The WORST Zelda Sidequest is Even WORSE Than We Thought...

[@MrDrBoi]

Wow, I honestly can't believe this quest is even worse than I thought it was! I guess it really is best to just bet one shell until the last few figurines, but holy crap is that tedious lol Great video, really enjoyed the editing and appreciate the shoutout as well!

[@ThePlantCommander]

Thank you, and no problem! Yeah, the one shell strat really is the only way to go. At the end of the day it's just a weighted coin flip, so even with the superpower that is cheating and rewinding time, it's not even worth bumping up your threshold when you can literally walk away and come back, and the game might just tell you you'll win by default, lol. Glad you enjoyed the editing, animating all of those charts took a ton of time and effort so it feels awesome when somebody appreciates it who knows how tedious the process can be haha

[@erlendvageskar3356]

Both of your videos were excellent! I had a fun time and now I know what to expect if I ever ATTEMPT this.

[@HamsterBaddy]

I still think the record player in the house makes the rewards better than only a single heart piece - yes, today it really doesn't matter, but when the gane came out, especially as a kid, it did - still not great tho

[@RobertJW]

I watched the original video a while ago but I had to re-watch it just now for the full context!

[@Bluedanes]

I used to speedrun this game and I can explain a bit what's going on here. There's a bit more going on behind the scenes, but you are entirely right about how the threshold and the two figurines (new vs repeat) work. However, the game actually does decide those values when you pull the lever. The reason you're seeing the same result in your experiment is because how the game's Random Number Generator (RNG) works. Minish Cap, like a lot of old games, uses an RNG algorithm that will take a number as a seed, use that seed to create a number, and then use that result as the seed for the next RNG call. And in the case of Minisj Cap, there's actually not too many things that will call the RNG and therefore change the seed. Examples of things that will call the RNG are item drops, enemy movement, and Link using certain items. This means you can get very consistent "random" results if you are careful about your inputs, since Link's actions are the only thing in this room that can change the RNG seed. In your experiment, you just weren't taking any actions that called the RNG other than the other pull. If you open up the same save state from your test, and then swing your sword, roll, or use the Pegasus boots before pulling, you will see a different threshold and awarded figurines. Another fun fact is that the game resets the initial RNG seed when the GBA powers on and starts the game. So this means that if you're careful and precise, you could power off and on the game, go to the figurine shop and get the same resukt each time. Speedrunners abuse this as well as how the RNG algorithm works to generate "perfect luck" to collect all the figurines as quickly as possible. Though even with that it still takes them over 30 minutes. If you want to know more about this little overlooked Zelda game that I really love, feel free to ask me, or I can direct you to the Minish Cap speedrunning discord where people who know much more about how the game works can give much more in depth answers!

[@ThePlantCommander]

I JUST replied to a comment where I mentioned the GBA power cycle, lol. I had a feeling it did something like that, but it's fascinating that you can actually abuse it that way. Interestingly, even though I didn't know this at the time, I did try some of those actions (mostly aimlessly rolling around the room, lol) and still got the same result, so it might be possible that either it didn't trigger the RNG at all or, somehow, just happened to roll the same thing. I'm definitely not a "dedicated" speedrunner, but I dabble in it here and there and I incorporate speedrunning elements into my let's plays, so always love learning stuff like this. It's incredible how much you can learn about how games work when you just wanna go fast!

[@Bluedanes]

@@ThePlantCommander Another quirk about the figurine pull is that the RNG doesn't get a number from 0-100, but actually 0-128, and if the number is greater than 100 it just calls the RNG function one more time. There are some times where it will just happen to get a number over 100 several times in a row, and when that happens it just keeps calling it until it gets a valid number. It's entirely possible that whatever your RNG seed was the game had to call RNG like 6 times when you pulled in order to get a valid number. Almost like a RNG call buffer. So if you, for example, rolled 3 times (each roll calls RNG once) before pulling, the game would still be in that buffer and end up needing to call RNG 3 more times, ending up in the same result.

[@ThePlantCommander]

@@Bluedanes Huh, fascinating. Let's just be thankful that it doesn't round higher numbers down to 100, or this would be even more unfair than it already is XD

[@erlendvageskar3356]

I would love to know the exact steps I meed to take, so that I get 100% guarantee everytime by just paying 1 shell every time :).

[@rkroll17]

I was wondering how speedrunners handle the rng. Do you guys have/allow a tool to be used to manipulate the rng?

[@PringoOrSomething]

99% of gamblers quit before they make it big

[@dusty0896]

That 1%:

[@abloogywoogywoo]

You couldn't win a figurine in a rigged game? Skill issue.

[@PringoOrSomething]

@@abloogywoogywoo Real

[@vinnyandlin8510]

And we all know that's some casino propaganda. Gambling is rigged. Best way to win is to not play.

[@Triforce_of_Doom]

Oh gotta love when the "odds" are just a weighted coin flip

[@tuluflulu]

Mario Party DS predetermines your dice roll.

[@Triforce_of_Doom]

@@tuluflulu there's a good few Mario Parties that do

[@tuluflulu]

@@Triforce_of_Doom I didn't know that. I thought it was just the DS one

[@jwjeff11]

​@@tulufluluevery video game does. The only difference is how complicated they wanted the math to be and what the inputs of the equation are.

[@emdivine]

@@jwjeff11 no, predetermining the rolls means they'd be consistent each time they're done regardless of whatever else happens. Very few games do that, since multiple different things call the RNG. But generally there will usually be one single seeded RNG, so the entire list of random numbers can be said to be predetermined. But each user of those numbers is sensitive to the other users.

[@WalkerSunriseChannel]

That’s literally how Fire Emblem rolls for whether attacks hit or miss, that’s fucking hilarious

[@ThePlantCommander]

In a game with permadeath that sounds like actual bullshit lmao

[@WalkerSunriseChannel]

@@ThePlantCommander I mean, you know what your odds are before committing to an action at least. If you play well you won't have to gamble on low odds often if at all XD

[@ThePlantCommander]

@@WalkerSunriseChannel that's the problem, I'm not that good xD

[@mauricesteel4995]

and that method of calculating RNG isn't going away as now Fire Emblem has limited rewinds incorporated into its gameplay.

[@sam7559]

As a fire emblem player I don't see the issue. If the game says 90% chance to hit, I have a (ignoring the fact that true hit is a thing and isn't something I want to explain right now) 90% chance that the RNG value pulled was under that threshold. It's still a 90% chance, just that I need to move an arrow around if I want a second shot at the RNG working for me.

[@octoink145]

I just got done 100% minish cap on switch, and I betted 1 shell and then abused the rewind feature to see if I got a new figure. Leave the room to get a different figure, then repeated it the whole time

[@chapinchancey7622]

Or roll around the room. That changes it too.

[@Lady_Papillon]

omg I'm so happy that I wasn't the only person who thought this was THE WORST sidequest in the game. I ALWAYS HATED THOSE SHELLS, HATED THOSE FIGURINES, HATED THAT GUY, i'm so happy to know so many of us lived that same collective rage-filled experience

[@reneethefox4797]

For what it's worth, this is basically how randomness/probability works in every game. Generate a random number from 1 to 100 if it's less than the probability -> success, if it's greater -> failure. The only difference between this method and the standard method is that it generates the number when you enter the room instead of when you pull the thing. Which is only important if you are using cheats to rewind time, or if you're TASing since you can see the exact amount you'd need to bet every time. For a casual player playing on a GBA, it's neither better or worse.

[@coreyhall5065]

If anything, with save states this games method benefits the player by having a static rng seed you're guaranteed to roll back to rather than have it tied to a certain frame. Otherwise like you said for a casual player there really is no difference and it still functions the same as most rng systems in games and I see no real issues with it

[@Ixmore]

That's most likely the case as my experience with programing. Sometimes you have to create an illusion of what your trying to create. Take Starfield for example, it's likely the main point on how the game was built was to create an illusion that the game is bigger than it actually is

[@williamdrum9899]

Funnily enough, since computers work in binary rather than base 10, percentage-based RNG is less efficient than using 1/(2^n) which a computer can do by applying a bitwise AND to the RNG, no actual division, exponent, or floating point calculations needed

[@joschwalb931]

Thank you so much reneethefox4797 for this explanation, because if you hadn't made it I would have but with much more hatred fueled language. my friend and I were screaming at this video the whole time while watching it, we both got so angry

[@brandyjohnson9272]

I mean... Unless you cheat this, in effect, is the EXACT SAME as what it was originally thought to be.

[@aughlnal7333]

yeah, I wonder what he would think if the repeat, new figurine and win chance where calculated when you press a on the shell menu instead of entering the door. It might seems more 'honest', but in effect the outcome is identical

[@nngnnadas]

It probably does that too, since you can bet again witout exiting.

[@trebmal587]

Yes, if you raised the odds of getting a new figurine to 80%, you have an 80% chance of the number determined by the game to be lower than that, and 20% that it isn't... that is, if any of the individual percent had an equal chance to get picked. So yeah, it doesn't change anything for the player... It's an okay way to determine the randomness of the side quest. There is no actual scam.

[@lukelcs8934]

@@trebmal587exactly, this is how ALL rng works lol. It’s only more frustrating for save-state users, but it’s still doing exactly what it tells you. (Not to downplay how frustrating it is by any means XD)

[@ThePlantCommander]

@@trebmal587 exactly this. More than anything this was just a fun little thought experiment and the KZbin algorithm loves when videos make claims like this, so I had some fun with it, lmao. The odds are technically accurate, but this does make it even more clear that betting only one is optimal (since a losing bet is still losing, whether that's 1 shell or 30). But that was already the case before knowing this just because of the math.

[@tuckerrouse4850]

There's actually a much simpler way to explain the cost inefficiency. With one seashell, you get the base percent chance, between 1% and 100%. With each additional shell, you get an additional 1%. So, if your base chance is 10%, and you spend 1 shell, you have a 10% chance. If you spend 3 shells, you have a 12% chance. You have *tripled* the amount of shells you've spent for a much smaller increase in chances. It only becomes statistically smart to spend more than one shell at the point when you're down to 1% chance. At that point, spending a second shell (doubling the cost) will boost you to 2% (doubling the percentage) Does this make sense to everyone?

[@ThePlantCommander]

This is actually a really good way of putting it

[@elliotlevy8610]

It is even worse than that though. As you said betting 3 when you start at 10% only gets you to a 12% chance. But now compare that to betting those same 3 shells one at a time. Assuming the base chance remains the same, statistically you have a 27.1% chance of getting at least 1 new figurine! You can actually graph this relationship and see just how bad it is. If we let the base percentage (b) be the theoretical value at 0 shells (so adding 1 gives you the starting percentage), the percentage chance of getting 1 new figurine (y) by adding more shells (x) can be expressed as y=100×((b/100)+0.01x). By comparison, betting 1 shell at a time can be expressed as y=100(1-[(1-[(b+1)/100])^x]). I know that looks a bit complicated in that format, but when you graph the two the first is just a straight line while the second looks like an inverted exponential decay function (which it is). What you find in these is that, for nearly all values of b, it never pays to bet more than 1 shell as you will have a lower chance of victory unless you bet enough to achieve nearly 100% chance. Even then though, you effectively are giving up the chance of getting 2 or more instead by betting so many. (e.g. Adding 1 to a starting 99% means you have a 100% chance of getting 1 instead of a 98.01% chance of getting 2. Betting 2 one at a time only has a 1.98% chance of getting exactly 1 and a 0.01% chance of getting 0 new figurines.) This only changes at very low starting percentages, where the difference in chances shrinks between the two methods. Even then though, it's not until precisely 1% that adding shells gives you better odds than betting the same number one at a time.

[@philippeklein7876]

Yes but since you have several thousands of shells during an entire playthrough, it's not practically the best strategy to bet 1 shell every roll. Betting more than one shell is faster, but it's more expensive. I'm looking at the best strategy, which optimally balance the cost and the speed

[@homestar92]

The thing is, the difference as to when it rolls is immaterial when playing without save states. It doesn't change your odds. It is only through emulators that this difference begins to matter or even be observable Your analysis is correct though - It is better to bid one shell each time. If you have 20% odds betting one shell, then there's a 90% chance you would get a new figurine after spending 10 shells one by one. Whereas bidding 10 shells on a single roll would result in only 30% odds of a new figurine. The odds don't tilt in favor of betting enough shells to get up to 100% until you're down in the very low single-digits.

[@emptylog933]

Just an implementation detail, the fact that the threshold is determined before you pull doesn't change anything in normal play.

[@williamdrum9899]

True, it only matters if you can use save states or see what the seed is. Then those percentage the game gives you suddenly become blatant lies. Probability is weird like that.

[@andreaquadrati]

The funniest part is that technivally the other brother in town has a gambling spot, but feels a lot of guilt and recommends you not to gamble at all. The real bad bro is this unapologetic dude.

[@DanielLCarrier]

Imagine someone offers to flip a coin and if it lands head, you win. It's a 50% chance of winning. Now imagine that instead of that, they flip the coin first, but keep it covered, and you have to guess what it landed on. What are your chances of getting it right? The answer is 50%. Unless you abuse time travel to know ahead of time what it landed on, it's equally likely to be heads or tails, and you have a 50% of guessing right. The same applies here. If you don't abuse save states, you'll have no information on what number the computer picked. It's a 1% chance of picking any number, and if you pay enough shells for a 53% chance of winning, that means 53 of those numbers will net you a win and you're 53% likely to win.

[@williamdrum9899]

Another interesting detail. The original Pokemon games based its RNG on 0.5^n rather than a percentage. (And any game that doesn't tell the user what the odds are ought to do the same, simply because it's the most efficient method. Despite the formula "0.5^n" the computer calculates it with a bitwise AND, no need for floating point, division, or exponentiation.

[@augustday9483]

Save states take a snapshot of the game's ram, so of the RNG call happened before you make the save state, the result will be identical every time you reload the save state. So in this case you'd need to make your save state before entering the room, since that appears to be when it does the RNG.

[@ThePlantCommander]

Yeah, that's pretty much the gist of it.

[@williamdrum9899]

Schrodinger's Coin: Simultaneously rigged and fair as long as you have no knowledge about the RNG seed value

[@MudakTheMultiplier]

I think you've confused "%chance to win" with "%chance to pick a winning number". Yes, of the ~80 options you can choose from you get the largest chance to win with the smallest spent by only spending one, but when you spend one for the 20%, you do have a 20% chance to win.

[@ZT1ST]

It's not that weird to me that the RNG works that way - with the rewind, you're rewinding the RNG seed itself - and the way most game's RNGs work with their seeds, that means you're going to roll the same number. Like if you were annoyed about a 52-card deck draw, so you put the cars you drew on the top of the deck, and then pulled again. The percentage chance of it being the trophy you want is going to be tied to what the next number in the RNG is, not what the percentage is. If you can find a way to rewind and then burn an RNG (Probably by cutting one piece of grass and then immediately coming back, or finding an NPC/enemy with randomized movement make one turn) then come back...the RNG would've changed, and then the number would've been different.

[@williamdrum9899]

In short, the percentages are a lie unless you're playing the game "normally" (i.e. no save states or ability to view the game boy's RAM)

[@williamdrum9899]

And that last paragraph only matters if you knew the seed. Funnily enough this same logic is handy in card games like magic the gathering, but in reverse. A card that lets you shuffle the deck is a useful effect in and of itself, provided you know what the top card of the deck is before shuffling. There's a card that says "Draw 3 cards, then put 2 cards from your hand on top in any order." The strategy is to put back two cards you don't care about, then use another card effect that makes you shuffle, which makes the next two cards you draw random once again instead of guaranteed poopoo.

[@kooriicolada]

I just did this quest and while playing around with the save states I figured out the fact that the figurine seemed to be chosen before the pull. It's interesting to see what was really going on there! Great video :)

[@lupislord]

This is pretty standard behavior for a Random Number Generator. Just think of it as savestates rewind your RNG just the same as they do your shell count, for instance. Now lets try thinking about how you can exploit the RNG! Here's my idea: its basically a modified "binary search" algorithm Step 1 - Try 1 shell (if that works your done) Step 2 - cut the odds in half (so if you were at 20% for 1 shell, jump up to 60%) If that fails, cut the odds in half again (jump up to 80%) If that succeeds, go back halfway from your previous guess (to 40%) Repeat step 2 until you find the exact % needed process might go something like 20%->60%->80%->70%->75%->77%->78% (end) This finds the target % much faster than incrementing 1 shell at a time You could also take shortcut and quit as soon as the shell count gets to high for you (and by quit I mean reset the room and try again) Not sure if this is actually faster than just doing 1 shell at a time and resetting, probably depends on how bad your luck is

[@ThePlantCommander]

Interesting thought experiment, but in reality that's probably gonna take longer. With how many text boxes and animations that play out for every single pull, rewinding that often is probably just a waste of time. Personally what I would do is create a save state every few minutes and if I get, like, 15 repeats in a row? Just reload it and force the game to re-roll to save resources. Your idea is probably really good in terms of resource conservation, but if you're just trying to get the sidequest done faster, it's more effort than it's worth.

[@ShyGuyXXL]

If they EVER remake this game, they HAVE to completely overhaul the interface of this machine, so you can just press 1 button to instantly buy another figurine. Then it wouldn't be so bad to just spend 1 shell at a time.

[@Aileil]

No kidding. I think the only two things that might take more time are tracking down kinstone fusions and beating Vaati at the end (thought that one is probably just me; my brothers generally aced the combat sections). I would also rejoice mightily if I didn't have to leave one of the three multi-colored sisters homeless. There is a perfectly good plot of ground for a third house and everything and this poor gal is left in the hotel reflecting about how it's more lonely with the other two gone. 😥

[@Knightmessenger]

Or just let you hold more shells at once so you don't have to keep coming back.

[@MajoraZero]

Spending one shell at a time increases your odd, but you're paying with your time!

[@willverschneider1102]

I have a better idea, change it completely! Make it so that when Link speaks to an NPC or defeats an enemy for the first time, it drops their "shell". Take that specific shell to the figure maker to add it to the collection.

[@Aileil]

@@willverschneider1102 I could live with that arrangement.

[@jeremycampbell4939]

I'm confused. You're saying the percentage is a lie, but explaining in great detail how if the game says you have an X% chance to get a new figurine, there is an X% chance you get a new figurine.

[@Shadowvanify]

The difference comes down to the resource cost. If the percentages worked as expected, then each shell you spent would go towards increasing your chance of winning some amount, regardless of how many or how few extra shells you spend. But with a predetermined threshold, spending more shells does nothing unless you happen to spend enough to hit the unknown threshold. Otherwise those shells are essentially wasted. It’s more cost efficient to spend one shell each time until the threshold is randomly chosen low enough than to spend multiple shells trying to improve your odds.

[@PS10111]

​@@ShadowvanifyI don't know if this is correct. Intuitively, it doesn't seem that the random number being pre-selected makes a difference. If you were to toss a coin and make a guess before looking, you'd still have a 50/50 chance. I wrote a quick simulation in Python and the odds are accurate to the player's input percentage.

[@Kafaldsbylur]

@@PS10111 It doesn't make a difference unless you have a way of rewinding time. Spending the shells to get that 75% odds *did* give you a 75% chance that the number the RNG had generated was under 75

[@xerneas0723]

@@Shadowvanify The fact that the best strategy is to spend one shell has nothing to do with the fact theshold is predetermined. Its is related to the concavity of chance of winning regarding the number of shell. Each shell you add increase the odd of winning but by a smaller and smaller percentage.

[@sam7559]

​@@Kafaldsbylurbut it does give you a 75% chance that it rolled at or under your number, the computer just did it's RNG before you picked your shells. Yes grand scheme is 100% or 0% binary, you get a new figure or you don't. Saying that it isn't chance like presented is like saying a coin flip isn't 50/50 because it's going to land on heads because you saw the future.

[@charleslee8313]

I'm quite surprised that Game Theory never did an episode on this.

[@benjaminhutz2405]

well the thing is it does not matter if its a 1% chance to get a duplicate or if the game rolls a 100. The Math is the same... because the chance for the game to roll the 100 is indeed 1%. The only difference is that you cant abuse savestates to get the figurines, because the number is set from the beginning

[@rawman44]

You can abuse savestates, you just have to save at the top of the stairs lol

[@Iamverylazy]

I'm a little confused as to what you're upset about. The game picks a random number and you set the odds of getting a new figurine. So what if it picks the rng before you set the odds? It doesn't affect the odds themselves. If you set the probability to 75% than you have a 75% chance of meeting/beating the randomly generated number. It seems to me like that's working as advertised. That being said, the real issue with this mini-game is that it's mathematically inefficient to bet more than 1 shell at a time, but that's not because of the inner mechanics of the game, it's just due to the fact that betting one or two shells give almost equal odds of getting a new figurine but betting two shells literally cuts your possible number of attempts in half. And this only gets worse the more shells you bet. You have to pay an exorbitant amount of shells to make any meaningful difference in your odds that it's just mathematically better to only bet one shell at a time.

[@ahernplays2239]

All these years, all those attempts, all the times people said I should just bet 1 shell and I ignored them because I thought "But it's percentage based". So much pain and it was all for nothing.

[@lyricjam39]

Maybe it was made to teach kids that gambling is a bad habit.

[@kurocat471]

if played the intended way, this makes zero difference. it's not lying on the probabilities, it's the exact odds it says it will be. if you cheat with rewind or savestates, you have to reset the RNG seed by swinging your sword or changing screens to reroll and get the 'intended' effect.

[@linkdx7079]

Swinging your sword changes the rng? That's pretty funny lol

[@simonO712]

​@@linkdx7079I would guess that it's because Link has multiple "voice lines" to choose from when swinging the sword.

[@popsiclesshubert9609]

I honestly did not think that quest could be that bad- thank you, this vid was great

[@ThePlantCommander]

Thank you!

[@HelpTheWretched]

Even though it's well-known that most "random chances" in gaming are pre-determined at some point, it's always funny (and a bit maddening) to see it laid out so clearly. This reminds me of one of the 3DS StreetPass games, Flower Town, where each seed you harvest can have up to 3 possible breeds that it'll produce, each with a percentage "chance". But it's totally pre-determined, and if you're watching closely, there are 3 different ways to narrow down the possibilities, so you can virtually always know what you'll get. I demonstrated this once on Miiverse by posting a screenshot of the three possible breeds, guaranteed that it would be the one with the lowest percentage (11%), then posted the follow-up screenshot where I was correct.

[@ppowersteef]

Correct me if I'm wrong, but I don't see what makes the difference here. Since the treshold will reset each roll. The % is indeed a guaranteed win or loss, but only for the current roll, but the treshold generation is still random. Without savestates, you can only make one guess before the treshold resets again It's just pre-determined rather than on the fly, and that system only breaks if you cheat.

[@ThePlantCommander]

That's true, I tried to make it clear in the video that this is really only relevant if you abuse save/rewind features to try to cheat your way through the quest, but even that is largely pointless so you might as well just do the quest normally. Even trying to cheat the system isn't worth it. Interestingly though, I would like to mention that a former speedrunner left in interesting comment on this video mentioning that it's possible to manipulate what the threshold it gives you is, so that's definitely something worth looking into!

[@ppowersteef]

@@ThePlantCommander Yeah, I've noticed the speedrunner's take too. This video makes for an interesting find and take on a % calculator though.

[@kabzebrowski]

Save states are what break the system, as you already pointed out, and of course, since you can't do that in the original game, the % is accurate there. You're taking a guess at what number (1-100) the game chose, the higher you bet, the higher your chances of being above the threshold. There's no rigging, it's just a different way of doing RNG.

[@EtheRenard]

My favorite games are both Wind Waker and Minish Cap, but wow, I hate the figurine minigames so much xD The part that I hate the most is that your max is 999 shells, so you have to think about coming back from time to time to avoid wasting treasures

[@TheOtakuArtist]

This turned out so amazing, it looks so professional!! So proud of you, you should make more essay videos like this!!

[@TakumiJoyconBoyz]

There's honestly a few games I've seen with similar gambling mechanics where they lowest bet is actually the best in the long run because even if you have to do more rolls you save way more resources. The Senran Kagura games come to mind for their outfit roulette.

[@ThePlantCommander]

Which Senran Kagura games, out of curiosity? Coincidentally, I've actually been playing through PBS right now as I worked on this video, lol. I 100% completed Burst Re:Newal a few months ago and I'm working on the same for PBS right now, but I don't remember seeing any gambling mechanics in these games outside of PBS' cards

[@TakumiJoyconBoyz]

@@ThePlantCommander In most of the games there's an underwear lottery.

[@KurtisC93]

Why it is most cost-effective to use a single shell every time, even when the odds are set at 20%: • If you use a single shell, and run the machine five times in a row with that singular shell, you are likely to land a new figurine at least once. In essence, you've only spent five shells (or however many shells it took) until you got a new figurine. • On the other hand, if you increase your odds by adding more shells to a single spin, you're likely going to need to enter a number that is _multiple times_ what you would have spent if you just kept running it at one shell per spin until you got something new. To make it more likely than not for you to win on that spin (>50%), you'd need to spend tens of shells. To guarantee a win, you'd need to spend tens more. Even if the game wasn't "rigged", this logic would still apply.

[@nodrance]

This isn't a scam, the odds are exactly what they say they are. If you don't cheat, that is. Imagine for a second the game doesn't decide your number until you pull. You picked an 80, now it picks a random number between 1 and 100 and gives it to you if your number is higher. You having an 80 means that 80 numbers will give it to you, and 20 won't. Guess what, that's an 80% chance. And the math doesn't change if you generate the number beforehand and keep it secret. Even though now it looks like you're trying to hit a threshold, from an outside perspective it doesn't matter if the number you're comparing against was generated before or after you chose, it's still a random number and a random chance. And it doesn't matter whether it chooses the existing or new figurine before or after you pull the lever, either. You getting a random pre-chosen figurine is the same as getting a random figurine that's not decided until you pull. You broke the illusion by trying to figure out exactly when it's decided, but it doesn't matter when it's decided as long as it's random. It's the same as a dice, the second it leaves your hand everything is up to physics, but that doesn't make the roll any less random.

[@chaosprime1629]

to be fair, save states generally have pre-determined results in many games. even a turn based rpg this is frequent where a random battle will occur after moving just one step unless you change the variables such as performing different actions.

[@coomtheslayer]

This is just programming stuff to make random things function smoothly, this is functionally identical to the advertised mechanic

[@movezig5]

I think you fundamentally misunderstand how the RNG in a video game works. Let me try explaining it, and why this game might not be as rigged as it first seems! First of all, computers don't do anything randomly. If you tell a computer to do something, it'll do what it's asked, and nothing more. Because of this, randomness in games has to be generated in a "pseudo-random" way--it needs to feel as though the results are random, even if they're not. This is usually done through the use of an RNG function. This function is a mathematical function that takes one number as an input, performs some operations on it, and produces a different number as an output. The RNG function is designed to give wildly different results for different inputs, such that the results are unpredictable and _seem_ random. Of note, however, is that *given the same input, the RNG function will ALWAYS produce the same result.* Whenever the game needs a random number, it will call the RNG function, but it needs an input. What does it use as input? _The result it got the last time it called the RNG function._ (If it hasn't made an RNG call yet, it uses an initial value called the RNG "seed," but I won't get into that here.) When you made a savestate, you saved a snapshot of every value in the game's memory, _including the result of the last RNG call._ This means that the next time the game needs a random number, it will ALWAYS get the same one! In order to get the result of a percent chance, the game generates a random number between 0 and 99, and if it's less than the percent, the chance succeeds. But remember, in this scenario, the game always generates the exact same number. This explains the "predetermined threshold" behavior you observed. Same input, same output. Every time. (This also explains why it gives you the same duplicate or non-duplicate figurine each time--it's generating the same RNG values for deciding those too.) The only thing this doesn't explain is why the game gives different results when you exit and re-enter the room. I can only guess the game burns a few RNs when this happens, but it doesn't look like there's anything happening in that room to cause randomness. Source: I have a masters in Computer Science and I play Fire Emblem.

[@ThePlantCommander]

This is something I actually have a bit of experience in as well (nowhere near as much as you say you do, in all fairness, but I've done a bit of game development and IT work) and all of this is correct. That being said, we actually are able to figure out why exiting and re-entering re-rolls the threshold--that implies that it's based on the game's current seed, and the RNG is decided when you interact with the shopkeeper. Everything after that point doesn't change the threshold because it's locked once you talk to him, and the only thing the player can do from that point forward is choose their bet or simply say no. Something a few speedrunners have clarified in the comments is that performing other actions like using an item or rolling also changes the result, but simply walking around doesn't seem to do anything. You can get it to change by entering a loading zone or doing any action that affects the game's seed before the moment of interaction. What makes it feel deceitful (at least to me) is the fact that it is entirely possible for the game to roll a high number, and everything below that is an automatic loss before you place your bet. That means that if the game decides you need to spend 80 shells and you choose to spend 50, you just wasted 49 because you were predetermined to lose before you actually spent that much. The number itself isn't entirely inaccurate, but it's definitely presented to the player in a way that's meant to trick you into wasting resources unnecessarily.

[@affegpus4195]

Thats just how Random number generators works. Given the same seed(which is likely set when you start the game) It will always give the same sequence of numbers. I repeat: RNG will ALWAYS give the same number sequence for the same seed. In this case it will get 2 numbers from the RNG, one to check if it is a new figurine or repeat and another to check which one it will be. So yeah, nothing wrong with that. Thats why is called pseudo random, not random

[@rkroll17]

I mean the percentage isn’t lying cause the only way you could know that it decides the value before is with save states or by looking at the code. There really is no difference between the number being generated before you roll or after you roll. I still liked the video though.

[@ThePlantCommander]

It's lying in the sense that by raising your "percentage," shells are being wasted without any way of the player knowing that without cheating. Yes, the odds don't actually change, but anything you bet above or below the threshold is meaningless, and without cheating it's impossible to know what that threshold was. Basically, it encourages being wasteful by giving you misleading information on how it works.

[@rkroll17]

@@ThePlantCommander I’m saying even if the value was determined after you would still be wasting shells. If you pay 50 shells for a 50% chance then the game rolls a 25 then you would still have wasted 25 shells.

[@IronMan3582]

If you have an EverDrive, there is a ROM hack that quite literally moves the heart piece to the other side of the room so you can just collect it and skip this whole nightmare and only do it if and only if you want to

[@ThePlantCommander]

why this wasn't how it is in the base game, I will never understand

[@IronMan3582]

@@ThePlantCommander It's truly an awful sidequest, and even moreso after seeing your video on how the lottery works against you. I thought getting all the figurines in Wind Waker was bad because of the limited number of pictures you can take, but this collection is way way way worse

[@ThePlantCommander]

@@IronMan3582 I did it in wind waker once... never again, and that's my second favorite game of all time lmao

[@IronMan3582]

@@ThePlantCommander I remember a work buddy of mine who got the Prima guide and I just saw all the color drop from his face when he realized how many hoops he was going to have to jump through to get everything in Wind Waker, and there were I think two or three you couldn't get at the same time, you had to choose one over the other? I dunno if they ever fixed that in the WiiU port

[@TheBreadPirate]

How on earth did this become worse???

[@drygnfyre]

Because it was at first believed the % displayed was accurate. It's not. The % is basically showing you the threshold for winning or losing, but the game is actually rolling a hidden number. In other words, spending more than a single shell every time is wasting your time.

[@TheBreadPirate]

@@drygnfyre I was being rhetorical. I know why it's worse. I'm just exclaiming my shock at how terrible the system is.

[@AeroDragon2]

This was such a fascinating experiment turned video from you dude

[@ZaffreRevolution]

my dude why have you been holding out on us? i haven't even played minish cap but this video was entertaining and clean as hell! great stuff

[@ThePlantCommander]

Thanks bro! I kinda decided to do this on a whim, it actually didn't even start as a video essay. Most of the footage of this video is actually from an unscripted live recording I took of the same concept, and after I did that I decided to go all out, so I wrote a script, dubbed over it, recorded some B roll and extra footage, and spent the next two weeks editing like crazy, lol. I had a blast making it though, it was super hard to edit and animate everything correctly but I think the end result was worth it. I definitely wanna start doing more video essays. Maybe even... a collab? 👀

[@yonatonheit8397]

I'm not sure i see the problem. If the game selects a random number between 1 and 100, there an 80% chance that its 80 or below. So when you selection 80%, there is in fact an 80% chance that you guess the right threshold. That fact that its a predetermine number doesnt make it not probability because you lack the means of knowing what it is without save states. Think of it like this. If I flip a coin, hide the results and ask you to guess. The coin is in fact before you guess 50% change heads and 50% change tails because from your perspective, it is unknown. The fact the true result is predetermined doesnt change that.

[@ThePlantCommander]

It doesn't, you are correct in that sense. The difference between this and a coin flip is that there aren't just two possible choices, and the fact that the game hides the fact that this is how it actually works actively tricks the player into wasting shells. If that information was clear, it would be obvious that betting more than one shell unless your odds are unreasonably low is extremely wasteful, but hiding that fact from the player encourages them to do so. For people who DON'T know that this is how it works, they'd think that a higher bet is actually worth it, but with a system like this, it isn't.

[@kabzebrowski]

@@ThePlantCommander That's.. not true, just like a coin flip, there are only two possibilities: you're either >= the threshold or < on your guess. You don't know the odds before hand unless you cheat, and the threshold rolls reset for each try, so it's working properly. Using two shells still awards you 2/100 chance of getting a new figurine per try, as there is 1/100 for the game to select 1 as the threshold, and 1/100 for it to select 2 as the threshold, and 98/100 for it to chose another value, so in 50 tries it statistically will roll 1 or 2, and each shell you put there ups your chances of being >= above that random number by 1/100, so there is no rigging nor wasting there. You're arguing that you're wasting shells because the RNG is predetermined on top of abusing save/load states, but the game and that style of RNG wasn't designed around save states.

[@simonO712]

​@@ThePlantCommanderI still don't get how you mean that the game is tricking you into wasting shells. Assuming no RNG manipulation, the optimal strategy (only spending a single shell each time) is _completely_ independent on whether the RNG rolls its number before or after you pull the lever.

[@ThePlantCommander]

@@simonO712 The timing actually doesn't matter, you are correct. It's the threshold concept that makes spending more than one wasteful, but only in hindsight; played as intended, it makes no difference. It essentially means there is no difference between betting 1 and betting 50 if the game happened to roll a 90. The only difference in that scenario is that you just threw away 49 shells for no actual gain. You were CLOSER to the threshold, but you lost anyway, so it didn't matter. In practice, this isn't actually deceptive, and without cheating since it would be impossible to even know that, but even then it's still beneficial to only bet one from just doing the math. 1 shell for 20% or five times the shells the shells for 25% simply is not worth it regardless of the mechanics behind it. More than anything this video was just an observation and breaking down how it works to those who aren't familiar with RNG mechanics than actually complaining about the mechanics themselves. Just a fun little thought experiment.

[@danielalvarez0]

Great video! The concept of replying to another vid takes me back to old youtube video responses. Nice editing, nice voice and nice script! Haha all great, really pro! Just subd

[@ThePlantCommander]

Thank you so much!! Yeah I miss the old school youtube days too lol

[@gneii6456]

This isn’t actually worse, it’s just the roll happens before you bet. For a real world analogy, imagine some plays a game where if you roll over 1, 2, or 3 you win a prize. The normal sequence would be you pick your threshold and then the roll happens to decide if you win. This is more like the roll happens but is covered up before you select your threshold. The odds don’t change, the two events are independent.

[@ThePlantCommander]

That's exactly the problem; the game basically lies to you by showing you a "probability" when that isn't how it works at all. It's entirely possible to walk into the room and get screwed over by the game getting a good roll, making your "bet" almost guaranteed to be wasted, with no way to know if the game screwed you over without cheating. Basically, the game lies to you about how the odds work to hide the fact that it can be (and often is) rigged against you. This encourages players to bet more shells to make them *think* they're getting better chances, but they really aren't; the threshold has already been decided, and you'd be wasting precious resources by trying to figure out what it is. It gives you misleading information to trick the player into betting large amounts on something that, with no way of knowing, could already be a guaranteed win or loss. Do they actual ODDS change? Technically, no, but at that point it doesn't actually matter because the result being predetermined means there's no point to wasting your resources. It actively tricks the player into doing something wasteful... which, to be fair, is accurate to how actual gambling is designed, but nobody's playing a Zelda game to get the casino experience, so that's not exactly a good thing. The 80% strategy still stands, but only when your base probability is so low that it's unreasonable to try to get a new for 1 shell. I'm talking below 10% here. Above that, spending more shells for higher odds (especially on a platform with save states) is a complete waste.

[@Monkeyman249]

@@ThePlantCommander I think all of this only matters if you want to use save states to try to get the figurines. You are right about it generally being wasteful to bet more than a single shell, but it doesn't have anything to do with the thresholds being predetermined. The probabilities the game shows are still correct given that you don't know the threshold, which you won't without save states.

[@javiersaneiro6412]

@@ThePlantCommander Is not because the results are predetermined. Computers can't generate true random numbers, and therefore they have algorithms to generate pseudo-random numbers. The problem is that GBA hardware is so primitive it has a very bad pseudo-random algorithm that creates a new "random" number based on the previous "random" numbers and it do not even change based on the clock. That is why when you rewind the savestate before pulling, it will have the same result. You are actually doing the exact same actions, and given that random numbers are determined based on previous random numbers, it is impossible to repeat the same action and have different results. To have different results, you must rewind before pulling and make Link move around the room and do things, just 1 or 2 seconds will be enough that will force the random algorithm states to generate new numbers, and so when you pull the trigger the result will be different. It is a trick that GBA Pokemon players learned years ago.

[@Londobelle]

I feel it's a matter of perspective. The way to look at it is to treat it as a betting game. You bet every number up to the % that you choose and play that against the target number that the game generated. so if you pick 60%, you bet that any number between 1 and 60 are equal to the target number. In other words, you bet that 60 out of the 100 possible numbers can win, which is 60%. The number you start with also seems to be determined based on the percentage of figurines you have, which can increase if you unlock a new set of figurines and lower when you obtain new ones. So while the chances don't directly decide what figurine you get, you still get either a new one or a duplicate at the chances you bet on. At the end of the day, the order you get new ones in or which specific duplicate you get doesn't really matter. Basically, the only thing that's rigged is the shell pricing.

[@ChaoticTeal]

Thanks for taking the time to dive into this! I *adore* Minish Cap, but would not wish unassisted completion on anyone. That said, I do have a couple of notes from my own NSO playthrough last year. While the added information from save states does make it feel weird, you *do* effectively have the odds the game presents. The difference is when the roll takes place; while you'd expect it to occur when you place your bet, it actually occurs based on the game's existing RNG state, which isn't affected by your bet. Because the game runs on a consistent random seed, though, you can alter the result by taking any other action which advances the RNG. The fastest one I found is rolling-the dust particle effect has a slight random variance. This doesn't make the quest any less of a pain, but at least it's easier to work with!

[@emdal9693]

well in fact the game is not lying to us.. it is just showing the chance that we are above the threshold aka the chance that we get a new thing in fact that only change when you have any information about the threshold which we don't unless using cheat... so no it doesn't really change anything as long as you're not using rewind

[@dasallmaechtigeJ]

That’s not the worst. The worst is planting trees in OoS/OoA randomly hoping for the heart piece.

[@ThePlantCommander]

That's RANDOM?

[@alecdashark]

Nobody tell him about sploosh kaboom.

[@ThePlantCommander]

I should test that actually, I don't know when the board is generated. At least that requires strategy and player input though. Still boils down to dumb luck but at least it's not a weighted coin flip :P

[@alecdashark]

@@ThePlantCommander TRU but it's really interesting! Because the community itself has an entire tool just to decipher the boards you get. Like it goes THAT deep. You should look into it it's super interesting!

[@Avigorus]

The instant I saw the repeats on the rewinds my first thought was that it was set when I entered the room, with the worst case scenario being a need for a full game reset to reset the figure & threshold. Now, I wasn't sure if exiting and reentering the room would reset it, but yeah "rewind to before entering the room" is where my instincts went lol

[@unknown000001000]

When I did this on the GBA, I dedicated an entire save file to that point so that on future playthroughs I would never have to endure it ever again. X(

[@psoslash]

probability is still accurate, it just does the rng calculation beforehand, if we don't take cheating in consideration it has no difference wether is decided earlier or after the bet

[@OffbeatDrill]

I don't think the game is lying to you. There's a 1% chance that the game chooses the threshold number as 100, so there's a 1% chance that you don't get a new figurine if you bet enough shells to have the game tell you that you have a 99% chance of a new figurine. The threshold number is just a way to program probability in the game. Am I missing something?

[@SuperJM9]

I bet enough shells for a 100% chance every time, refilling by using picolyte to farm mysterious shells and rupees to buy picolyte. The great spin attack in Western Hyrule Field helped hugely with this. It was tedious but at least I knew I’d get a new figurine every time.

[@mage1over137]

Without save states this is functionally the same.

[@AngryEyesJustInCase]

idk, i dont think it "lies" to the player in a way that matters. Just because the money was put into the briefcases *before* the game started, doesnt mean Deal Or No Deal isn't a game of chance (unless you're clairvoyant, but then what are you complaining about?) The randomness behaves as it claims to, with a seed that changes based on player inputs, in a system that's only known to speedrunners and enthusiasts. To everyone who isn't savvy to this, it's functionally random, just like pokemon catch rates (where the randomness seed is increments once per second).

[@ThePlantCommander]

Pokémon catch rates are some of the most obscenely and unnecessarily complicated things I have ever seen in a video game, so this is nothing by comparison lol. It's kinda scummy in the sense that it can lead you to wasting resources (although in all fairness, that's still true even without knowing the mechanics at play here) but it's not really "lying." I just had some fun with making this video since the KZbin algorithm loves claims like that :P

[@The-Pizza-Zone]

I recently replayed this game on GBA 100%, and this quest, as well as being given the mirror shield post-game, knocked the entire thing down a point or two for me.

[@link_64n]

Wow!!! This is insane and so helpful, thank you!!!

[@tjeerdtrekkie1030]

You know I'm starting to think that Nintendo made this entire heart piece shell quest to teach Children about the dangers of a gambling addiction

[@karrengoodman7165]

Man I had too much free time as a kid

[@Himmelschenk]

This heart piece in question was one of the last things i needed to do i had nothing to lose.

[@Florete_FE]

If anything I think this actually makes it better. It's manipulable. A lot of games, especially older games, use RNG systems like this (I thought of Fire Emblem very quickly) since computers can't be truly "random." There's always an RNG string in code somewhere. And as far as the player is concerned, this *is* still random. It ultimately works the same as any other % chance for one of two outcomes. Any event that has a % chance of happening is either going to happen or not, and the chance that the game rolls 100 is still 1%.

[@Keanine]

For the first half of the video you're explaining that if I select 60%, I have a 60% chance of being over the threshold? I don't see the issue there. If I pick the lowest possible percent, I have the lowest possible chance of being above the randomly generated number which is accurate to the rules they provided. It just so happens that they're kind enough to give us a minimum bet I do see what you're saying with 20% being the minimum bet and the odds being in the players favour regardless, and since that first shell is worth 20% while any subsequent ones are worth 1% there is a decent chance you'll win by picking 1 shell every time without using as many resources overall but with a longer time commitment as you'll lose more games than you'll win Overall you raised some good points, but I really don't see how the game has been lying to us, it follows its own rules exactly. Regardless, it still sucks for a single heart piece lmao

[@ThePlantCommander]

More or less, yes, that's the gist of it. As many other comments have pointed out, it isn't entirely accurate to say the game is being lying, but it IS deceitful in the sense that a losing threshold is predetermined BEFORE selecting your odds. In practice, this doesn't actually matter since the player is not supposed to be able to know that, but it does mean that a losing bet is more wasteful in hindsight. If you bet 50 shells and the game decided you needed to bet 80, you just wasted 49 shells for no reason because you were guaranteed to lose that bet before you even made it. So yeah, the actual numbers aren't WRONG, but they are depicted in a way that's designed to trick you into being wasteful.

[@alexanderspencer4194]

That editing timeline would have made my computer start sparking and emitting steam, lol. And here I had no idea it got worse than the “Hit Orca 500 times” heart piece in ww

[@ThePlantCommander]

At least the Orca one is relatively easy, it just takes a while. This one is nearly an hour at minimum, but that's with absolutely perfect luck. It's more likely gonna end up taking nearly 10 hours xD

[@LitchiBorrower]

The end result is still accurate, unless using save states. You still have the indicated % chance to have picked a number bigger than the threshold. The best strategy of using a single shell would still be true no matter what; You show using 66 shells for a single 85% pull in the vid, for that same cost you could get *66 pulls* at 20%, which have much, *much* higher odds of giving you a new figurine. You don't need deep experimentation using savestates to realize that using more shells is a time-saving feature and not an optimal strategy, just basic probability.

[@Knightmessenger]

Do you think the lottery in Smash Melee works similarly?

[@ThePlantCommander]

Most likely. This is a very common way for games to program how RNG works. That being said, Melee's lottery takes place entirely within the game's menu UI (which has a lot less going on mechanically than an actual room you can run around in), so it's probably much less manipulable than this. That's just a guess though.

[@VGamingJunkieVT]

It's not as bad if you think of it as a model viewer with the piece of heart and music player as bonuses. I happen to like model viewers, especially in the Metroid Prime games.

[@sagacious03]

In theory, for each number there might be a series of guesses that can tell you what number the game picked in as few guesses as possible. Doing that might be able to lower the time spent exiting & re-entering the room for rerolls. Anyway, neat analysis video! Thanks for uploading!

[@果従莞友]

I don't think it matters how "rigged" it is. The game gifts us with more shells than we know what to do with anyhow. I have all figurines and 999 shells. I never feared spending them all.

[@ThePlantCommander]

You are the first person I've ever heard claim this, you must have insanely good luck or just be good at RNG manipulation. Spending one at a time will lead to that with how much the game gives you, but... that takes hours, so either way this sidequest is absurdly tedious lol

[@果従莞友]

@@ThePlantCommander Idk I didn't do it all in one go. I would go in, spend all my shells, leave and play more out in Hyrule, then come back and spend em, etc and so on. You don't have to play it for hours on end if you don't want to. Obviously can leave the mini game any time you want.

[@guanchampo]

The other thing that wasn't mentioned in this or MrDrBoi's video is that he'll buy duplicate figurines back from you for 5 rupees. In other words, if you only bet 1 shell and get a duplicate, you're essentially selling him a shell for 5 rupees. Well, you can also buy 30 shells for 200 rupees from the Hyrule Town shop, i.e. 150 shells for 1000 rupees. Meaning for every 200 attempts betting 1 shell, you can get 150 shells back, so a net cost of 50 shells for 200 attempts. This turns your 999 shells into almost 4000 attempts if you only bet 1 shell each time and properly cycle your rupees into shells. This is what I just did when playing through - I just watched some youtube videos or streamed tv show while mashing betting 1 shell each time, until I got to like 10 figurines left and bought them at 100%

[@YokoX23]

Nintendo odds are either 50% or 0% if it's not 100% See Pokemon for the most obvious example.

[@xenxander]

i like when people reveal that there is 'no reason to play' a game. The Yugiho card game online.. the AI draws the best card from its deck it needs, on its turn. So the cards 'drawn' in its hand mean nothing. It 100% cheats, having 'deck crawl' or whatever you call it,100% of its turns. Therefore I realized there is no reason to play because if the right card exists in the computer's deck to counter you, it will play it... guaranteed. and if it doesn't, means you got lucky and chose those cards before the match.

[@ThriftyFangirl]

What you're describing doesn't change anything though. When you spend more shells to increase your odds, you're doing so with no knowledge of where that random number landed. It doesn't actually matter when the number was generated as long as you've made your choice without any additional information. You choose how big to make your "basket" and the rng decides where its going to land. Now the actual reason it's a scam to increase your odds is because unless you're down to 1% for 1 shell, the amount that you increase your odds is not proportional, and you would net more wins over time by spending only one shell. That's because the first shell is worth however much, say 20%. If you spend an additional shell on the same roll, it's worth 1%, but if you save it to spend on a second roll, then the second shell is worth another independent 20% or 19% if you won previously

[@LuckyNumber39]

TLDR, it's not betting on dice, it's the card game War

[@benny4162]

I learned about the repeats when I was collecting all the figurines. I had two or three figurines left to go and I was savescumming like mad to get them. At that point, if I didn't bet enough shells to have 100% chance I would just get the same Darknut figure over and over and over.

[@cookshowtrevor]

It’s like it’s simulated probability. I wonder if true probability was a bit too taxing for a GBA game so they used a fake percentage as a threshold marker to make it a simple yes or no equation

[@ThePlantCommander]

It actually has nothing to do with GBA limitations, this is a very common way for even modern games to handle RNG. A lot of games either decide an arbitrary number on the spot, or it's pulled from the game's current seed, which isn't entirely random but it's remarkably specific enough to make it *effectively* random to the average player. This is what makes things like wrong warp and RNG manipulation possible; if you know how the game's seed can be changed, you can force some things to happen.

[@shadeknight6537]

The best strategy for getting that Heart Piece is to play the game on emulator, use a walk through walls cheat, and just skip the whole gambling nonsense.

[@Eddies_Bra-att-ha-grejer]

I remember finishing this quest as a kid and I don't remember it being particularly tedious. I'm pretty sure I collected the shells using a combination of two of the picolytes.

[@Zolerian]

So this proves that it is even more worth to just use 1 shell each time, because it doesn't really matter if the odds are 1% or 99%, as it is the same as long as they aren't 100%.

[@JimiGosu]

So we are essentially rolling a d100 against the game EVERY TIME WE TRY TO GET A FIGURINE regardless of what the percentage says our chances are.

[@ThePlantCommander]

Sort of, but not really. Behind the scenes that's what it looks like, but in practice, the percentage is technically accurate. It's only inaccurate if you're abusing save states or rewind features. Think of it this way: the game shows you 20%, and you bet one shell. There IS a 20 out of 100 chance that the game will roll lower than you, so that percentage is true. However, let's say you somehow are aware of the fact that the game had actually rolled a 50. Everything between your starting bet of 20 and 50 are a loss, so your odds were actually 50/50. BUT, this only applies if you are aware that the game already rolled. Imagine someone flips a coin and tells you to pick heads or tails. The coin already landed on tails, meaning you have a zero percent chance to be correct by picking heads. But, since you don't know that yet, from your perspective this is still a 50/50 guess. If playing the game as intended, the odds are accurate to what it shows you, but that falls apart if you look behind the curtain. The strategy remains largely the same either way; betting only one is optimal both because of how the mechanic works and because of the math. Different thought process, same conclusion, so in reality none of this actually matters.

[@id-f8679]

I am doing this rn and the best way to do this is using one shell and use a save state and do some random things in the room like swinging your sword until you get a different figurine. Each sword swing changes the figurine you will get.

[@SporeMystify]

Okay, this is just normal psudeorandom numbers. With the same savestate, you generate the same number. It doesn't even have to have already chosen it, a pseuorandom generator is more like picking from a deck of cards - you don't know what the next sequence is, so every draw is random, but the order was chosen during shuffling, not drawing. So when you savestate to repeat the same chance, yeah, the random part is effectively fixed, and you are changing the threshold it's checked against. But that's still exactly equivalent to setting the threshold and the. Getting a random number to compare against if you aren't replaying that event. It's like if somebody flips a coin, and you can call heads or tails, but afterwards you rewind time and watch that same coin flup over and over

[@vulpes6144]

These and the fact that some Kinstones can only be matched at a certain point of the game and then NEVER again is pretty rigged 😂

[@AnonymusAxolotl]

I did this on NSO last year and figured out that you reroll your rn every time you do a sword swing, so I would bet one shell every time and just rewind and do a different amount of sword slashes every time until I rolled a new figurine. At least for numbers around 20% it helped but it was still tedious. Just a little less so

[@skottlee8959]

I guess this is the next poorly understood zelda trivia to be memed to death until the end of time

[@PixelOverload]

I assume it also rolls a new threshold whenever you pull, not just when you leave and enter, else once you enter and get a number below 20% you can simply keep pulling with the 1 shell for indefinite new figs, no? Unless the game boots you from the room after a pull automatically, I don't recall 🤷

[@natanmaia3575]

If you forget save states and rewinds, the odds are actually correct: the order between rolling a number from 1-100 and choosing a difficulty class from 0% to 100% to match doesn't matter, you can test that by rolling d100 dice. The problem is that the dice has _already been rolled_ so save states don't change this; just like GBA fire emblem has your numbers already rolled in a giant list and picks one by one, so if you rolled a 90, if none of your weapons hit 90%, none of them will hit regardless of rewinds. NOW, I haven't finished the video, but if the game rolls one random number when needed after you get your prize, you can do rewind rerolls by rewinding 2 pulls instead of one. Similarly, if the game keeps a list of X numbers, you can change a roll by rewinding X+1 rolls backwards and changing the RN (by pressing buttons differently or waiting different amounts of time).

[@RPG_Hacker]

I know some people have already pointed this out, but as a programmer, I still wanted to share my perspective on this. What you observed here was basically how any randomness in computing works (I'll ignore some exceptions here for simplicity). Computers aren't really able to generate random numbers. They're only able to generate pseudo-random numbers, which really just means they're pulling numbers from a pre-determined sequence of numbers. Simply put: They're using math to calculate a number that "feels" random. Which number from the sequence they're pulling depends soley on the current state of the random number generator. Usually, the state is just a single number (called a seed), and it's initialized once at application start (usually using the current time, but since the GBA can't measure time natively, the game likely uses a different method). Now what you're essentially doing by using save states is that you're rigging the test. Save states function by taking a copy of the game's entire memory, and then restoring that memory whenever you load the save state. Since memory is also what contains any seeds used by RNGs, you're essentially using the same seed for every single attempt. This test case doesn't quite replicate how an average player would engage with the feature while not using savestates (aka how players on the original GBA would've played the game). With no knowledge about the RNG seed and no way to roll back state, the percentage numbers are (essentially) representative of what results a player will actually get. What's worth noting in this regard is that not every game uses RNGs the same way. This game, as you've determined with your experiments, seems to pull the random number for the lottery every time you enter the room. Other games might do it differently and pull a random number every time you interact with an object. Also other games might share a single RNG between many functions in the game. This could lead to the situation where in those games, you could get seemingly different results despite using save states, because something else advanced the RNG in the meantime (but really, under the hood, the game is still just pulling numbers from a pre-determined sequence). This was actually the case in Final Fantasy XII on the PS2. The game had a very strong weapon that could only spawn in chests in a single room and with a very low chance of 1 in 1000. However, the game only pulled the number for the contents of the chest upon interacting with it, and any combat in the game used the same RNG sequence as chests. With some tables, it was actually possible to find out the current seed of the RNG, and then manipulate it using attacks and magic to end up with exactly a seed so that the next time you'd open the chest, you'd always get the weapon.

[@ThePlantCommander]

I really wish that FF12 example was a more common thing. I know a lot of games have stuff like that, but I'm just surprised I don't see it *more* with how crazy speedrunners get with stuff like this. Certainly would make getting the Sword of Kings way less of a pain in EarthBound. I *think* that pulls from the seed the moment you enter a battle, not during the battle itself, so I don't know if much could be done in that case.

[@RPG_Hacker]

@@ThePlantCommander For speedrunning purposes, that'd definitely be practical, yeah. I guess most of the time, programmers don't really think about their game that way when implementing RNgs. They probably don't consider emulation a valid use case, and don't consider speedrunning a priority. In some cases, programmers even intentionally try to make RNGs unexploitable. Newer Pokemon games seem to do that. From what I understand, they use a secure RNG from the system to generate what Pokemon appear next. It's an RNG that is not exploitable by simply observing its generated numbers (though because of some programmer oversights in the games, there's still usually one or two areas where RNGs can be exploited). Aiming for unexploitable RNGs kinda makes sense when considering what purpose randomness usually serves in games. It would somewhat defeat the purprose to intentionally make one exploitable just for the sake of speedrunning - but of course, since randomness is unpredictable, it always has a chance of making a mechanic more tedious than it was intended to be, like here in the Minish Cap. It's why I generally prefer speedruns that are more skill-based than luck-based.

[@Pintobeans74]

Sometimes game developers add patches so that you can’t “cheat” by using an “unsave” to farm. Maybe try accepting the gift and using a new shell to see what the odd are like!

[@tadesubaru1383]

I once betted with a 95% chance of getting a new figurine and still got a repeat. I almost screamt

[@firakagegetsu1101]

You are correct in everything except the abuse of the save State part because it depends upon how far you go back as to whether it chooses a different one for example if you choose to go back to before you talk to the guy in order to pull your stuff and then talk to him that's when it'll change what figurines it pulls but if you always only go back to after you've talked to him then you are correct I just thought you'd find that interesting

[@ThePlantCommander]

From what I can gather, it's talking to the guy when the RNG is decided. Actions before that (including doing things in the room like rolling or using an item) will re-roll the RNG, but as far as I can tell it's locked once you interact with the shopkeeper.

[@shawnbay2211]

makes me glad i abandoned trying 2 get all heart pieces !

[@CatsLilaSalem]

As far as i know lots of games use that type of RNG, and only some actions, like leaving the home in this case changes the RNG

[@Beannie7777]

Wait, so does that mean it could be subject to rng manipulation?

[@darkychao]

I mean, whether the rng is seeded before or after you pull the lever dosn't really matter, since, unless you're cheating it's just a black box. also, statistically even if the rng was decided after the lever was pulled you still should only be doing base odds rolls anyway. you'd be wasting shells on raising the odds by 1 percent each rather than using the 1 shell for an entire exta roll with whatever base odds. going just by a raw shells per figurine standpoint it only makes sense to start using extra shells when you're base odds are at 2% or lower, at which point you should always pump it to 100% every time.

[@VilasNil]

This makes no sense. Using shells DOES give you a certain probability to pull a new figurine. It doesn't matter if the "machine" knows that you will fail the probability check or not beforehand. What is changing with every time-rewinding shenanigan is your knowledge of something that wasn't intended to be known, it's like you've gone back in time and you know of the events of the future, if it's outside of the scope of your capabilities you won't be able to change it (and why would you?). If the game is using a pseudo-random number generator, it could even be that the number is being picked AFTER selecting the number of shells (which in your eyes seems to make it more fair), but due to how pseudo-rng works, as nothing else is consuming numbers from the sequence, the sequence is predetermined and the same number will be picked every time when going back in time (and consequently going back in the sequence or more accurately returning the pseudo-rng to the state it was at that point in time, think of it like when in Minecraft you give a Seed to the world, the results are always the same, because the same starting point for this sequence is being picked). Then, we can go to the numbers part of the equation. Given that we can assume that the pseudo-RNG is almost random to a regular player without time-bending powers, it's a good implementation. Each pull has a 1/100 chance (or close to, binary is weird sometimes) to give each possible threshold, meaning that when you pick a percentage, you are placing a bet that the threshold is BELOW your selected number, which does indeed have its given chance. If picking 75%, there would be 75 thresholds that would give you a new figurine and 25 thresholds that wouldn't, so indeed you have a 75% chance to get a new figurine. Sure, the threshold could've been at 5 and you'd have wasted a lot of shells, but without that knowledge from the future it could just as probably been a 75 and have just barely gotten the new figurine, you just don't know where the threshold is at every pull and with every shell you are increasing the amount of valid numbers, each having their 1/100 chance of being pulled.

[@VilasNil]

What I will give you though is that the shell number is a game of diminishing returns, and beting just 1 shell gives you the better value. And to save time, using shells to get to 100% for the last few would certainly save time, as there is no way to have a 100% chance of getting non-repeats by just betting one shell, you could have a streak of bad luck and eternally pull in the range of 21-100.