first two items pretty much say when you die it is not fun. Skill level doesn't necessarily equal fun in a platformer game. That block that almost kills you and scared you? Turnes out it was scripted and there was no way you could have died no matter what your skill level was.
There are a handful of platformers on Linux and they all have the problem of being too hard. The games in the first few levels are so hard that the player quits before they can explore what the engine can even do or find the "fun" in the game.
Maybe difficulty is simply the easiest nob to twist when designing platforms which is why so many first time platform developer turn that first, but I would have to think that the fun found in a platformer is not from the difficulty, but in other aspects. No one would dare say that Mario 2 (japan version) is better than Mario3 because it was more difficult.
I love hard games, but there's a very fine line between hard and frustrating and pulling it off is difficult. For one, if I'm playing well and don't make many mistakes (how many depends on how far in I am - if I'm almost at the end, zero mistakes may be tolerable or even a good thing), then I expect to advance. For example, if I play really well and then lose because of an unforeseeable event, that's not "hard" - that's just frustrating. It's also closely tied in with fairness. If a game is only hard because the AI cheats (perfect aim, always knows where you are, always outnumbers you, infinite resources, whatever) then that is probably not going to be fun. It can be, depending on the game (eg infinite amounts of enemies are a fun challenge in one game but frustratingly unfair in another), but again, this is difficult to pull off. Repetitive stuff is rarely fun, no matter how hard or easy it is.
What makes hard games fun is the sense of accomplishment when you finally outwit or outplay the game. A constant sense of impending failure is a good thing. Failure is also not a bad thing, as long as you can learn from it and do ever so slightly better next time. Constant failure is not a good thing as it quickly leads to frustration. Anything out of your control is also a bad thing and leads to frustration (example: if I died because I made a mistake, that's fine. If I died because I didn't do exactly what the designer wanted and there's no other way to do it, that's probably not fine, unless it was obvious what I should do). Some games also expect you to die a lot, but dying isn't particularly painful, so its not all that frustrating (ie in VVVVV).
Finally, gamers often (maybe even "usually") make terrible designers, so unless you have proven yourself, designing a fiendishly hard game is going to be a pretty big gamble.
Our goal isn't actually to make a hard game. Our goal is to make a game that suits your skill level, whatever it is.
As for whether we've succeeded, we've tested the game on a few hundred people, both hardcore and casual, and have only had one dissatisfied tester (it was a 5 year old who was playing an early build that was way too hard. Honestly I'm impressed he played it for as long as he did). In particular the good folks at Microsoft and Valve seemed to have a great time playing today. (Hooray!)
I have to disagree. I think making a hard game that is still fun is very challenging. But when you accomplish it, it's an even better platformer than the not-so-hard one. Take Donkey Kong Country Returns (Wii) and Rayman Origins (Wii/360/PS3). Both are very hard, especially DKCR. But they are very hard in fair, legitimate and satisfying ways. Beating the later levels in DKCR takes a bit of dedication, but is oh so satisfying when you get in the zone and find that beautiful path through the level.
Also see Super Meat Boy. It's basically a hard as nails platformer where you can and will die hundreds if not thousands of times trying to beat it. It's another great example of very hard (it's way harder than both RO and DKCR), but still fair, fluid, and "zen-like".
I don't think any automatically generated levels could ever reach the level of flow that the games I mentioned have, though.
Hard and fun are completely different things, I agree. I spent about 3 months on the basic algorithm, making sure the levels generated were feasible and interesting. The remaining 3 years have been spent making the game fun. The real trick is tailoring the difficulty to suit the level of the player. When you nail this correctly it leads to a very satisfying experience (and the algorithm does have a sense of 'flow' when it designs level, although I can understand that's hard to believe without playing the game)
Tell that to Dark Souls. I recently replayed ST:FU on it's harder setting. I think perhaps the difference between "dying" and "not fun" is a couple of things: first, there are regular and predictable save points so the level doesn't have to be started from the beginning; second, there is still some progress saved, experience earned, etc., so that even in death there is some progress; third, a way to track community progress (via achievements or social features) so that it's clear the end is attainable by something other than luck.
> second, there is still some progress saved, experience earned, etc., so that even in death there is some progress
I accidentally realized how important this is while playing Bioshock 2. The fact that when you're trying to kill a BigDaddy and you die, they keep their damage when you restart so that you can still kill him even if you don't have any weapons left and have to repeat 3 or 4 times.
If that wasn't the case it would have probably demoralized me from continuing the game when things got tough.
Therein lies the pendulum swinging the other way. When dying doesn't really disadvantage you, it takes a lot of enjoyment out of the game. Particularly in Bioshock where it can be an advantage to die (free health!)
Some people honestly do love nigh-impossible platformers. I am not among them, but the Linux games are going to be made by people who 1) are making it for themselves (and possibly a few others; that community's relatively small) and 2) aren't looking for massive distribution.
People who love platformers that much usually like them to be hard.
Very few platformers can pull off "hard" well because it can easily become re-playing the same 1/2 of a level for extended periods of time. This will result in users putting the game down, not telling their friends about it and game making less money. As for the open source examples my hunch is that they just wanted to make a mario clone and having play tested the game themselves throughout making the game didn't realize how hard the first levels were and never sat down and tried to built up any progression of difficulty that might match the players skills.
I love insane platformers like N+, Super Beat Boy, the upcoming cloudberry (from the OP), and of course the super hard mario levels.
I think what's interesting about cloudberry is that they are dynamically generating the levels, so you can play the game easy or hard. Properly implemented, it could be attractive to both casual and hardcore platform gamers.
Watching the (30 second?) trailer I saw a lot of the same stuff repeated which typically doesn't equate to much fun. With the hype of more levels than you could ever play seeing the same thing over and over in a trailer made me cringe. With dynamic level generation it is a bit odd that they would even have difficulty settings and instead adjust the difficulty on the fly like Crash Bandicoot's engine did to make the game more fun for the user.
> There are a handful of platformers on Linux and they all have the problem of being too hard. The games in the first few levels are so hard that the player quits before they can explore what the engine can even do or find the "fun" in the game.
When you're developing a game by yourself, you play it 1000s times to test every small change, debug, etc. So you're becoming pretty good at it.
So you balance it to be challanging, and it's too hard for regular players. At least that's what happened to me.
As a counterpoint: Spelunky is great fun, extremely hard, and very addicting. It's levels are randomly generated, and if you die you start over from the very beginning; it's perma-death.
A few years ago I started playing a lot of mario style platformers writing down notes including what I found "fun" (for the day I would get around to making my own platformer).
http://benjamin-meyer.blogspot.com/2010/04/what-makes-fun-ma...
Skip down to the "What is fun" section and you will find the first two items pretty much say when you die it is not fun. Skill level doesn't necessarily equal fun in a platformer game. That block that almost kills you and scared you? Turnes out it was scripted and there was no way you could have died no matter what your skill level was.
There are a handful of platformers on Linux and they all have the problem of being too hard. The games in the first few levels are so hard that the player quits before they can explore what the engine can even do or find the "fun" in the game.
Maybe difficulty is simply the easiest nob to twist when designing platforms which is why so many first time platform developer turn that first, but I would have to think that the fun found in a platformer is not from the difficulty, but in other aspects. No one would dare say that Mario 2 (japan version) is better than Mario3 because it was more difficult.