Randomness in Games: A Long-form Analysis
Vložit
- čas přidán 17. 03. 2015
- You can help support Game Soup:
Patreon - / gamesoup
Paypal - GameSoupLP@gmail.com
Bitcoin - 1FMPuCoEeXrPxdMnjx1b86nSDEmKem4rzt
We have an analysis-only channel (no Let's Plays): / @gamesoupanalysis
Thanks for watching! ツ
GameSoupLP at gmail.com
What is the role of randomness in the average video game?
In this analysis we'll take a look briefly at the philosophy of randomness, then spend most of our time looking at the ways that games choose to implement randomness, or RNG. We'll check out methods of implementation, calculation and revision, and the benefits that random elements can bring to the table to augment and improve existing game mechanics. - Zábava
Actually the deterministic 'random' battles in FF1 and 2 increased my immersion I felt like those monsters were out there, rather than just randomly appearing on top of me.
Any one playing X-com will understand how randomness can seem be unfair. 95% chance to hit an enemy misses, sometimes even 2 or 3 times in a row. the 5% enemy chance to hit your operative crits and kills your best guy.
What a lot of players don't know is that the number is actually higher than indicated, as stated by the lead designer Jake Solomon in an interview on gamasutra (and other places). Meaning the numbers are skewed to always be more in favor of the player.
But the 'problem' is that if you miss a 95% chance to hit shot at point blank it feels bullshit, especially since (unlike white a physical dice) we can't see the random roll. In addition, when we see such a high percentage as say 95% our mind doesn't really see "so it's a 5% chance to miss" but instead pretty much interprets it as "a guaranteed hit". These things (and more) combine to make us experience it as "oh cmon, I should've hit that!" instead of "Well... I did have a 5% chance to miss, oh well".
note at around 27 minutes you make the remark "this creates a negative feedback loop" referring to how the result of unskilled players losing will quit, making a larger percent of all players skilled, making it harder for new players, constant spiral until only very skilled players want to play. This is actually a "positive feedback loop" as the feedback into the system pushes it further in the direction it was going (into an unstable or undesired state), a negative feedback loop would be self stabilizing. Control systems theory, your concept is completely correct. it's just the wrong word. In terms of feedback loops "negative" and "positive" don't mean good or bad. positive means in the same direction, negative means in the opposite.
+Diamonddrake Oh, wow! This is a really good observation - I certainly got the terms mixed up here.I didn't realize there was such nuance to these concepts. Thank you for pointing this out!
A very common and understandable misunderstanding. I actually learned aboot that from Game Maker's Toolkit before I read any books on cybernetics
Also, your random number was likely 7.
QuickNET Tech LOL so true
it was actually 5 but suddenly i went like "that would be too obvious, i'll pick 4"
QuickNET Tech
Fuck, how did you know?
O.o
I didn't know about the seven set rule for Tetris RNG. That's really helpful.
love your long form analysis videos, keep killing it my guy
Very well researched!
Subscribed for sure!
+Willy Goat Thanks! :)
This channel sure is exceptional. Keep up the good work!
Wow. Very nice analysis, and honestly, a solid thesis. Loved the wrap-up about randomness bringing color. I know you focus on video games, and it would be cool to hear if/how you think these principles apply to tabletop gaming (RPGs, board games, etc). that use different methods of allotting or assigning RNG via dice of various sizes and numbers, cards in a deck, and the like.
In practice, I think regular players won't distinguish between true randomness and pseudorandomness if you seed the pseudorandom with something like a millisecond time measurement.
What is much more useful is things like using a card stack for Tetris pieces vs. having independent odds each time it picks a piece.
Or the random number of steps between 12 and 32 vs 6% odds for each step. Very different odds that will give very different experience for the player.
+r hill But using a card-stack thing will prevent you from, for example, getting the same thing twice in a row, or almost twice in a row but with another thing in the middle. The only way this could happen is if you're about to run out of cards and are close to a re-shuffle. However, the ideal thing would be to make all points equal: every result (not only the border ones) should have the same chances of repeating twice, unless that result has happened in the close past (2-3 before).
imo
Nicely done video.
Really like the thorough analysis that covers most aspects of RNG.
One aspect you didn't touch though is the Desire Sensor mentality which is prevalent in games such as Monster Hunter and Diablo where a player feels like he doesn't get the drops he wants just because he wants the drops.
Man. This is amazing content. Keep it up!
Always enjoy your Longform Analysis videos, the internet needs less juvenile attempts at lets plays, and more of this!
The internet's pretty big. There's room for both!
Great video, thanks for sharing!
Well that went deeper than I expected... :D
Great video!
In Final Fantasy 4: The After Years, the original game determined the composition of encounters based on the position of party members and the actions taken and enemies encountered in the previous battles. People were able to use this to encounter very rare bosses and get rare items very easily.
In the PSP version, they changed it. Encounters are determined based on the PSP's clock. This is much harder to manipulate.
Wow, again, great video
At an arcade I saw some people complaining to the manager that the claw machine wouldn't grab the prize enough to lift it to the exit. The manager then said that it only grasps it hard enough every time another $50 has been inserted. That means that is wasn't random at all.
There seems to be some misunderstandings in regards to the random encounter example. Firstly, player input is rarely taken into account in PRNG. If you wanted a cheaply "secure" PRNG you could fold in elements of the input provided you can measure them sufficiently accurately enough (ex. how long did a player hold down the last button down to the number of clock cycles or even just microseconds), but I don't think anyone actually does anything like that. PRNG generally just takes the current random number and feeds it into a mathematical function. One of the egregiously broken examples cited, Golden Sun, does exactly this. What breaks Golden Sun is not that the player's input is taken into account, it's that the RNG is reset on reset, the first few numbers of the RNG happen to be particularly beneficial (because of particularly poor seeding -- fyi "0" is often a dangerous choice to seed an RNG with), and RNG is not constantly consumed ("per frame" RNG). Secondly, the function given at 17:40 is functionally identical to the FF1 example if you don't reset. You could also make it identical to the FF1 example when you do reset fairly easily, though at that point you might as well just do it how FF1 did it instead. You could also mathematically generate a number with a non-uniform distribution to make one identical to the FF6 one, but again, at that point you might as well just do it how FF6 did it.
The FitnessGram Pacer Test
is a multistage aerobic capacity test that progressively gets more
difficult as it continues. The 20 meter pacer test will begin in 30
seconds. Line up at the start. The running speed starts slowly, but gets
faster each minute after you hear this signal. A single lap
should be completed each time you hear this sound. Remember to
run in a straight line, and run as long as possible. The second time you
fail to complete a lap before the sound, your test is over. The test
will begin on the word start. On your mark, get ready, start.
Amazing video, keep it up.
This is why I don't like it when non-statisticians talk about randomness. You guys always mix up randomness with independence, and uniformity. They are not the same thing by any stretch of the imagination. The main limitation saying that randomness does not truly exist is chaos, but I'm no philosopher, just a humble statistician.
It is not anymore random to have uni_c[0,5] as opposed to exp[3]. It is also not any more random to pick a permutation of 5 cards from uni_d[1, 5!] as opposed to picking 5 random cards. The outputs are almost certainly different but they are equally random.
+The Count of Menea Calais Hi, and thanks for the great explanation here. I'm definitely not a math-oriented person, so whether the topic is randomness, independence, or uniformity, I'm basically trying to relay my thoughts with a "layman's" use of the word random. I feel that everyone has a basic understanding of randomness as a concept (even if I can't explain it in mathematical detail) and as the topic is really interesting to me, I wanted to share it as best I could. Sorry if that turned you off from the video, but I really appreciate the feedback regardless :-D
Randomness is very delicate to describe because what people think as as randomness is as above except in two discrete cases: probabilities of 0% and 100%. However, randomness can still be expressed on a spectrum. Hence, it's not simply reliability or predictability or independence. It has to be very precisely defined (something I don't trust myself to do off the top of my head) else very fundamental philosophical problems arise. the most notable one is: what is the probability of something that has already occurred if you don't know what the outcome is?
Obviously the probability of an outcome that has already occurred is 1. Precisely defined indeed.
So what you're saying is that the probability that the other guy flipped a heads is 1.0 and the probability that the guy flipped a tails on the same flip is also 1.0 because the event is in the past? Remember, you don't know what the outcome is. That breaks one of the 3 rules of probability: the sample space must add up to 1.0
The two most popular answers are something akin to Schrodinger's Cat and the Gaussian(?) Method. In the first, from your perspective (this is key), you don't know what the outcome is, so you can treat the probability as either 1 or 0 simultaneously until you figure it out. In the second, the outcome is still 0.5 except we stop calling it 'probability' and start calling it 'confidence,' which is also pretty fair since that's how we approach parameter estimation anyway.
Is that to say, then that probability is relative? That my probability is different than your probability if we have asymmetric information? How is this not conditional probability? And why is it worth redefining probability to do this?
Neat video. I personally didn't see the point of discussing pseudo-randomness (not to be confused with constrained randomness) vs. true randomness for such long portion of the video. Even if true randomness could efficiently be implemented into a video game, you couldn't tell the difference. On the contrary, I missed a more in-depth discussion on constrained randomness (e.g. replacing a dice-roll with a card-draw figuratively speaking).
Also your notion about "solved" games misses the point of them completely. Chess, although technically solved, remains a highly regarded psychological strategy game: you play your opponent, not the game. I suspect this is the case with many other "solved" games as well, if you're sufficiently skilled at them. But of course, "solved" and simple games like Connect Four exists, and suck.
Putting limitations on a random playlist doesn't stop it from being random. It's just random drawn from a different distribution, or the draws are not independent and un-correlated.
Gambler's Fallacy is more-or-less the bread and butter of mechanics-based Pokémon internet rumors. How many times have we heard online that pressing the A button at a specific time during the capturing animation supposedly gave you a MASSIVE success boost.
I kinda wish you had discussed the desire sensor in monster hunter, it's a very interesting mechanic for combating RNG.
In Pokémon TCG for the Gameboy Color, the coinflips are predetermined. If you use an attack, but the coinflips are unfavourable, you can restart the game and use a different attack, an item, or switch Pokémon.
I agree that true randomness is hard to achieve, but here are few tips to achieve closest possible result:
- Uptime. Of course it's staple of getting random seed. But it's also increasing at a constant speed.
- Microphone noise. Most people, unless they're into streaming/youtubing/audio producing have pretty much crappy microphones. Which is to our advantage as they're picking up much random noise as well which can be used as random number source to the pretty good effect.
-Mouse distance since last frame. Calculating distance mouse traveled since last frame may as well give out random results since you'll neber be able to move mouse with constant distance since last frame, even with a steady hand.
- Same for keyboatd. Scancodes of few recently pressed keys are good sources of randomness.
- Window handless and pids of currently running software. Because of how they're given out they're unpredictable for a person without knowledge on how his/her OS works.
None of above will get you anywhere close to the true randomness if used alone, since either can be affected (with os handles being hardest to afgect for average gamer), but combination of those may get you unpredictable numbers that are hard to affect. Cheers.
Plus none of these require any exotic hardware on part of the user. Just stuff he likely already has.
One example of a really bad pseudo-RNG is the one from Monster Hunter 3 ultimate. The seed is selected when you start a game based on the current time and is stored in your save file.
The bad part is that if you are unlucky, the seed will result in a RNG with a really short period, like a few hundred. In practice it means that there may be some random items (charms) that you will never get, no matter how much you play, unless you start a new game. These are called "charm tables", but for me, they are more like a bug.
The RNG can also be manipulated but it is not an exact science. This manipulation is also made possible by the short periods. Note that even "lucky" seed only have periods in the few thousands.
thanks for the info! -Joe
Great analysis! Makes me think a little deeper about what I'm inflicting upon my players.
Your bit about Mario Kart makes me feel like you've never played Mario Kart 8, though. The item grants' RNG system is so completely borked and unbalanced, it really does feel like it's ACTIVELY working against you 99 percent of the time. There's no way to easily predict what you'll get, which is fine, but if you predict "I'll get something useless" you're probably right regardless of your place.
mario kart items are not random they have rates depending on your placement at the time witch i agree is unfair and you do often get crap while the comp gets red shells and stars mario kart would actually be more fun if the items where more random and blue shells where in special boxes only like mario 64 that or lower the odds to get them to 5% because they are op 64 was alot more balanced which made it more fun it felt more chaotic however the karts and charaters have become too balanced they don't even feel different anymore it feels like character section is purely cosmetic in mario kart 7 and 8 they don't have any personality
good video
I get what you're saying about Tetris randomness, but...
"It's entirely feasible that a player could get 60 line blocks [in a row]"
(1/7)^(60) = 2×10^-51. Not feasible, likely, or even remotely probable.
As a side note on randomness. Subatomic particles were tested (in a variety of ways) and all tests so far produced results that suggest that with modern theories some behavior is inherently random, so there might be some true randomness in the universe, although "true" randomness isn't particularly useful for gaming puposes
i use something i call the clock randomness, i give the player a number and keep on increasing it but for example everytime its 64 i make it 1 again and becouse the numbers increade fast enough i dont think any player would be able to catch the number they want
The game I'm making needs to have "randomness" in it, the only thing I have so far is a set critical hit chance. However, I'm going to have some points where it SEEMS random but there is a set outcome, like a "luck" stat, (or like how light/luck works in Homestuck)
Great video and all, was the footage captured on a potato though? Seeing Minecraft and D3 lag like that, reminded me of that old shitty Laptop I had with Intel HD integrated graphics. Those were dark days
I mostly agree with most of your video, except when you get to Hearthstone. And don't get me wrong, I love HS to the core. However, the huge power of Random elements of the game is way over the top, particularly in professional play.
Tournaments matches among the BEST players in the world have been 100% DECIDED by Shredders pulling MillhouseManastorm (a win) or Doomsayer (a lost). Viable-yet-random like Ragnaros, Unstable Portal or even KnifeJuggler made or break whole brackets in a single coin toss.
You linked to ExtraCredits, which is weird because they took a lot of time to explain this. Randomness in Poker works because there are MANY hands, so getting a few lucky hands is no big deal. However, getting EdwinVancleff from UnstablePortal after using TheCoin pretty much gives you the entire match on turn 1.
Random cards that may give you small edges and can be managed to give you the best outcome, like BombLobber or StampeddingKuddos, are incredibly reasonable. Random cards that require a lot of setup to be played (like Yogg-saron) are incredibly fun and completely harmless. Random-yet-viable cards with the potential to win a game on the spot are neither of those things.
P.S. I Find it weird you didn't mention Dota2. In Dota2 they actually changed the random distribution in some skills to make their outcome feel more fair. The newer semi-random distribution makes the chances of getting a positive result higher every time you get a bad result. That way, players have more tools to adapt to randomness.
I like these type of videos do you program ?
+Justin Wilson We both do a little bit of programming in our spare time using a program called Construct 2. Neither of us does programming professionally though.
I dunno about the cluster illusion. When 90% of my legendary shield drops result in the same crusader shield (salvation) in diablo 3, and at least 5 in a row, i am very skeptical of the randomness at play.
These analysis videos are great! Consider me subscribed.
thanks!
You did explore some aspects in detail but this analysis lacked completeness or purpose for me. It was great to hear such details of specific games and perhaps with a video title closer to that topic and a point focused on that, it would have more impact. Thanks for the time, effort and skill put in.
I think that positive reinforcement and learning through a variable interval ratio could be a big driver behind how games become addictive.
Funny you should mention that. Our next new analysis video focuses on exactly this subject... --Ed
Definitely, there are personality traits that influence the willingness to learn through a variable interval ratio of stimulus. But I am a huge Starcraft fan and I think that games such as League of Legends are really impacted by this clicking experience, you click multiple times and then in interval ratio you win, effectively winning this strategy. I see you have a psychological background, me too, it's great to see games explained. It is definitely understudied.
3:21... have to stop you there, coins and dices are random unless they were tampered with. Also, while ANY event has to have a causal trigger for it to come into existance, the resulting act CAN be perfectly represented by a given random distribution, from the way dropplets fall the sky when it rains, to the half life of an isotope. Yes, human are VERY BAD at understanding random that's completly true.
What humans are VERY GOOD at is figuring out patterns... so maybe the Tetris example is completly wrong on the part of the designer... if the game were truly random, an experienced player might lose against a novice if they had a really bad batch of figures... but the experienced player is STILL going to win more matches against the novice... so yeah, that's why I say humans are bad at understanding random.
8:29... wrong... ish... there are systems that are chaotic, that means that even if you know your starting setup, a really small variation from that can lead to a completly different result. Add to that the fact that physics is at it's core and observational science (well... most sciences are), and you land into problems you can't solve exactly.
ooooh, what is that game with the spaceship shown at the end again?
FTL
Faster than light
It's so hard
to say goodbye
to yesterday
Sub'd. Note that dice rolls effectively are random from a human point of view, because as you explained most people aren't trained to roll specific numbers. Although yes they are deterministic, so not random at the same time.
+DJ Owlz Thanks for subscribing! Some day I'll learn to roll a 20 and then all my D&D adventures will be a cakewalk.
Gambler's Fallacy: Why I never use certain Fire Emblem characters others swear by.
Thanks
Just an aside: the example you provide with the gambler's fallacy is not entirely accurate, at least where I live. In my country the law dictates that slot-machines must have a certain pay-out-ratio. I'm not sure, but I think it's about 5%, i.e. for every 1000 bucks put into the machine, it has to pay-out at least 50. So the observation: "the machine hasn't paid out in a while, I'm bound to get lucky the next try" is not wholly untrue. I've tried this many times myself: slot-machines would usually be more generous on the afternoon after a busy night in the pub, because they were more filled and therefore more likely to pay out. I guess some games might also apply this tactic: "Event x is going to happen at a random interval, unless y or z is satisfied, then it will trigger right away."
this must've been asked a lot, but what is the name of the intro song?
+Max Wuup I think you're the first person to ask, actually! It's an original song that Ed (the narrator) wrote many years ago. The track can be found here: soundcloud.com/ecvgm/on-the-arcadian-expanse
The Tetris part was interesting but partly incorrect. Not all versions of Tetris use the deck method of randomization and NES Tetris (the one that just had a huge tournament in Portland, the CTWC) is a lot closer to randomness. Though the skill of the players is still obviously important because 1 person has won 7 out of the 9 tournaments that have ever been held.
Though they do use a seed to get the same piece sets during tournament games. So I guess your luck point does still stand lol
If you're interested in a technical view of some pRNGs used in production code: citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.53.3686&rep=rep1&type=pdf
+CharacterString This is fascinating stuff! I wish that it wasn't mostly over my head in complexity >.< Thanks for sharing this!
RNG Manipulation is why i would always have the RNG be based on the System Time.
or if possible use some strange vairable aviable that is hard to control. like on PC you could use the CPU Temperature as part of a Seed for an RNG Function
whats the game at 0:30
I would watch a Hearthstone long form.
I actually began writing one long ago, but so many people have already analyzed the game to a great degree. That's not to say it won't ever happen, but it might be hard to say something that hasn't already been said (or said better) by others.
Any recommendations for those videos?
WOuld you talk about League of Legends's crit chance RNG?
You are mixing up two very different things. Having true randomness does not prevent a developer to determine or manipulate the brick order in tetris. You always take the random number and make something with it. In this case you'd take the random number and say if it's over X give brick Y. Then you add that if you already have brick Y four times it can't add a new one to the order. So using a true random number generator in tetris does not mean that the bricks will be in a total random order, it just means that instead of a dice-roll a real random number is used to determine the order. And while you could only notice the difference statistically there are many cases like betting where true randomness would be quite nice.
The reason humans can't choose randomly is the same reason I have trouble believing in the concept of free will. Seems more causal than something magic.
This was great, I am very skilled at creating controlled randomness but knowing how to use it properly is an entirely different skill. I base a lot of what I do on D&D so I have a lot of rolls that use if(random(0,100)
+obscurelobster It's really fascinating to dig deep into randomness because there are countless ways to approach and implement it. I'm always amazed at how sophisticated the RNG systems in games can be.
From all I've read, it seems the case that players like RNG that doesn't make them lose. RNG rewards are exciting, since you never know what you're going to get. But everybody feels like the computer is cheating when it kills you with a high-roll critical hit.
Or if RNG consistently gives you bad results several times in a row. (ex: Such as in Pokemon where you can sometimes miss two or three times in a row with a 90% accuracy attack.)
so RNGesus mind our actions
btw thanks of teaching us :D
subbed
thanks!
I don't think the absence of RNG pushes away bad players. I don't know about you, but I've never thought "gee I keep getting my ass kicked in this game, I wish there was an element of randomness".
Especially in games where you play against another person, a game with more randomness will generally increase the win percentage of the worse player more than better player. Think about a chess grand-master, I could play them 100 times and I would lose 100 times. In hearthstone playing against a pro player if we each play the same deck with a lot of random effects I could probably win like 45% of my matches against them. I know winning isn't the only thing that keeps people playing games, but I certainly would not want to keep challenging the chess master.
I can already tell that part of me is going to hate myself for watching this. Came from the Mario RPG vid...
Seriously phenomenal.
But I have a few comments. You've mentioned if you limit the set of a playlist by a certain condition, you lose randomness. I'd have to bring up the concept of infinity - the set of all natural numbers is infinite, whilst the set of set of all real numbers is also infinite, however, greater than the infinity of naturals. However, infinity is infinity, so having a larger set of values doesn't negate the fact, just like randomness with restrictions is still (pseudo)random. rand(6) and rand(12) will both generate "randoms".
Also I don't know if you know about Diablo III's system, but how can the clustering effect occur when the legendary weapon encounter timer resets every time you encounter a legendary weapon, and increments every minute you don't encounter one? Is the clustering effect continuously observed because the programmers somehow increase the rate of encounter once you get a legendary item? Why force the cluster instead of simply heightening the rate of encounter universally? (moreso a rhetorical question)
But really rad informative vid dude. Gaming needs more channels like this
P.S. would REALLY love to see a Psychology in Video Games vid lol
+Slick Sludge
The clustering illusion is just that: An illusion. Randomness randomly forming a pattern that happens to be somehow disruptive to player experience.
Randomness entails an element of chance. When the range of possible outcomes grows, individual outcomes become less likely, you are less likely to get lucky and predict the outcome; So, you could legitimately say that a die has more randomness than a coin. Restricted randomness is literally less random. Imagine a 1-sided die: It will always come up "1". No uncertainly, no randomness, no entropy.
I guess I've misinterpreted the video a bit. I thought the game drop rate was affected after the player received their first legendary drop. Thanks for the clear up.
I agree on the less/more random part, however, as long as possible output states > 1, we still have a [pseudo]random output. Yeah it will get less random, as I've mentioned with infinity, narrowing the amount of values you have will literally make the infinity less infinite. However, a smaller random output set is still random. And you can increase that output set by simply repeating the random output procedure: try predicting the output of 10 consecutive 2-sided dice rolls. You have a 1/(2^10) chance of getting it right. Pretty random.
+Slick Sludge True randomness does not exists, and your argument that there are different sizes of randomness is flawed.
firstly infinity is a concept, it's not a thing. You make the remark that one infinity is larger than another infinity and this is also a fallacy. Infinity has no bounds, so the idea that one could be larger than another is broken math. Example, 2 * infinity = infinity? wrong. infinity is not a number, it doesn't equal anything, your can't multiply it by 2. More over, if you had an infinite box of balls that each had an infinite weight, it would not weigh more than an infinite box of balls that all weighted 50 grams.
Having more options has ZERO effect on randomness only the appearance of randomness which is what the video was trying to explain in the first place.
In a later comment you say "try predicting the output of 10 consecutive 2-sided dice rolls. You have a 1/(2^10) chance of getting it right. Pretty random." your math is correct, that would be your chance of guessing all 10 coin flips (that's what a 2-sided die would be, a coin) But actually it would be even smaller chance of getting right if the order of those flips mattered. BUT if you flipped a million coins, (or 2-sided die?) your chance of guessing the ONE next value would still be 50% The only way of using your data of consecutive flips to get more randomness would be to take the expected value of the consecutive flips. But the expectation would not be any more random that any on of the original flips, because you only want 1 value (heads or tails, zero or 1, or whatever) and when you scale that back down, it will be in perceivable from the original random value. number of options is not what dictates randomness, as long as there is 2 options with equal probability, then randomization is preserved. Setting rules is what breaks the probability, and that is what destroys randomness.
example, If I want a random heads or tails. I can generate a number between 0 and 90823742938487809240120394710892734089127304971298347908127439081274098127098471290834790812347890124908123749801279803471290840921340912470912374092174091207402394789 If that number is divisible by 2, then its heads if not then its tails. The fact that I generated this data from a HUGE pool of possibilities is irrelevant. It will be just as random as if I randomly picked between 0 and 1. If you don't believe me, try it. write a quick program.
If you put a restriction on a random number it is no longer random. Because the probability of each value has changed, thus it is now deterministic (we can guess with some certainty). Pseudo random refers to the fact that the random number is generated in such a way that it appears random. This has to do with the transparency of the rules. If you figured out how numbers are generated, then its still pseudo random you just know the rules. If the consumer of the values are informed, (such as knowing that it won't play 2 songs by the same artists in a row, or won't play the same song twice) then it is considerable less random. In-fact, if you know that the playlist is 4 songs long, 2 by one artists and 2 by another, you have a 50% chance of guessing the 2nd song, and after the second song 100% certainty of knowing the last 2. So only the 1st song played will feel random, the 2nd will be random with probability 50% while the last 2 are completely deterministic and not random at all. The more "public" rules you apply to a pseudo random system the less random it is.
The whole point of this video was, the false randomness present in good games is tailored for the best player experience, not the truest of randomness. If I told you to draw random dots on a sheet of paper, it's likely that you would space them out almost uniform, but true random dots would overlap and leave large unsettling empty gaps on the page, and your brain wouldn't like it. You want something that 'feels' random, not something that is random.
The idea that restricting randomness doesn't make things less random is false, but the idea that true randomness is valuable only holds in simulations not in game play. As GameSoup said, if we knew all the rules to the universe, even the seemingly random interactions of background radiation from space would be deterministic, so it really comes down to frame of reference. If the observer knows nothing, its random, if they know anything, then its not.
True Randomness doesn't exist???? - Heisenberg's Uncertainty Principle
Infinity not quantifiable???? - as I've mentioned, the set of all integers vs the set of all real numbers. Both infinite, but the set of all real numbers is infinitely larger than the set of all integers. Georg Cantor's Theory of Infinite Sets.
"If I told you to draw random dots on a sheet of paper, it's likely that you would space them out almost uniform, but true random dots would overlap and leave large unsettling empty gaps on the page, and your brain wouldn't like it. You want something that 'feels' random, not something that is random." - why? I don't think even Xandros legitimately explained how it is an illusion if there's a clustering effect happening in many random cases. Why would random number generation always cluster about a similar value???? How is it an illusion if I'm staring at the graphs, and I'm drawing a fairly blatant conclusion pretty much all observers could agree on???
It seems what you're trying to say can be distilled to "perception is reality". And I agree - the end user doesn't want an actual random system, but a "seemingly" random system, which would naturally be pseudorandom.
So why the hell does a genuinely random system have some sort of illusive clustering effect??
Heisenberg's Uncertainty Principle has nothing to do with randomness, simply limits the accuracy of measurements as the act its self changes the result.
Your 'bigger infinity' concept is invalid based on how you tried to apply it. Cantor's theory is about types of infinity. I have had this discussion with math professors at my university, they have stated that infinity is a concept you can 'tend toward infinity' but you can never reach it. The idea that one infinity is bigger than another only holds for different KINDS of infinity. Not different expressions of values of the same kind of infinity. The infinity counting all natural numbers is smaller than the infinity counting all real numbers, because there are infinite values between any 2 real numbers. Infinity cannot be quantified, countable infinities are smaller than uncountable infinities. That is to say continuous real numbers have a growth problem, if you start with 0 and say count every possible value forever, you would never even get to 1. This is uncountable. When you are talking about 2 infinities from the SAME TYPE, one cannot be larger than the other, it can just be tended toward faster. If you are comparing reals infinity to natural infinity then yes real infinity is a larger set because it's uncountable. But you can't use both types of infinity in the same problem, it invalidates the system. Which is the argument you seemed to be trying to apply.
There is no solid evidence than anything is truly random. Background radiation and timing of radioactive decay are about the only things that are considered random while many other things can be modeled as random and get a valid result, they are not truly non-deterministic systems. Even the most advanced random number generator hardware sometimes has anomalies that show the possibility that all randomness is an illusion, on the grounds that we don't know all the rules and starting conditions they appear random.
How can you come to the conclusion that RNG (particularly, cases where RNG decides between a clearly superior and inferior option, such as getting/not getting a particular random drop) enhances games, after listing several occasions where people spent serious effort to dismantle and circumvent said RNG? Clearly, for these people, the RNG made the game worse, not better. Also note that the main dividing line in people's preferences lies not between truly random and pseudo-random RNGs, but between ANY kind of RNG and a purely deterministic system.
"But if you you'd rather have hundred line blocks in row, then RNG just isn't your thing..."
I'm not even sure where to start:
1) It's a false dichtonomy.
2) While a Tetris game with predetermined piece order sounds ridiculous, the Candy Crush Saga is a very successful example of applying handcrafted level design onto an originally RNG-based formula: czcams.com/video/Sz4WXVNq7v8/video.html
3) RNG is not necessary for replayability. The Deus Ex games, AFAIK, has completely deterministic gameworlds, yet considered highly replayable. There are several other borderline examples, where RNG / dynamic systems only determine minor things, like what kind of consumable will drop from this monster.
4) RNG doesn't entices skill or informed decisions... it diminishes the effect of skill and informed decisions, because even the most skilled player can be screwed over by bad luck. An RNG loot system also diminishes player agency by denying them the use of a particular loadout, or forcing them to spend time grinding a particular activity. Such negative effects of the RNG-based systems are sorely missing from your video.
He also claims that tetris isn't random because it comes in blocks of seven. Which is false. Having rules for random outcomes doesn't make it less random. A dice have rules as well, you can't get a 7 out of pure chance. You could also have a dice with two fives and no six, changing the expected outcome. That doesn't change the fact that it is a random order. He even states that the randomness was intentionally reduced". It wasn't it is still random. the VARIANCE was intentionally changed.
Variance is a property of randomness. You can not talk about degrees of randomness or something being more or less random than anything else. What he really is talking about is variance. And the way he is describing is, I believe the variance is increased. What kind of block that comes up have even more variance, making it as likely as possible that you do not get any double blocks. Increasing variance doesn't mean that tetris is less random. The entire video is riddled with misconception of what randomness is.
If you really want true randomness quantum physics works fine. You can not predict quantum randomness no matter how much information you have.
actually the randomness succumbs to the same knowledge that you talk about with the coin. if you know how the dice is weighted and throw the dice in the same way every time you will get the same result
Lance Grabow not necessarily true since there are factors beyond your control that can slightly alter the die roll. maybe in a vacuum though
RNG is a lot like salt, it adds flavor to the games. However, too much of it can ruin an experience. Random-based progression in games really get my goat, and even games that have RNG for 100% completing games also irritate me. There is ONE heart piece in The Legend of Zelda: The Wind Waker I have not got because it requires you to play that Battle Ship style minigame flawlessly. But since this is Battle Ship, it requires guesswork and luck rather than actual skill to complete this game in the minimum number of moves it requires. The first round of that minigame, however, is fine in my opinion because, even though it is still random, it at least gives you a fair sized margin for error so that it doesn't feel like the game is just refusing to let you win.
I'm sure there's some guide on the internet somewhere that has what all possible outcomes for that battleship minigame are, but I have a slight personal bias against using guides. Which is an odd thing for me to say considering you need a guide to do most of any of the optional things in Final Fantasy 12, yet I enjoy that game.
Slot machines are required by law to have certain return rate(80%-95%) so there is a reason to play after someone who lost a lot.
+peterruskov But over what period of time?
I know lottery and scratch-off companies have to send their production code for third-party review and thought slot-machines have to do the same thing. It would be possible to mathematically prove what return rate your machine will have without programming it to compensate for winning or losing streaks.
+r hill on the slot machines I played it was not timed ,but after set amount of spins they had to be restarted.
+peterruskov
Man, that sounds pretty fascinating.
Thank you for the video. I think your dead wrong when it comes to randomness not affecting the player skill. At 25:25 you stated it rarely impedes upon the skill of each player. At a certain point, there is too much RNG, and the outcome can not be determined by player skill. For example, it would be like playing rock, paper, scissors and each time you made a choice, that choice changed randomly to either rock, paper scissors, when do you finally say "what's the point of playing, if I can't influence the outcome?"
+jchart7 This is a great point! I think that Hearthstone has managed to retain a good balance of skill and RNG, though over time the more recent expansions have added significant amounts of randomness. But all the skills you use in a card game like Hearthstone still apply (card advantage, playstyle, board control, synergy), and the best players can consistently reach the top of the ladder. If the legend rank was full of a random assortment of Hearthstone players each month, that would definitely be a problematic amount of randomness.
16:25 - 19:06 Yeah I like FF2's the best. One, highest minimal steps per encounter. Two, FF6's example of it being anywhere from 2 to 92 steps is fucking ludicrous.
22:39 Hah, nope.
So, you say in Hearthstone, any player can beat a skilled player? I know what game I'm avoiding like the plague.
Hearthstone has some randomness to it, like any card game. The cards themselves have become a bit less random with the last few expansions. Over time, the best players always rise to the top, but there are some frustrating tilts where you lose to bad players because of your bad luck and by no fault of your own. I still play the game a lot and enjoy it. -Joe
Game Soup If I want to play a luck-based card game, well, there's always _Yu-Gi-Oh!_
But uh don't mind me, just bitter about bad RNG. Could be a inverse Gambler's Fallacy or something to do with the Availability Heuristic, but I think it's something more than that.
how is the 92 steps calculated? isnt the formula "192x/256>255" to calculate max steps (given you roll 255 over and over)? That gives me x>340
procrastinator Small nitpick but I think you mean "x>=340" and if the max were indeed 92 it would be "192x/256>69"
Not sure why this is marked as the answer. The video says the max is 92 steps, which would mean there had to be a 0-69 roll. Yet there is a 0-255 roll. I still dont get how thats supposed to create a 92 max step situation.
The number I thought of was seven.
Well, quantum mechanics is most likely truly random. See Bell's theorem.
Needs more xcom
I think my copy of Binding hates me.
Laplace Demon
Three minutes in and the BS was too deep to fathom, I bailed at that point.
Hearthstone vs. Magic
Input-driven RNG - I recommend the channel pannenkoek2012. He mind-controls the entirety of Super Mario 64.
Quantum physics.
You are conflating constrained randomness with pseudorandoness, which are not related to each other.
True randomness is not an unfalsifiable claim. See Bell's Theorem. It is not correct to say that apparent randomness is actually just hidden causality, and it has actually been proven that this is not the case. I understand that this is not intuitive, but it is absolutely not a God-of-the Gaps argument. The "if only we knew every detail" supposition, which is stated in this video and required by the assertion you're making, is pure fantasy and is forbidden by the Uncertainty Principle (as well as Gödel's Incompleteness Theorems). String theorists know this, and operate under the requirement that any formulation of their work has to be compatible with Bell's Theorem and the Schrodinger Equation, and thus has to be able to produce truly random results. Source: I haz physics degree.
Oh, and hal is right about what he/she said, too.
***** "I know that science presupposes the uncertainty principle." - This is false. The Uncertainty Principle is not presupposed; it is mathematically derived from the Schrodinger Equation, which is the most durably accurate finding in all of science. I agree that it is not something you can take as an absolute truth in the strictest possible sense. However, it is an inevitable consequence of the best understood and most accurate framework for physics that we have.
"I'm responding to people saying that whether something is truly random or not, can be known with absolutely certainty." - Are you? Who said such a thing? Not the video, not hal, and not me. The irony of this is that my own comment was a response to your overconfident assertion that all randomness is an illusion engendered by our ignorance. It is this claim, and only this claim, which matches the criteria you described. So...you were arguing with yourself?
"How do you know that quantum fluctuations is purely random? As a scientist, you must base your conclusions on what you observe about it." - Yeah, that's exactly how I do know, because a huge amount of data and mathematical deduction points to exactly that conclusion. I hope I don't have to spell out the fact that when I say I know, I am not saying there is literally no room for error, nor ruling out solipsism, nor anything like that. I am saying only that it is the overwhelmingly likely conclusion based on the best available data and methods of inference, and that you have misrepresented it.
"Whenever something is too "random" to understand fully, you can only prove the fact that it's 'random' as per the scientific definition of randomness." - You say "the scientific definition of randomness" as if there exists only one such definition and as if you can say definitively what it is. You also implied the same thing in your earlier assertion that no randomness exists. However, this is misguided. When I say that there is evidence of true randomness, I am not saying that the universe manifests any possible interpretation of the concept that you could come up with, nor that it requires you to have no information whatsoever. I am saying only that there exist rules by which the observable universe apparently works, with a very high degree of statistical certainty, which prevent you from having ALL the information about the position and momentum of any system.
"But you can't know anything about what that randomness is, or else it would be known to you and thus it wouldn't be random." - You are making the same mistake that this video makes in saying that constrained randomness is not truly random. Randomness does not require that you have no information whatsoever; it only requires that you cannot have it all to an unlimited degree of precision.
"The uncertainty principle dictates that nothing can ever be known as absolute truth." - No, it doesn't. It says only that the position and momentum of any particle (along with any emergent properties of systems that involve these) cannot be known with absolute certainty. But that's irrelevant anyway because I never claimed absolute certainty about anything. In fact, I was pretty sure you did, when you claimed that randomness doesn't exist. Perhaps I should have given you the benefit of the doubt and inferred a lower level of certainty, but I was simply taking you at your word.
"Doesn't Bell's Theorem assume that the current understanding of Quantum Mechanics is correct? That we understand how it works?" - No, it doesn't.
"Like any other theorem in existence, it obviously assumes something about science. Obviously we too need to assume that the theorem is correct, because that's what the data shows. But a theorem isn't itself true. It's just a statement being made that coincides with the data we have - or else we wouldn't have called it a theorem." - That is not what the word theorem is generally understood to mean. More pertinently, there is nothing about the title of Bell's Theorem that implies what you're inferring about its place in scientific epistemology.
"And you're basically saying that there cannot exist any unknown data that we could learn at some point?" - No, I never said any such thing.
"Now who's ignoring the uncertainty principle?" - Yeah, that would be you, because apparently you don't understand what it even is, and neither did you display a good understanding of what I've said about it.
"I honestly don't think you understand what I'm talking about here." - That's correct. It makes little sense to me, and therefore I cannot claim to understand it. If there's anything else more specific that I'm misunderstanding though, then I would still appreciate you pointing it out to me.
"There's an underlying epistemology and ontology behind the concept of science. There's a way to acquire that data, to recognize it, to contextualize it given the human mental faculties and perceptions.
When you lose track of that foundation, you're no longer being scientific." - In order to lose track of that foundation, you'd have to first know exactly what that foundation is, and I really don't think I, nor you, nor anyone else, can honestly claim to know such a thing. I'm not sure I even know what your point is, since I doubt you and I actually disagree about scientific methodology in any way that would manifest in practice.
"More to the point: Explain to me in what way a pseudorandom number contains less entropy than a random number?" - I didn't claim that it does, and such a claim wouldn't even make sense given the accepted definition of entropy in physics. Hint: It's not "disorder," and if you think it is, then you are working with a rather crude approximation. In any case, a number itself can't be random nor pseudorandom; the use of such terms, at least in programming, refers to their relationship with the user and not with any property of the number itself. This is emphatically NOT the case with entropy in the thermodynamic sense.
"Obviously there's grades or complexities of randomness. But people are generalizing the less complex ones into the term 'pseudorandom'. So where do you draw the line? What random number is just random enough to not be called pseudorandom?" - If you are looking at randomness vs. pseudorandomness as a gradient, then we are talking past each other because you are not using the terms in a way that would be applicable to QM or thermodynamics at all.
I manipulated RNG in Suikoden 1 for drops.
+Sarek Mather I tried manipulating RNG in Final Fantasy 12, but it is unbelievably complex. There are excel spreadsheets with thousands of entries that you can use to "locate" yourself in the game's RNG loop, and then you can force certain items (Zodiac Spear, perhaps?) to drop 100% of the time. Loopholes in RNG can be pretty entertaining sometimes!
+Game Soup
Wow, I checked that out. That's pretty amazing!
+Game Soup What happened in that one corridor full of monster on Final Fantasy 1?....the game was rng no?...then why there is a corridor where each step is a monster? T-T (I knee after a while it was a thing everyone find themselves against.....and god I got out alive somehow)
Beware the idea that hidden mechanics enhance a game. If the player isn't allowed to understand the game then that's a problem.
The person that deciphered the enigma machine was killed shortly after the end of WW2 because he was gay. I might be remembering wrong.
Alan Turing was imprisoned and sentenced to chemical castration.
He was later found dead in his lab, official statement says it was suicide, friends and family disagree. Note he manipulated cianide salts in his lab, which are deadly even in small amounts
if you go too far with random not existing theory and you land on predeterminist philosophy
I don't know how you make a video like this and not even say perlin noise.
im glad life does not use true randomness. and at the same time no randomness at all would suck.
Only lazy developers use rgn.
oxis77gas Did you listen to, like, ANY of the ending? Sure, lazy developers will sometimes use RNG poorly, but more experienced developers can use it to a good extent. Take Spelunky. If every map in the game were pre-made, the game would be less interesting, not to mention a much larger download size compared to a level making algorithm. What happens when you run out of levels? Do you go to Steam Workshop and HOPE there's something interesting? And then there's another question; How do you keep veterans of the game interested? If your game has pre-scripted surprises, your every day game hacker can look into the files and see exactly when, where, how, and why everything will occur, no more, no less.
Party Puppiez Rng maps will be good as long as you have enough texture variety, take for example Elite Dangerous, 100 bilions of star system, but only cca 100 different planet and star texture. But the real reason why I hate rgn is WOT.
Randomness in videogames is cancer, end of story, no need for a video.
I'm 1 minute in and already turning the video off. Not talking about the role of RNG but instead talking about the logistical implementation as a software feature? Then what's even the point of this video?
RNG is lazy game design, end of story. I was hoping to see another person's perspective on this topic but am disappointed. Please don't use such misleading titles in the future.
+GodOfReality He did talk about the role of RNG. And he compared the "lazy" RNG that just uses a random number with the more controlled RNG that's suited for the best experience