So far, there seem to be confusion about what a "TAS-proof" game means. It could mean:
1) A game which is pointless to TAS.
2) A game which is difficult to reach perfection.
3) A game where perfection is so close to normal play (might be #1).
4) A game which aims to depress TASers (might be #2).
5) A game which aims to discourage anyone from even playing it normally.
JXQ wrote:
A chess game with a takeback feature seems "TAS-proof" to me.
Or reaching the end of level 1 in less than X seconds makes level 2 proportionately harder.
That wouldn't make TASing impossible. It would simply cause the TASer to need to find the optimal strategy, the optimal compromise between speed and difficulty which minimizes the overall time.
Tub wrote:
on the topic of a fixed seed for the PRNG that is saved: having predictable and unchangeable randomness once the game is started will be abused by unassisted speedrunners or even regular gamers. Memorisation replaces reflex. Don't do that.
I think that the idea was that the game creates a random seed (from user input delays or whatever) and then initializes a RNG with that seed, after which the RNG is not affected by anything anymore. Changing the initial seed will create a completely new and unpredictable (well, for a human at least) random number sequence, and since it can't be affected anymore by anything, TASing won't help.
on the other hand, there are games with a fixed PRNG where randomness can still be manipulated: by polling more or less random values from the PRNG. Like creating dust-sprites in Mega Man X.
The idea was that there would be several such RNG streams. For example one stream would be used exclusively for randomizing the time between encounters. Since there's no way to pull numbers from this RNG stream (because separate independent streams are used for other things), it cannot be affected.
In other words, it's like at the beginning of the game each single random encounter would have been fixed, and there's nothing that can be done to affect that after this initial randomization.
This should be perfectly possible to make even in a NES. There exist simple formulas (of the form "seed = seed*constant1+constant2") which, using eg. 32-bit numbers, go through the entire 32 bit value space in pseudorandom order.
I think that the idea was that the game creates a random seed (from user input delays or whatever) and then initializes a RNG with that seed, after which the RNG is not affected by anything anymore.
right, but he also said that the current PRNG-state would be written to savegames. So it can be abused in unassisted speedruns/regular gaming for every point after the first save.
It would hinder luck manipulation in a TAS, but at what costs?
The idea was that there would be several such RNG streams.
keeping the PRNGs seperate would prevent the MMX-like problem, but could still be abused depending on the game and situation.
Should I slash this enemy to get more damage against the bigger enemy coming up next?
Should I get another encounter before passing this border, so I'll have less encounters in the more difficult area?
If you want to prevent luck manipulation, you'll have to make a game where luck isn't used. Enemys have to be blind as well, because otherwise you could manipulate the enemies directly instead of manipulating the PRNG.
You'll get something like DDR or SMB, which are nice and easy examples of TASing.
If you just want to make luck manipulation hard.. well, Minish Cap is being TASed as well. So is Breath of Fire. And I believe we have a couple of SNES Mega Man runs somewhere. The game can still be an entertaining TAS, even if not mathematically perfect. (but we should probably agree on a question, see FractalFusion's post)
SXL wrote:
sorry, had to reply that Tub's post looks a bit sadistic. you're a funny, yet a bit scary guy :°)
Tub: Random.................not really.. I still don't see how a game can not be not random, there just is no way to do it, you can make all the pseudo-random number generators you want. The code needs to work from something (electricity cycles all the way to your input and beyond), nothing can just be derived from 'nothing' as it would destroy the world!! Unless school has failed me, wouldn't be surprised. :(
Joined: 11/27/2004
Posts: 688
Location: WA State, USA
L4yer wrote:
Tub: Random.................not really.. I still don't see how a game can not be not random, [...]
The console could have something radioactive and a gieger counter (or something to that effect). Then again, the emulator would probably just use an RNG based on the player's input.
Nach wrote:
I also used to wake up every morning, open my curtains, and see the twin towers. And then one day, wasn't able to anymore, I'll never forget that.
Joined: 5/23/2006
Posts: 361
Location: Washington, United States
What about an RPG with the following applications of the RNG:
-Random encounters will always occur in a set order (like, 5 steps, 7 steps, 3 steps, 4 steps, 9 steps, 6 steps, etc., then repeat the pattern)
-Battles are determined at the start of an encounter, but the attack patterns (i.e. miss, hit, critical) will follow one of, say, 20 patterns with a set number of misses, hits, and criticals. Another possibility, which would make the game a bit frustrating, is to make the enemy hit whenever you hit, miss whenever you miss, etc.
But, most of this just makes the game pointless to TAS (i.e. a speedrun would look about the same as the TAS), not truly TAS-proof.
Hmm, interesting that Tub said "he" to refer to me yet later mentioned Breath of Fire. (Myria is the evil goddess of Breath of Fire 1 and 3.)
--
I think it's basically impossible to have a random generator that can't be manipulated. Final Fantasy 5 GBA is a good demonstration of this.
Let's say that you're TASing FF5 GBA. You're in one area, fairly close to another area. You've just loaded a save file or quicksave (FF5 GBA bug: the encounter stream cipher gets set to a constant). You know that the bug in the game will cause you to fight the rarest monster in an area in your second battle after loading the file.
With the current encounter countdown, you can get to the next area with one encounter in the current area along the way. However, the rare monster in the next area is very nasty and you don't want to fight it with your underleveled characters. The answer is obvious: fight a second fight in the current room to get an easier rare monster before proceeding to the next room.
Even if you can't cause generated numbers to be consumed elsewhere (Rockman X as described above), you can change the context in which the numbers are interpreted.
A potential counter to this is to give a separate stream cipher to every room in the game, but that's a lot of extra data for the save file.
--
What is the significance of a take-back feature in a chess game?
Melissa
What is the significance of a take-back feature in a chess game?
There's a glitch with the take-back feature that can give a player an extra move. That's it, really.
It's not the glitch that causes the game to be "TAS-proof"; rather it is the game itself. The glitch has nothing to do with a TAS.
I just meant that having a takeback feature negated the need for savestating. And a chess game isn't about real time reflexes, so TASing it seems redundant to regular play of the game.
<Swordless> Go hug a tree, you vegetarian (I bet you really are one)
That wouldn't make TASing impossible. It would simply cause the TASer to need to find the optimal strategy, the optimal compromise between speed and difficulty which minimizes the overall time.
Definitely. I just interpreted "TAS proof" to mean something that closes the gap between TAS and speedrun, while maintaining the characteristics of a game.
Hardcode the ONLY frame-specific solution to a battle into the game. This game will never be beaten in real life, and it would (probably) never be beaten by TASers. Nonetheless, there exists a unique, albeit difficult to find, solution to beating the game.
Edit: Hey, with this said... I was reminded of something from a much older thread. I just recently started reading about quantum computers. What types of advantages would they provide in brute forcing a game's optimal solutions? Would those advantages allow for the game to be solved in a reasonable time (a human lifetime or somesuch)? Numbers people.
The best idea I can think of is to write your games for a console that is either really confusing to emulate, like the DS, or contains hardware that most people's computers won't have the power to emulate, like any of the next-gen systems.
I just recently started reading about quantum computers. What types of advantages would they provide in brute forcing a game's optimal solutions? Would those advantages allow for the game to be solved in a reasonable time (a human lifetime or somesuch)? Numbers people.
Yes, this seems reasonable to me. I heard of computation times being the square root of their classical counterparts. So the numbers I was thinking of, as you pointed out, would still be too huge. OmnipotentEntity likes getting my hopes up, though.
http://tasvideos.org/forum/viewtopic.php?t=2471&postdays=0&postorder=asc&start=60
Definitely. I just interpreted "TAS proof" to mean something that closes the gap between TAS and speedrun, while maintaining the characteristics of a game.
I'm not sure if it would close that gap at all. The same "you are playing too well, let's make the game more difficult" rule applies to both types of runners. The advantage of the TASer is that he can search for the optimal compromise while the regular speedrunner needs to go by the feeling.
Yes, this seems reasonable to me. I heard of computation times being the square root of their classical counterparts. So the numbers I was thinking of, as you pointed out, would still be too huge.
Actually, they're only that huge if you bruteforce the movie as a whole. However, if you split it in short sections (say, separate stages or arbitrary frame ranges with partly predefined ending frame goals), it would be nothing for such a computer. Actually, it is already feasible now, but still requires a lot of time and manual work setting up the goals and general strategies. I'd really wish for such a run to be created, though, and LoZ, as suggested by xebra, seems like a good candidate.
EDIT: On the other hand, that would reduce the possible timespan of a given section to a few seconds at most… Well, there's still something to hope for. :D
Warp wrote:
Edit: I think I understand now: It's my avatar, isn't it? It makes me look angry.
Make it so that certain activities that can only be done in TASes increment a TAS counter in memory. These include frame-perfect button-mashing or abusing certain glitches that were known to the programmers but never fixed because they are so hard to reproduce.
Much later in the game, add lots of lag, which is proportional to what the value of the TAS counter was a few levels back, so that each TAS-specific exploit loses seconds or even minutes over the course of the movie. Turning the console off for 5 minutes resets this, so it's usually not a problem in regular play.
Since the TASer doesn't know about this in advance, he is stuck dealing with mysterious lag starting on level 3 and getting steadily worse over the course of the movie. Furthermore, he can't figure out how to avoid it, even though it wasn't there in his test run. If the first few levels are optimized enough, a "perfect" TAS will be easy to beat in real time.
The best part is that the fastest TAS of this normally entertaining game looks a bit sloppy, doesn't take down bosses as fast as possible, and avoids all the cool glitches and sequence breaks.
Trying 127.0.0.1...
telnet: connect to address 127.0.0.1: Connection refused
telnet: Unable to connect to remote host
I'd create lots and lots of options, and radomness for determining large chunks of the game. The random variables would be influenced by the options chosen. The game could last 8 hours or 1 hour, depending on the randomness. (hopefully I described that right).
In other words, it would be so long and difficult to maipulate the randomness that no one would bother to do it.
There are lots of interesting ideas in this thread, but I think you're all trying too hard to work around an issue that doesn't exist anymore. If an emulator for any recent console was relatively accurate, it would include some version of a time chip independent of whether a game had been "inserted," right? At least on the PC, randomness is nearly always seeded from time, and I can't imagine it's different on the PS2 (though I can't imagine a lot of things that are nonetheless true).
So barring "cheating" by including the exact time and date in the movie file, or forcing the user to otherwise set the clock back before playing it (neither of which I consider to be in the spirit of a TAS at all), wouldn't nearly any PS2 game be completely and literally impossible to properly TAS because, no matter how unoptimized you were willing to be, it would still desynch?
hum well on the emulator there will be some config file, which you would delete before starting the game, so that you always start the game at the same time. And I think that Frame advance would block the clock so that it actually advance at the same time as the clock.
