When I was a small child playing Soul Silver on release I became so addicted to Voltorb Flip that I never actually beat the game, I just stayed in the Game Corner all day losing to 50/50 guesses. I knew the game was BS but I loved it.
Stand proud king, your girlfriend stucked with you BECAUSE you beat fire red blind and deaf on offical hardware, not despite it. I'd tell it to everyone I meet tbh. Also matlab is truly rated M.
Fun fact: The EU was the one who pitched a fit about gambling in pokemon. The yanks get off scott-free on this one (for a change). Very nice video on a minigame I completely forgot about.
thank you for delivering justice to the people. i love the way you stitch in bits about your life in these videos, they are much more compelling for it. looking forward to the next one.
No joke, not only a great pokemon video, but a great video in general english is not my first language so i dont know the exact words, i just wanted to say that the final product turned out to be really great!
I've recently become monomaniacally obsessed with an obscure (albeit less whimsical) project myself to the point that it's started to invade my dreams, so this was very cathartic to watch. Now you've got me wondering whether anyone's tried to patch Voltorb Flip to make all boards solvable... also damn those are some big honkin' caterpillars
@@TongyBacala Nice cover story to have ready before the administrators see this, Mr Bacala. FWIW I’d be happy to run whatever (sigh) Matlab crowd compute project you want on my sometimes idle 9950x/192GB workstation if you put out a short video asking for volunteers.
Another great video. If you need some extra compute, I might be able to spin up an extra node or two on my local kubernetes cluster. No guarantees, but I should have some extra unused processing power… - a fellow New Yorker
I feel like something that generally helps is to ignore fields you dont need to care about. For example when a row has 1 voltorb and the sum of the other tiles is 4. That row only has ones and voltorbs. Ones dont help you so if you dont know where the voltorb is you can simoly ignore tte whole row. Doesnt help with 50 50 guesses though
This about lines up with what I would have expected. I played through SoulSilver earlier this year and initally thought Voltorb flip was really fun, but eventually found myself running into these minesweeper-like guessing games all the time, which really took the fun out of it. It seems to happen more often at the higher difficulty levels, too. Question: does the game actually generate of of the theoretically possible puzzles that you checked for solvability? I remember finding patterns in how the game generated its puzzles: something along the lines of two tiles with 2s in level 1, two 2s and a 3 or two 3s in level 2, and so on. I'd be curious to see what proportion of puzzles at each level are (un)solvable.
In HGSS proper, there are several board configurations that the game can select. 10 for each level from 1 through 8, for 80 in total. These configurations consist of five numbers: the amounts of voltorbs, twos, threes, how many "free" multipliers (i.e. without a voltorb in the row/column) can be in a single row/column, and how many free multipliers can be on the board total. I made another comment earlier, but I think it got auto-nuked because I made the mistake of linking to a solver tool I made.
neat! I figured there was something like that going on, but I wanted to tackle the "theoretical" game rather than what's really in HGSS. Thanks for the comment :)
A similar question that I have had about Voltorb flip: what is the probability of winning while playing optimally. That is, for each time you have to guess, simulate all game possible game boards and choose the tiles least likely to make you lose. Sequel?! Great video though, I dig your style
I think if you really wanted to go deep into what it means to play optimally, one would have to use a technique like information theory to determine which guesses yield the most information relative to their risk. The @3blue1brown Wordle video comes to mind
You should be able to reduce the column search space using two invariants. Sum(Voltorbs in Rows) = Sum(Voltorbs in Cols), and Sum(Cards in Rows) = Sum(Cards in Cols). No need to check every single combination of columns for every set of rows, only those where the invariants hold.
did you account for the weighting though? as in: since you are eliminating permutations, you *should* then count them as many times as there are permutations for them. otherwise the percentage solvable might be off
@@theaureliasys6362 yup, i did! generally didn't get into all the details but you can calculate the number of combos that lead to a given unpermuted row combo pretty easily and weigh by that
I think your example may be wrong, there should be 2 voltorb in column 4, the number of voltorb in the rows and columns don’t match. But what do I know, I only grinded for dratini not ice beam
you are right. I mentioned this in a different comment thread - editing error on my end. I actually just updated the description to say as much, but I think everyone gets the idea :D
@@AJMansfield1 yeah i know. there are a lot of discussions about this in the comments already. it's more a pen-and-paper theory thing. hope you enjoyed the video anyway
I think you severely miscounted. Even without knowing the game specifics, I highly doubt all boards are equally likely. First, the game probably generates a limited number bombs and then fills the rest. Second, you counted the number of row/column configurations. Every solvable round has one configuration, however every non-solvable has 2+. By that, non-solvable row/col combinations are at least 2x as likely as solvable ones. On the other hand, only from the rows/columns it's not always possible to tell whether a board is solvable. It could be that there is a safe square and the game becomes solvable if it is a 2 but unsolvable if it is a 1.
@@patrickwienhoft7987 i don't go into all of the details in the video but i account for how likely each score combo is and i do account for this "safe 1 vs 2" thing. it's a good point, i just didn't want to crush the audience with details the only thing i don't account for is how likely the Pokemon game is to generate a board. it has some sort of seed, of course, and may even select from a fixed set of puzzles. the treatment here is more theoretical, as if it were a pen-and-paper game
I always looked at this minigame as a hybrid of minesweeper and picross, hah Which the stock minesweeper that came with Windows (and probably the one on Google) also funnily have a tendency to succumb to RNG vibe checks so that's fun All the while the one picross game I grew up with, Picross DS, has a puzzle editor that comes with a solvability checker that stops you from saving a puzzle if there's an ambiguous set of clues, it's pretty damn robust too iirc
I appreciate the work but a lot of it seems… unnecessary? You can look up the total number of 2s, 3s, and voltorbs based on the number you’re playing. Then, based on that, you can calculate the odds a square of death exists.
Alternative title: New York man proves that a casino game involves chance
(Great video as always)
@@AnUsername bahahahaha i love this. thanks for the consistent viewership
When I was a small child playing Soul Silver on release I became so addicted to Voltorb Flip that I never actually beat the game, I just stayed in the Game Corner all day losing to 50/50 guesses. I knew the game was BS but I loved it.
@@Duker_Dude some say he still doesn't have enough coins for an ice beam tm
Stand proud king, your girlfriend stucked with you BECAUSE you beat fire red blind and deaf on offical hardware, not despite it.
I'd tell it to everyone I meet tbh.
Also matlab is truly rated M.
Fun fact: The EU was the one who pitched a fit about gambling in pokemon. The yanks get off scott-free on this one (for a change).
Very nice video on a minigame I completely forgot about.
thank you for delivering justice to the people. i love the way you stitch in bits about your life in these videos, they are much more compelling for it. looking forward to the next one.
No joke, not only a great pokemon video, but a great video in general
english is not my first language so i dont know the exact words, i just wanted to say that the final product turned out to be really great!
incredible video, I certainly did not expect to stumble upon absolute cinema when clicking on this
I've recently become monomaniacally obsessed with an obscure (albeit less whimsical) project myself to the point that it's started to invade my dreams, so this was very cathartic to watch. Now you've got me wondering whether anyone's tried to patch Voltorb Flip to make all boards solvable...
also damn those are some big honkin' caterpillars
welcome to my hell ;)
This is a weird way to confess to abusing access to the high-performance compute at university.
@@c6m in one sense, yes
in another sense, it was largely a joke. i did almost everything on my home computer lol
@@TongyBacala Nice cover story to have ready before the administrators see this, Mr Bacala.
FWIW I’d be happy to run whatever (sigh) Matlab crowd compute project you want on my sometimes idle 9950x/192GB workstation if you put out a short video asking for volunteers.
@c6m Bacala v. Academia, 2024
excellent music throughout the video, and I loved how you described the discovery process. really good!
Solid upload. Can't wait for the next one
An 18 minute tongy video is exactly what i ordered for my lunch break
new yorkers can do anything
@@zapbolt8 except consistently beat puzzles for children
STARS AND BARS MENTIONED
@@foiledthrice tfw order doesn't matter 😜
Another great video. If you need some extra compute, I might be able to spin up an extra node or two on my local kubernetes cluster. No guarantees, but I should have some extra unused processing power…
- a fellow New Yorker
I feel like something that generally helps is to ignore fields you dont need to care about. For example when a row has 1 voltorb and the sum of the other tiles is 4. That row only has ones and voltorbs. Ones dont help you so if you dont know where the voltorb is you can simoly ignore tte whole row.
Doesnt help with 50 50 guesses though
He’s back with some quality content!
s/he knows the vid is good within 30 seconds, a true fan
This about lines up with what I would have expected. I played through SoulSilver earlier this year and initally thought Voltorb flip was really fun, but eventually found myself running into these minesweeper-like guessing games all the time, which really took the fun out of it. It seems to happen more often at the higher difficulty levels, too.
Question: does the game actually generate of of the theoretically possible puzzles that you checked for solvability? I remember finding patterns in how the game generated its puzzles: something along the lines of two tiles with 2s in level 1, two 2s and a 3 or two 3s in level 2, and so on. I'd be curious to see what proportion of puzzles at each level are (un)solvable.
In HGSS proper, there are several board configurations that the game can select. 10 for each level from 1 through 8, for 80 in total. These configurations consist of five numbers: the amounts of voltorbs, twos, threes, how many "free" multipliers (i.e. without a voltorb in the row/column) can be in a single row/column, and how many free multipliers can be on the board total. I made another comment earlier, but I think it got auto-nuked because I made the mistake of linking to a solver tool I made.
neat! I figured there was something like that going on, but I wanted to tackle the "theoretical" game rather than what's really in HGSS. Thanks for the comment :)
Makes sense. The aforementioned RNG solver tool can reach the cap in about 10 minutes, if you ever feel like skipping the grind next time :-)
A similar question that I have had about Voltorb flip: what is the probability of winning while playing optimally. That is, for each time you have to guess, simulate all game possible game boards and choose the tiles least likely to make you lose. Sequel?! Great video though, I dig your style
I think if you really wanted to go deep into what it means to play optimally, one would have to use a technique like information theory to determine which guesses yield the most information relative to their risk. The @3blue1brown Wordle video comes to mind
I grinded for a dratini once I discovered the game, seems about right lol
You should be able to reduce the column search space using two invariants. Sum(Voltorbs in Rows) = Sum(Voltorbs in Cols), and Sum(Cards in Rows) = Sum(Cards in Cols). No need to check every single combination of columns for every set of rows, only those where the invariants hold.
@@Chockrit yup, i did that! just didn't get into those details
@TongyBacala ah gotcha, surprised that the search space wasn't further reduceable
@@Chockrit it might be if you are smarter than i am go for it ;)
Thank you for this video. It was very enjoyable :3
damn bait! when I saw "core dump", I thought you was gonna say the lab computer took fire because of some sub-optimal code. 😂
@@minirop the climax is i get fired and they take my PhD away
MATLAB MENTIONED
did you account for the weighting though?
as in: since you are eliminating permutations, you *should* then count them as many times as there are permutations for them.
otherwise the percentage solvable might be off
@@theaureliasys6362 yup, i did!
generally didn't get into all the details but you can calculate the number of combos that lead to a given unpermuted row combo pretty easily and weigh by that
I think your example may be wrong, there should be 2 voltorb in column 4, the number of voltorb in the rows and columns don’t match.
But what do I know, I only grinded for dratini not ice beam
you are right. I mentioned this in a different comment thread - editing error on my end. I actually just updated the description to say as much, but I think everyone gets the idea :D
I expected many thing when I clicked this video. Daddy Yankee was not one of them.
@@ivan1793 DURO!
Two posts in one week? Early Christmas?
@@ptolemy6616 the stars just aligned 🫥
W
It doesn't matter what proportion of theoretical voltorb flip puzzles are solvable -- the actual game doesn't sample from that distribution uniformly.
@@AJMansfield1 yeah i know. there are a lot of discussions about this in the comments already. it's more a pen-and-paper theory thing. hope you enjoyed the video anyway
@@TongyBacala Oh, definitely did! Hardcore math + storytelling is such a good combo!
I think you severely miscounted. Even without knowing the game specifics, I highly doubt all boards are equally likely.
First, the game probably generates a limited number bombs and then fills the rest.
Second, you counted the number of row/column configurations. Every solvable round has one configuration, however every non-solvable has 2+. By that, non-solvable row/col combinations are at least 2x as likely as solvable ones.
On the other hand, only from the rows/columns it's not always possible to tell whether a board is solvable. It could be that there is a safe square and the game becomes solvable if it is a 2 but unsolvable if it is a 1.
@@patrickwienhoft7987 i don't go into all of the details in the video but i account for how likely each score combo is and i do account for this "safe 1 vs 2" thing. it's a good point, i just didn't want to crush the audience with details
the only thing i don't account for is how likely the Pokemon game is to generate a board. it has some sort of seed, of course, and may even select from a fixed set of puzzles. the treatment here is more theoretical, as if it were a pen-and-paper game
I don't understand the puzzle at 6:36. Isn't the fourth column from the left only supposed to have one Voltorb?
yeah that was an editing mistake on my end. number of voltorbs should be read 2
Heartgold > Soulsilver
on this channel we respect all opinions
except for this one
I always looked at this minigame as a hybrid of minesweeper and picross, hah
Which the stock minesweeper that came with Windows (and probably the one on Google) also funnily have a tendency to succumb to RNG vibe checks so that's fun
All the while the one picross game I grew up with, Picross DS, has a puzzle editor that comes with a solvability checker that stops you from saving a puzzle if there's an ambiguous set of clues, it's pretty damn robust too iirc
Tomgy
savestates are a wonderful thing
official hardware only !
I appreciate the work but a lot of it seems… unnecessary? You can look up the total number of 2s, 3s, and voltorbs based on the number you’re playing. Then, based on that, you can calculate the odds a square of death exists.
@@shadowcage72 yeah but to do a statistical analysis you need to do that for every puzzle
36
i messed up and said 32 in the voice recording and had to use the robogirl voice, i just had to
Lol just reverse engineer the seed, then the puzzles are skill again, you have all those computers.
i mean sure, but that isn't the point. it's about the theoretical game, as if it were on pen and paper or in front of a child playing a ds