What are other anime you'd like to see included? I want to do an analysis for a future video that requires some more anime, so what would you like to see? More diverse the better
I mean how long is long? some of my personal favorite run about 5 or 6 seasons. MHA (My Hero Academia) has 7 with an 8th coming soon, and it's infamous for having an impressively large cast of side characters (who's utilization is up to debate). There's also Bungou Stray Dogs which has 5 seasons and it's very faction focused, in fact I think the whole series (animated) only has about 50 characters because it's all a bunch of tight knit groups rather than monster of the week which I think would be rather interesting to see how it diverts from the massive casts of the other shows. I know FMA:B (Fullmetal Alchemist Brotherhood) is another longer series that's very popular however I've not watched it so I can't comment on it Another one that I think would be interesting because of its uniqueness is Haikyuu!, as a sports anime it's cast consists of teams, so rarely do you ever just get one character popping up, the whole team does. I'm curious as to how that would appear on the graph! Those are just a couple suggestions!
@@notdavidsecondary9958 That's another great comparison to make yeah! How the two versions differ in cast makeup God I love being a stats nerd NUMBERS ARE FUN lmao
I think it's weird that you didn't mention Jojo's obvious difference to other anime, the Parts. Jojo doesn't have any characters that stick around for the entire series, and it barely has any characters that show up in more than one Part. Jojo is basically 9 different series with some crossovers, and i think if you make graphs for each individual Part, they'd more closely align with other anime series.
Yeah I agree I should have mentioned that explicitly. I talk a lot abut parts in the main channel video and here I just kind of alluded to it but didn't say it explicitly (thought it is pretty evident from the Jojo raster).
@@truthcrusader360 Series is usually how you would refer to a manga, which they would have to be since they said 9 and JoJo only has 6 parts adapted to anime. You are right that the whole thing is also a series but there isn't really a better word for each part (aside from part) If it was an american comic they could use the term "run" but that isn't really a thing in manga.
DIO in some way plays a major role in all of the first 6 parts and is referenced in 7-8 with his alternate universe counterpart being a major player in part 7 with 8 being indirectly affected. It’s not multiple separate shows it’s 2 deeply interconnected worlds that make little sense if you don’t have the knowledge of the previous parts to build on to the next ones. it’s more like really long movies in the same series, kind of like the original StarWars trilogy, or Heisei era of Godzilla movies, all of them are decent stories on their own but only truly shine when they’re in the context of one another in a linear viewing experience from each part to the next.
@@seviiaal Yeah!! They are briefly shown (in present time) during the Egghead arc. I forgot which chapter, and I dont wanna be too specific n spoil anything.
@@Attack_Pillow1. They made it pretty clear during baratie he was kind of a freak that could survive it and 2. no not really. Oda doesnt really use death for realism and imo drops the ball on a handful of occasions with it. but its also just not his thing to kill off ppl
Those two show are complete opposite with each others. One Piece focus on continuous saga storytelling meanwhile Jojo had each distinct story. One Piece emphasize world building, Jojo emphasize chronological lineages.
Exactly. This is why I tend to say skipping One Piece arcs is wayyyyyy worse than skipping Jojo parts. Like, let's say you jump from DiU to Stone Ocean, if you skip Golden Wind you wouldn't find yourself confused by anything at all since SO has absolutely 0 to do with GW (hell the whole concept of Requiem doesn't even return either even). I'm not recommending people to be part skippers, but what I'm gonna say is that if anyone happens or so chooses to start JoJo with certain parts other than Phantom Blood, they won't be having that hard of a time making sense of everything happening in that part unlike the feeling of someone skipping Punk Hazard or something because literally every single arc connects and it's an basically an epic.
@@Glatier however skipping one piece is also not very important, so many one piece ideas have just changed on the fly like haki so skipping some PRE TS or POST TS arcs like fishman or long irng long land or whatever its called wont cause much or any damage at all
@asdasd-dm1hd You are wrong. Fishmen Island has lore that is very important to the overarching story. I think you have not even watched/read One Piece am I right?
@@Glatier I strongly disagree with this idea, every single part connects in important ways, if you skipped GW you’re just missing out on a great story but that’s besides the point. When it comes to continuity between parts you’re missing out on what happened to Polneraff, the introduction of the concept of DIO having kids, what Jotaro was doing between DiU and Stone Ocean, the concept of stand evolutions being unfathomably powerful depending on the obtainment method, new worldbuilding that expands our knowledge of a different part of the world and how it’s been affected by stands, we get the origin of stands and stand arrows, and most importantly more information on how the concept of fate works in JBA. To say that any part is just skippable is to say that any 1/6th of a story made by any great author could be ‘cut out’ and you ‘wouldn’t lose very much’ because by all means, part 5 is very important to our knowledge of the JBA world and foreshadows/explains a lot of future events in part 6 indirectly.
@@notdavidsecondary9958are you stupid? sogeking is sooooo obviously Zoro. His nose is bandaged so much it looks like Usopp’s, and we all know that Zoro is almost always injured and handicapped. And, if you’re wondering about how Sogeking fought alongside Zoro, it’s because he was constantly using the demon thing that triples the amount of bodies he has. Furthermore, we never see Zoro use the demon thing and Sogeking in the same room, therefore there is no proof that Sogeking isn’t Zoro.
The graph is exactly what I expected, and also why I love Jojo so much. It isn't like any other show on the list, the main protagonist changes with every single part, and aside from Jotaro, Joseph and Dio, there is almost no character that is present in more than one part. While One Piece always brings back old and forgotten characters and shows the ramifications of the arcs that happened (and I love it), Jojo keeps a small cast per part.
@filipemartinho1753it's similar to pokemon where every generation is a different cast but there are still easter eggs that refer to previous characters
it's probably because jojo's has several protags and a main cast that changes every part so the characters that appear the most usually only do in their own part
I see anime have three Protagonist stucture styles 1. "Main" Protagonist Structure - This classic format centers around one central character, with the story primarily focused on their journey. Side characters may have their episodes, but the narrative revolves around the main protagonist. Examples include One Piece and Dragon Ball. 2. "Ensemble" Protagonist Structure - Here, the story balances its attention across a group of characters. While individual episodes or arcs might spotlight specific characters, no single protagonist dominates the narrative. Examples include Macross and the more recent Ishura. 3. "Shifting" Protagonist Structure - In this format, the main character changes between seasons or arcs. While each phase has a central figure, the overarching story accommodates multiple protagonists over time. Notable examples are JoJo's Bizarre Adventure and Uma Musume. Of course, hybrid models blur these lines. Durarara!! is largely ensemble-based but also highlights three key characters who drive the plot. Similarly, Teen Titans (the original animated series) operated as an ensemble but gave each season a shifting focus on a particular character. Or the shifting ensemble of Digimon. You might get a different curve with an ensemble show or with one of hybrid styles.
Bleach is more like 1 but with a bit of 2. It is very centered around the MC group, but the other groups do get a lot of spotlight without the main characters, and get their own narratives. Heck in the last arc the MCs are not even present in half of it
@@diablo.the.cheater I find very common that very common in late stage shonen action anime/manga. The world has become so big, the main character loses primary focus. That might be an interesting thing to research, the exact moment when the MC stops being the MC.
@@Zoro-gk5ow I agree, One Piece has a strong ensemble. But the story generally bounces between Luffy and someone else. Maybe a better example of an MC focus structure.. maybe like Future Boy Conan
@@bennypmo but the thing is that the story of one piece doesn't necessarily revolve around Luffy...sure he's an integral part but the world keeps on moving and changing without much influence from Luffy....soo I guess the first one is correct but it's also not kinda?
My favorite part of JoJo is the blood lineage structure as opposed to the continuous Journey and/or tournament. It's kinda nice to see one story end for the next family member to continue it for them.
Hunter×Hunter would (probably) turn up a very different looking graph because of how each arc introduces new characters that sometimes become POVs, in occasions they even become regulars, and some will pop out of nowhere in a future episode after appearing very briefly earlier. And that's just talking about the anime, since the manga basically changed protagonists in the most recent arc.
Also worth mentioning would probably be World Trigger, which seems to add characters without that much of an introduction and make them regulars (to the point that it's kinda difficult for me to follow the story sometimes because of the massive cast).
Jojo’s is kind of a villain of the week show. In part 3 and 4 enemy stand users usually get like 1-2 episodes before losing and never being seen again. Part 4 has characters show up more, because it’s centered in a town and not a journey. One Piece typically has even minor antagonists appear in several episodes. I think that has a lot to do with why they’re so different.
I think the reason for jojos abnormality here is that jojo is like several different stories that all take place in a shared universe. characters from previous jojo parts can show up in later ones, but fundimwntally its more a bunch of mid length shows all wrapped under one umbrella. to put it simply, its an entirely different category than a short anime where all the characters are shown all the time or a long anime where a core group is the focus in a world of characters. I imagine if you found a series that is also made of multiple stories in one shared universe it would follow a similar trend try the "neon genisis evangellion", "code geas", or "a certain (magical index, scientific railgun, etc.)" series as they all fit this criteria. maybe they would align with jojo more than the others
@@someonehavinganidentitycrisisa bit more if you count the one created by made in heaven, the one re-created after Pucci died and the various that Funny Valentine visits while trying to escape the infinite spin, also 3 more that time they shot Johnny.
@@Glatier the light novel is written very smart, but the series as a whole can be chopped up to an action harem with a lot of "lolis". grossed me out. apparently it gets uber good when the odin arc happens and stays that way, but thats not animated and I dropped it before that.
As someone who started watching/reading One piece when Water7 was getting written and started reading jojo after part 3 was animated... I really cant decide which is my favourite anime, but they are both in my N1 spot. They are completely different but for me its like Reading the Odissey and going to see the Parthenon, they are both works of Art but in very different categories and they each give completely different vibes.
@@notdavidsecondary9958 yeah, OP is only the 16th longest by episode count, with Conan coming in at number 15. Doraemon is either number 1 or numbers 12 (~1407 eps) and 5 (1787) depending on if you want to treat all the adaptations as one long franchise or 3 separate series (the first isn't worth assigning a number to due to being only ~50 segments long). The symbol of constancy and persistence though is Sazae-San, which has been running weekly since '69 with only around 9 weeks of hiatus (4 in 1975 due to an oil crisis and 5 in 2020 due to obvious reasons). The episode count is ~2771. If it keeps going you'll eventually need to watch 3 eps a week to catch up to it in your lifetime... Stats taken from Wikipedia.
Conan has to have thousands of characters that show up less than 5 times, right? Can't imagine there would be many mysteries that last more than five episodes (usually it's like two)
@@hiurro I mean, there's a little shy of 700 pages in the "characters" category on the (non-fandom) wiki, so I think it's safe to say that that's probably a good approximation of how many characters there are. Which makes sense if you have a few recurring suspects that reduce the number of new characters per case. If we assume there's some frequently returning patsies, it's easily possible to have case with no new characters, and then there's the various case-less eps (filler). With only "a little"* over 1000 eps, it makes sense that they're not even three quarters of the way to their first thousand characters. *If you can call almost 150 eps (or roughly 11-12 cours) "a little"...
this tool is actually really interesting. i hope i can structurally make models of character relations as well in relation to the plot or arcs. thanks! great vid and analysis.
I think it's just a thought-provoking research based on how important characters to the show really are. The point isn't exactly informative, but you know its just a silly diagram. It's up to you if you find it pointless or not.
you should do thiw with manga, since these stories are planned in chpaters not eps. Few of these long anime restructure the story to match the eps. really nice
That’s precisely why I love JoJo, it’s not everyday you make 9 unique main characters, have each of them get their bro or sis that fits them and have a story that improves along the way with it
1:22 Oh I was thinking about the Whole Cake to Wano crew splitup, Chopper was with the crew on Whole Cake while Usopp stayed on Wano and sold toad oil lmao Same deal with Nami and Zoro, thats why he ranks below her
@notdavidsecondary9958 Easily the most interesting character, and I'm really saddened that Oda sort of just forgot about him. I really do wonder WHERE sniper island is though.
JoJo is super unique when it comes to everything, despite being one of the most commonly immitated or even plagerized anime ever. One thing that makes it unique is the parts structure, which is the main reason it looks so different than the other shows. Even the protagonist, main cast or main villain don't stay constant. Plus, while Naruto, One Piece and Bleach have some villain of the week, it's not as heavily emphasized as in JoJo.
@@javierslytherin9898 Have you never watched another shonen in your life? Without JoJo, all other battle shonen that came after it wouldn't exist. They all copy hamon and stands to some level. There is a reason why so many people said the phrase "is that a JoJo reference?" after seeing JoJo to the point of it becoming a meme.
The parts structure and the main cast changing isn't unique at all, it's been around since Kamen Rider (1971) and Super Sentai (1975). There was also this little-known Super Sentai ripoff called 'Power Rangers' that was really popular in the 90s, way before Jojo's became mainstream in the West. Edit: Ultraman did it since 1966
Jojo's has basically a timeskip after each part, most returning characters become secundary or mention only to give way to a new roster. No character has appeared in every single part, and part 7-9 are a completely new continuity with close to no recurring characters
originally that is what I was going to do, but I realized it would be a much more interesting video to do that for a larger variety of anime and see if you could do some sort of clustering based off pair-wise distances. This was still KS-distance-esque so I left it as is. I have read papers that KS distance isn't that great for heavy tail statistics though, will have to think about that
Think this happens because Jojo has different parts and like only 4 characters actually meaningfully appear in more than one part and never more than 3(yes I know Jotaro also makes a cameo in part 5 and part 2 which would make his count 5 but it's too minor to count) parts so it makes sense it would be drastically different from the others.
To me this divergence suggests that any given part of JoJo's operates as a story with a smaller central cast, like the 4 protagonists of Diamond is Unbreakable vs the 10 straw hats Interesting stuff, further research is a must!
I wouldnt say "its not ideal" that so many long running anime/manga have such a significant overlap in characters vs appearances. I'd argue that's a really interesting quirk of storytelling especially at length. I feel since the other anime you pulled were "similar" long running shows, that the amount of and length of arcs is probably the reason they are so close. I predict that if you tried other "Monster of the week" style anime that they would follow a pattern similar to that of jojos, with a much quicker falloff with a shorter graph. This series of videos are so cool. Genuinely good work and great finds!
I think the structure of the story is most important for character relationships. Most often we only follow one main cast, which travels to many different locations always meeting new people. These location specific groups rarely interact with each other. But they are often only one or two main character relationships away to any other person in the world.
You should include Legends of the Galactic Heroes (Ginga Eiyu Densetsu), and the reason for Jojo's abnormality is because effectively each 'part' of Jojo with a few exceptions is essentially a new series with a whole new cast of both protagonists and villains, often with only one or two ties to previous parts.
Aside from what a lot of people already said about Jojo's story being separated into very clearly different parts that feel like individual stories, there's also the fact that out of all the other anime you compared, Jojo is the only one to have been adapted when the full manga story was already done (up until what is already adapted, that is) and in the current seasonal format instead of an on going series throughout the whole year. This helps to take it even further away from these other shows, specially One Piece and Pokemon, that were made not only while the manga (games in Pokemon's case) was still going, but a lot of times became almost caught up to it, which lead to either a lot of fillers and/or dragging and pacing issues that Jojo did not had. A Jojo character that appeared in 5 episodes could appear in 20+ episodes if it had from the start the same adaptation format that these other anime had (just an example, but there are a lot of Jojo episodes that adapt about 4 manga chapters, so the math kinda checks out in most cases). And an unironic fun fact is that even though I just said that One Piece and Pokemon are the most further away from Jojo in this comparison, they also have the the closest story structure to Jojo, in which an arc lasts for a couple years and the next one feels almost like a soft soft reboot of the story (2 softs because they still have more implications and connections to the main overarching story than Jojo, that feel more "rebooty" in this aspect), with the introduction of an entire new cast of characters to that specific setting and plot line and maybe an addition/switch of one of the main crew.
I've just started my research, but I have a theory that many of the big writing issues in the most popular shonen are due to their veneration for and inspiration from Dragon Ball, which was further solidified by the popularity of the 'big three,' either in general or in the western market specifically. It was very cool to see a graph that visualizes the similarities between the character appearances of One piece, Naruto, and Bleach. Thank you!
I think the trend your seeing may come from the average number of characters per episode being relatively stable. A histogram of the characters per episode would show how true this is, along with a moving average of the number of characters per episode over time. If the number of characters per episode is roughly fixed then we can use that to establish an exchange rate between characters based on the number of episodes they'll appear in. For example, ten characters who appear in ten episodes each would be worth a hundred characters that appear in just one episode each. Put more generally, the number of characters and the number of episodes they appear in should be inversely proportional.
oh that is a super interesting idea. I'll have to think about that tomorrow. I think for the most part if you look at the graph at 1:03 the number of character per episode is mostly constant on the whole (given it spans approx 1100 episodes). I havn't checked it for the other shows but I like that idea!
@notdavidsecondary9958 I was partly making a joke 😂 but actually, I wouldn't create this graph using the One Piece anime since the pacing is all over the place and all the extended recaps at the beginning of each episode would mean some characters would technically show up in episodes in which the chapter counter part doesn't include them. Also, the manga has "cover stories" that the anime doesn't adapt however, the normal/colour spread cover pages might also confuse the data collection because they often use characters that are not in the current story. But yeah TLDR: Anime repeats/adds scenes that are not in the equivalent manga chapters which are more concise. But also I was joking 🤣
This difference is likely due to the pretty strict segmentation of Jojo. With only a couple rare exceptions most Jojo characters are only seen within their part/arc, whereas you're much more likely to see a random previous character from an early episode in other shows.
Jojo's curve is evidence of the power of smaller scale, focused stories. When I think about the casts of the other stories, vast quantities of those 1 or 2 appearance characters are forgettable and unimportant. They are mostly background. Jojo being divided into smaller parts that follow different core casts on interconnected, yet smaller stories, make for a long running series where even the characters who only appear once or twice EXTREMELY memorable and impactful. there are many Jojo villains and side characters who I can vividly remember, quote, and recall important scenes and direct impact those 'smaller roles' still had in a massive, long running story. Example: When you (presumably a Jojo fan) hear "N'doul", you probably immediately remember how impactful he was to the plot and how his stand worked, etc. He barely had any screentime, but he MATTERED. The problem with so many long running stories' bloated casts is that there are interesting, potentially compelling characters who will unfortunately fall to the wayside. There are so many characters who barely appear in Jojo but they are so interesting and special that there are fans who would cite those characters as being among their favorites
What you can clearly guess from this visualisation is how big the main group of protagonists is and how well they hold together. Basically the longer the graph needs to drop the bigger the main group. (higher number of characters appear more often)
That's interesting!! I do think that JoJo having very distinct parts with almost no shared characters is what makes its graph so different. All the other shows tend to have a main cast that occasionally gets expanded, with a bunch of side characters that only appear for a season or less, so that's probably why they look so similar. I think it would be interesting to quantify what I think makes One Piece special, which is how characters can become relevant again hundreds of chapters after their first arc ends. With your current graph that information is removed: it doesn't matter if a character appears 20 chapters in a row, 10 chapters in a row and then 10 more four arcs later or a single chapter in 20 different arcs. That's shown in the raster, but it's hard to read there, so having a graph showing the maximum/average period of episodes between character appearances after they are introduced would be interesting. I'd expect One Piece to be significantly different there, but I would like to see what the numbers actually say.
Hey this video just popped up randomly for me and while I didnt understand anything you were actually saying (not your fault Im just smoothbrain) it was really cool to look at the graphs for some reason keep it up buddy
I probably would never think about this (despite never actually watching either, but if I did wanted to get into one piece for some reason the newly animated series could make it a short watch) Nice job.
I find it interesting that Pokémon follow roughly with the other anime, but it makes some sense considering it's nature as a long running anime. It's interesting in that it has three characters who appear in almost every episode, being Ash, Jessie and James, and if the Pokémon are counter Pikachu and Meowth, Brock would follow in not far behind considering his prominence in the Anime since Indigo, and apparent from Orange Islands, being a major character in every season up till the end of Diamond and Pearl. It's interesting in that much like One Piece, beloved characters make intermittent reappearances, but for the most part the seasons take a general JoJo like approach in that they have new small groups being the main characters when they switch over too a new season.
You should add fairytail to the list. There is an abundance of characters that make their way in and out of different episodes. I think the graph for it would turn out prety interesting
5:05 Maybe you could normalize the X axis, as % of episodes. since 200 vs 1000+ is quite a difference even in a loglog plot. I would also be interested in seeing a semi-Log plot where only the y axis is a log plot. I know that would use all the space on characters that appears a lot. But I think that makes sense since that is also how watching the show feels. Characters that appear in nearly every episode is a big part of the show. So fair that they take up a lot of space on the right side of the plot 🤷♂. 4:09 I don't understand the Y axis 10^0=1 that much I get. But 7 appearances 50% that would be about 10^(-0.3)~0.5. I guess that could be -0.3. So the Y axis is % of characters if I got it right. So the Y axis is already normalized to a value between 0 and 1. It would be nice with ticks, minor ticks and maybe even a grid. It would make the plot easier to read at least for me. 0:27 This plot mush be normalized somehow. Cause one piece have many more episodes than JoJo. I also would have thought one piece had way more characters, but on the plot it looks like each character take op the same width. Maybe it is just the first episodes of of one piece on top of all the episodes of jojo. (am I confused?) It is interesting that jojo stands out. I have not watched the show. But it looks like the show changes main characters as the show goes on. I guess the others looks very close due to the loglog plot 🤷♂. Or maybe shows have a budget for a number of voice actors thus they limit the number of new characters in an episode and the number of recurring characters in a season. And that is why all the other show look so similar. It would interesting to not count episodes but minuts on screen, although that data would be much harder to obtain. Guess that could be a job for a AI in the future. Or maybe a system like the once they use for sports data like football where they record every contact with the ball, every meter every player ran etc.
great points (as always) I'll reply point by point lol 1) I had thought of this, and I think it is something I'll think about more but the issue that I had was that it makes the graphs even more confusing to read. Here they all start at 1, but if you normalize by the total number of episodes, each graph starts at 1/NumberOfEpisodes which is not the same for each show. This made the graph extremely confusing without more explination. However, nothing fundamentally changed. 2) semi-log could have been good to check. Admittedly that was sort of force-of-habit. In my IRL work I use log-log all the time and so here it was just the first thing that came to mind. 3) Your're correct on the y axis stuff. The lack of tick marks was an annoying thing that happened when I tried to import the .svgs from python to blender. I need to look into how to do that better in the future. I tried to mark them using those little horizontal and vertical bars to compensate, but I agree some tick marks would have been better. 4) Oh yeah here the actual vertical height is technically meaningless. I mainly just wanted to be able to show the overall structure, but if I was to plot them so that the dots were the same in both shows, you'd be barely able to see Jojo because of how large one piece is. I could have done a compromise to animate jojo from small to large to make that explicit. 5)regarding your last point -- thats something I've been thinking about quite a lot. Maybe there could be some feature recognition one could do but im not sure. I know in some arcs too the characters in one piece also change their clothing which would make it hard for an AI to track them. I think its a super interesting problem but looking into it would also be wayyy more effort than I can afford right now haha. I'm just having a lot of fun doing these little side videos right now during the holiday break lol
@@notdavidsecondary9958I agree with OP's first point, since if you have an anime with say only 12 episodes, its plot would dive down earlier than the plot for JoJo, if the horizontal axis is the number of episodes. If you divide by the number of episodes, the horizontal axis will simply become the proportion (or percentage) of episodes the character appears in. The leftmost point of the plot would be different depending on the anime, but that doesn't matter (or you could connect the leftmost end of the plot to the vertical axis with a horizontal segment, which might be clearer; just a thought though)
@@notdavidsecondary9958 As always on this channel I get a much better answer that I had any right to get. 🙏 1) It is true that for the first one it might be more confusing. And it might also end up just showing differences in show length even more, I can't really imagine how it will look. But I think it might make it easier to see if a show have many repeating characters or many new characters compared to another show. No matter the length of the show. 2) I always just get straight lines when I am using loglog plots. And I am addicted to plots. So if I already have the data in python I often make both linear, semi-log's and loglog. Just to see the data from different angles. I often also look at the derivative or a cumulative plot. They are all easy and fast to make. So I might as well 😂. 3) That is a good excuse. I have not tried importing matplotlib plots into blender. I would probably just have save the plot as a high-res PNG pic. (Cause that is what I normally do). And then just imported the PNG on a surface. But now I want to try to import a svg file. (I guess you cannot import a pdf into blender, can also be vector based). I have never tried importing vector graphics into blender, but I have often had problems with importing vector graphics in other programs. 😬 5) Yes tracking characters from video would be a huge undertaking. Maybe it is possible to use the subtitles... But often the subtitles does not state who is talking. That would have been super helpful. But still not perfect of anything. But manually going through every scene would be quite an undertaking. 😂 But well this is always the problem with data, there are always some details missing. 😂
@@notdavidsecondary9958 Also, the algorithm loves this video. It took less than 24 hours to overtake the "Comparing Massive Anime Character (and Brain) Networks" 🤷♂
@@Petch85 Good questions are hard to write, they deserve good answers. Just to comment on your point 3 -- the reason to use an svg over a transparent png is that you can animate the datapoints in it (e.g., when the curves appear, or when the dots are made bigger and stuff like that). Some of that is possible with a png with some clever cheats, but it limits you a lot. The big issue is that it imports everything as a curve without a profile, and curves without profiles don't render. So for the tick marks I would have to go to each tick mark and assign them a profile, and as far as I can tell you can't link curve profiles like you can material (though now that I think about it I should double check that...). But also yeah, I agree its been interesting seeing the algorithm do its thing so differently. I'm actually a little worried because this was genuinely meant to be just as a side channel where I can upload random quick-to-make stuff, but now the audience is very different to the main channel. That does give me an idea for a future follow up to the 'youtube subscriber network' video on the main channel though...
Definitely interested. My theory was initially that it came down to the continuous arcs vs villain of the week structure you mentioned but Pokemon as I remember it was more a of the week sort of show than a shonen arc type show and it is pretty close to One Piece. Then I thought about it more and you mentioned the generational divide structure of JoJo also makes a compartmentalized raster and network diagram, so maybe it has to do with JoJo's kind of unique generational structure. But Pokemon also has generations though not as harshly divided and once again it's not even the closest to JoJo. As for anime I'd suggest, maybe YuYu Hakusho or Yugioh. Not an anime but I'd love to see it and it'd be as big or bigger than Pokemon and One Piece, I'd love to see this done for Monday Night RAW.
Actually you hit on something I want to talk about for pokemon and this would be a great opportunity to test something -- how much of the pokemon anime did watch? Like mainly early pokemon, or more the newer stuff? Or the whole thing perhaps?
@notdavidsecondary9958 I watched through the Gens 1 - 3 stuff as a kid. I would then peek in every now and then to see what's up with new moves. The structure seemed to be Ash traveling across the region encountering new places, people, and regions until Alola where the structure shifts to him staying at the school with a larger cast of consistent characters with little excursions.
@@jarekguevara6563 That's perfect, thank you so much for replying and letting me know :) Yeah my experience was more or less the same (I think I fell off around the start of gen3). And yeah exactly, there is this very clear shift from the original one-off characters format to a larger more consistent cast in later seasons, and that's why it follows one piece so well. I'm hoping to do a video on that in the relatively near future.
A hypothesis: if you were to perform the conversion of each individual JoJo arc to this appearance table it would also follow the same trend albeit at a smaller scale as total appearances would be lower. I would also posit that this trend would even exist outside of anime, perhaps it would be worth pulling samples from popular western shows with similar run lengths like House, Supernatural, etc. with a diverse enough data set we may see an overarching trend amongst all shows that have a core group with supporting characters that may only ever be in a few episodes.
very much agree, I suspect the same would happen. I think House would be an interesting one to compare against. I wasn't a devout watcher but I've seen some here and there and from what I remmeber it's got a core cast that appears almost every episode and then the patients which are almost always one-offs. That's actually why I included pokemon thinking it'd be the same but in the later seasons they really changed up that format, at least from what the raster shows.
You should use the "physics" method. First make a list of all the variables that are important, for example "number of chapters" "number of characters" and "total appearances". And then analyze separately how each of these things should affect the final value. Should the number of chapters positively or negatively affect the final value? Should the number of characters make the final number bigger or smaller? Finally, you multiply all the values that should increase the final value, and divide them by those that should decrease the final value. This gives you a basic formula with a high probability of being precise (although not exact).
This video is a pretty good example of how easy it is to misrepresent things with statistics, just by organizing a graph in a way that muddies our understanding of the data
@@Lazypackmule The number of itterations… as in… one? Changing the histogram to a cumulative distribution function to address the ways in which a histogram misrepresents heavy tailed data? I guess I’m curious what your remedy to this would be then - Leave it as the misleading thing?
Apologies, the tone of my comment was perhaps not the best reading it back. I am genuinely asking out of a sense of wanting to understand how it is a misrepresentation. If your comment had been "this was not presented well" then I'd be like, okay yeah I can see that, CDFs can be unintuitive and I'll work to fix that for the future. However, your comment was that this was actively misrepresenting the data, and that is what I was hoping to understand. Typically the step (iteration?) of transforming a histogram to a CDF is not viewed as a misrepresentation of the data (in fact, often the reverse would be more true because of how poorly the histogram represents heavy-tailed data).
Jojo is like 9 diffrent series in a threnchcoat. hell starting with part 7 the series isnt even set in the same canon anymore. Get the number of apperances of Baruto characters and Combine this list into a Baruto+Naruto Set and youll see something closer to the JoJo line then the standard trend. Love this video. Try adding smaller series like Death note to see if the data changes depending on series length. It might also be a shonen trend exsclusively, sailor moon would be a good example to test that against.
Im curious why you think my data is "all weird"? I purposefully chose it for the contrast so I'm not sure I understand the comment. Also regarding the shonen comment -- that's a great hypothesis and that's what I thought too, but that is why Pokemon was included
@@notdavidsecondary9958if jojo is different because it is a combination of multiple stories featuring their own characters, then if you combine a few unrelated animes, would you get similar results as jojo?
@@notdavidsecondary9958 im not critisizing you, i like your data. Im just trying to hypothesise on the data. Sorry if that wasnt clear. This is a great video.
@@inverted_paradox4170 haha no sorry, I did not in any way read your comment as criticism, tone just doesn't carry over well over youtube comments. It was very clear you were giving genuine ideas/ feedback :). I was just curious for clarification as maybe I had missed something that was obvious.
I mean one thing that sticks out to me is how the graphs dont have the same max number of appearances, with jojo having the least, so of course it ends earlier you could probably just do something along the lines of what you did for the number of characters and theyd be even more similar
Nice video! Though I wonder how did you count the Pokemon themselves in the Pokemon character list? Do all appearences of a species count as the same character even though they may or may not be different? If not: do you treat legendary Pokémon differently (as lore may imply that they are unique)? Looking forward to future work!
Yeah that's a great question. The way the code is currently structured, I believe every pokemon is just counted as the 'generic' version of it. i.e., 'ash's pikachu' would just count as every 'pikachu'. Bulbapedia does often denote unique versions in parenthesis, so I believe it should be fixable and if I do make that follow up looking at the pokemon raster that is something I'd fix. For the question about the legendaries, that one im not sure -- it really depends on how the fan wiki stores that information.
@@notdavidsecondary9958 Interesting, thanks for the answer. Sorry if this is already explained in another video, but how do you actually get the data? Webscraping from the wiki?
effectively yeah, I use the beautifulsoup python module. Some wiki's (e.g., wikipedia) let you download the entire thing, in which case I'll do that so as to not stress the host server, though if that isn't an option I use plenty of delay timers to slow the program down.
I think what this shows is that one piece, more or less, established the normal distribution for anime given its length and age. Most anime will begin and end in the time One piece has existed, so it sets the bar for Shonen anime and Manga to "follow".
Honestly, since you hinted at it, I'm now really curious about how different genres compare, especially since you hypothesized that it may be the nature of this genre/demographic/longrunning nature of the shows you compared seeing which other genres' graphs JoJo's is more similar to might highlight something about it
50% of characters appearing for only 1% of the show to me does kinda show that maybe One Piece has a bit of a problem with drawing things out too much to the point where there's enough time for some characters to be introduced only to be never seen again almost immediately afterwards.
mimics real life though, you can meet someone on the street and have a lovely conversation and never see them again, i think it's a service to enrich the world, also this data comes from the anime which it shouldnt but too late i guess
@@ultratronger it's a fictitious story though. If I wanted a story that gave me the experience of real life I'd go outside or read a biography or something.
@sarbe6625 what story gives you the feel that the world is alive and breathes and feels so full of life and it's always moving even when you're not looking like One Piece? Certainly none in manga. It's easier to compare it to something like A Song of Ice and Fire.
The Jojo divergence probably comes from the fact that fewer of the main and supporting characters consistently across the show due to the "Part" structure
It might be useful to take this idea and use it to compare genres of shows. Something like Detective Conan, while it would have more episodes, is a lot less serial and plot driven iirc. Doraemon might be a good comparison with it. I wonder what would be revealed for other shows out there, like sitcoms? Seinfeld versus Friends we might see a similar curve, but what about something like Days of Our Lives the longest running daytime TV drama? The soap operas? Those things run long enough on human life scales that characters literally grow older and have children through the series as a staple for the longest running ones.
:Turns abstract understanding of anime structure into data analysis question: "I have carefully set this up to be sure I'm asking a question that makes sense." :Sees data analysis: "I have no idea what this means." Been there, buddy. In this case, it might be interesting to take one step back to the log/log plot and compare each show on that. The CDP approach is useful for *finding* these kinds of divergences, but not especially helpful in drilling down. My theory is that there aren't any characters in Jojo who persist through the entire narrative, so you get this stack of "casts" from each arc that are there for that entire part of the story and then just disappear forever. The other shows slowly expand the main cast so you get a more gradual termination, lifting the entire line up a bit and extending closer to zero. You can at least see *that* in your comparison.
Legend of the Galactic Heroes(1988) might be a good point of comparison because the cast is pretty evenly split between both sides over 110 episodes. So it might be interesting to see whether there's a bias between either side. Ranma 1/2 or Avatar the last Airbender could also be interesting pics to see whether other genres fit the shounen trend.
Another anime that might be worth checking out is Gundam, especially the first Gundam series - while a lot of villain of the week stuff does exist at the start of the very first run of Gundam, even that early, Char and a lot of the higher Zion generals are introduced early, though the show does use them wisely and sparingly. I'd be very interested to see what its stats look like
JoJo's is, more or less, a world, and each part within is it's own story, tied loosely to another by a couple strings and the place it is charted upon, unlike most other series, JoJo's is practically an anthology, explaining it's divergence for the norm, especially with SBR and ahead being essentially on a whole different plane.
Instead of having total number of appearances, why not try percentage of episodes the character shows up in, this will make the list more standardised instead of having lines that end in different places. It would also mean you don't have to log the x axis and it will work better for shows with varying numbers of episodes.
I did try this, it doesn't change anything. Actually it just makes the plots much harder to read as they no longer allign anywhere. The smallest value would no longer be 1 episode, but rather 1/N where N is the length of the anime, which is different for each. This wouldn't be a problem if, say, each anime had a character who appeared in every episode, but that isn't the case for any of them, and especially not the case for jojo. Moreover, since this is just a uniform scaling, it doesn't prevent the need for log axis, it just shifts it from the domain.
It would be interesting to see how other genres of anime stack up. Like adding ranma 1/2, sailor moon, and card captor sakura would be interesting shoujo picks. Detective Conan or Sgt. Frog would be good for a childrens anime. You could also compare it to western anime-inspired shows like ben 10 and avatar
I connject that Jojo looks so different because it regularly ditches it's cast. Luffy and Goku and Ash never go away, but . I predict that Digimon will closely match Jojo, if not be more extreme because across the 12 seasons, there are 8 separate continuities, that exactly 2 characters have crossed (In the show, more have in the manga and games)
3:36 Why not? That perfectly exemplifies Zipf's law, a linear regression with slope ≈ -1 ! It also predicts word frequency, city sizes, popularity of chess openings, etc. There is a Vsauce video about Zipf's law mysterious appearances, you should look into that
Apologies I'm kind of confused by the "why not?", I give a few reasons for why it is not a good representation. Was there something there you disagree with? Regardless though, Zipf distributions are really cool! and I was going to talk about them in this video but they are also very tricky and you need to be careful in seeing Zipf where there actually isn't any. If you're interested to read it, I'll also include my reply to another similar comment here: I’ll caution against jumping to conclusions and thinking its zipf (actually this is not zipf but rather something called power law), especially when plotted in that form (what you want to do is logarithmic binning). To conclude that it is "zipf", you need to perform maximum likelihood estimation to get the exponent (caution: Linear regression to get the slope is *really bad* for this do not use it), calculate its pvalue and then alternate hypothesis testing against other distributions (see Clauset 2007). I haven’t done any of these so it's premature to tell, but keep in mind that the cdf of a powerlaw is itself a powerlaw and our CDFs are… doubioisly powerlaw (which is not to say they cant be powerlaw, cdfs can over exaggerate differences especially when in log log scale).
@@notdavidsecondary9958 Yeah, I agree it is too early to declare Zipf's law; any straight line on a log-log graph boils down to a power law. But extrapolating a line through the first ~20 points (where the bulk of sporadic characters are) crosses axis at around ~(1, 10^2.5) and ~(10^2.5,1) i.e. slope -1, which is suspiciously Zipf-like. I don't have any problems with using a CDF, as integration does help smooth out noise. But using the cumulative obscured a bit that connection. Also I was half expecting for no one to read my comment, I'm glad to be heard
It makes perfect sense why it is that way. Luffy, Naruto, Ichigo, Goku and Ash all are THE MC of the show. They have a few characters that stick around, some longer some shorter, but the core of the show's crew rarely changes. Jojo's on the other hand has Parts, meaning that there really is not singular MC like that (tho the Joestar Bloodline does connect the Parts, as well as Dio who has "something to do" with every single villain/antagonist) Id say the show thats probably the most opposite to this (i guess) is Cowboy Bebop. That being said the Show is also very short lol
![Seungmin "그렇게, 천천히, 우리(As we are)" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/kAzmhLHePqU/mqdefault.jpg)
What are other anime you'd like to see included? I want to do an analysis for a future video that requires some more anime, so what would you like to see? More diverse the better
Haikyuu or other sport animes could be interesting, since you can probably see the matches and rematches against specific teams really well
I mean how long is long? some of my personal favorite run about 5 or 6 seasons. MHA (My Hero Academia) has 7 with an 8th coming soon, and it's infamous for having an impressively large cast of side characters (who's utilization is up to debate).
There's also Bungou Stray Dogs which has 5 seasons and it's very faction focused, in fact I think the whole series (animated) only has about 50 characters because it's all a bunch of tight knit groups rather than monster of the week which I think would be rather interesting to see how it diverts from the massive casts of the other shows.
I know FMA:B (Fullmetal Alchemist Brotherhood) is another longer series that's very popular however I've not watched it so I can't comment on it
Another one that I think would be interesting because of its uniqueness is Haikyuu!, as a sports anime it's cast consists of teams, so rarely do you ever just get one character popping up, the whole team does. I'm curious as to how that would appear on the graph!
Those are just a couple suggestions!
ohhh yes, Haikyuu is a great idea. Sports anime in general would be interesting to look at in general
I meant to include FMA:B but it completely slipped my mind. Maybe even a comparison between FMA and FMA:B would be interesting (at least for me)
@@notdavidsecondary9958 That's another great comparison to make yeah! How the two versions differ in cast makeup
God I love being a stats nerd NUMBERS ARE FUN lmao
I think it's weird that you didn't mention Jojo's obvious difference to other anime, the Parts. Jojo doesn't have any characters that stick around for the entire series, and it barely has any characters that show up in more than one Part. Jojo is basically 9 different series with some crossovers, and i think if you make graphs for each individual Part, they'd more closely align with other anime series.
Yeah I agree I should have mentioned that explicitly. I talk a lot abut parts in the main channel video and here I just kind of alluded to it but didn't say it explicitly (thought it is pretty evident from the Jojo raster).
thats actually a very good point!
I think you meant to say 9 different shows, since a series would consist of the whole thing.
@@truthcrusader360 Series is usually how you would refer to a manga, which they would have to be since they said 9 and JoJo only has 6 parts adapted to anime. You are right that the whole thing is also a series but there isn't really a better word for each part (aside from part) If it was an american comic they could use the term "run" but that isn't really a thing in manga.
DIO in some way plays a major role in all of the first 6 parts and is referenced in 7-8 with his alternate universe counterpart being a major player in part 7 with 8 being indirectly affected.
It’s not multiple separate shows it’s 2 deeply interconnected worlds that make little sense if you don’t have the knowledge of the previous parts to build on to the next ones. it’s more like really long movies in the same series, kind of like the original StarWars trilogy, or Heisei era of Godzilla movies, all of them are decent stories on their own but only truly shine when they’re in the context of one another in a linear viewing experience from each part to the next.
One Piece Data: 50% of characters do not appear in more than 1% of the show
One Piece Fans: "GIN AND DON KRIEG ARE BACK"
Even fking Pearl broo 😭😭
WAIT WHAT REALLY
@@seviiaal Yeah!! They are briefly shown (in present time) during the Egghead arc. I forgot which chapter, and I dont wanna be too specific n spoil anything.
Does nobody die in one piece? Gin got freaking poisoned and had limited time.
@@Attack_Pillow1. They made it pretty clear during baratie he was kind of a freak that could survive it and
2. no not really. Oda doesnt really use death for realism and imo drops the ball on a handful of occasions with it. but its also just not his thing to kill off ppl
Those two show are complete opposite with each others. One Piece focus on continuous saga storytelling meanwhile Jojo had each distinct story. One Piece emphasize world building, Jojo emphasize chronological lineages.
Exactly. This is why I tend to say skipping One Piece arcs is wayyyyyy worse than skipping Jojo parts. Like, let's say you jump from DiU to Stone Ocean, if you skip Golden Wind you wouldn't find yourself confused by anything at all since SO has absolutely 0 to do with GW (hell the whole concept of Requiem doesn't even return either even). I'm not recommending people to be part skippers, but what I'm gonna say is that if anyone happens or so chooses to start JoJo with certain parts other than Phantom Blood, they won't be having that hard of a time making sense of everything happening in that part unlike the feeling of someone skipping Punk Hazard or something because literally every single arc connects and it's an basically an epic.
@@Glatier however skipping one piece is also not very important, so many one piece ideas have just changed on the fly like haki so skipping some PRE TS or POST TS arcs like fishman or long irng long land or whatever its called wont cause much or any damage at all
@asdasd-dm1hd You are wrong. Fishmen Island has lore that is very important to the overarching story. I think you have not even watched/read One Piece am I right?
@asdasd-dm1hd Well uhh... don't be surprised when you see Luffy destroying FMI and some random mermaid girl calling upon a horde of sea kings then.
@@Glatier I strongly disagree with this idea, every single part connects in important ways, if you skipped GW you’re just missing out on a great story but that’s besides the point. When it comes to continuity between parts you’re missing out on what happened to Polneraff, the introduction of the concept of DIO having kids, what Jotaro was doing between DiU and Stone Ocean, the concept of stand evolutions being unfathomably powerful depending on the obtainment method, new worldbuilding that expands our knowledge of a different part of the world and how it’s been affected by stands, we get the origin of stands and stand arrows, and most importantly more information on how the concept of fate works in JBA. To say that any part is just skippable is to say that any 1/6th of a story made by any great author could be ‘cut out’ and you ‘wouldn’t lose very much’ because by all means, part 5 is very important to our knowledge of the JBA world and foreshadows/explains a lot of future events in part 6 indirectly.
Sogeking is a man of mystery and we may never know his secrets
next video should be a network analysis to try to figure out the true identity of Sogeking
@@notdavidsecondary9958are you stupid? sogeking is sooooo obviously Zoro. His nose is bandaged so much it looks like Usopp’s, and we all know that Zoro is almost always injured and handicapped. And, if you’re wondering about how Sogeking fought alongside Zoro, it’s because he was constantly using the demon thing that triples the amount of bodies he has. Furthermore, we never see Zoro use the demon thing and Sogeking in the same room, therefore there is no proof that Sogeking isn’t Zoro.
damn sogeking unmasked in my very own comment section I cannot believe this
@@notdavidsecondary9958 lmao he can’t escape cold hard analytics😭
He's a menace and he must be stopped .
The graph is exactly what I expected, and also why I love Jojo so much. It isn't like any other show on the list, the main protagonist changes with every single part, and aside from Jotaro, Joseph and Dio, there is almost no character that is present in more than one part. While One Piece always brings back old and forgotten characters and shows the ramifications of the arcs that happened (and I love it), Jojo keeps a small cast per part.
Even recurring characters change role and personality plenty and feel like completely new characters
You forgot to mention the most imprimant character who’s in multiple parts… Speedwagon!
@filipemartinho1753it's similar to pokemon where every generation is a different cast but there are still easter eggs that refer to previous characters
I find it so hilarious that you can just see Joseph and Jotaro standing out on the character graph.
it's probably because jojo's has several protags and a main cast that changes every part so the characters that appear the most usually only do in their own part
I see anime have three Protagonist stucture styles
1. "Main" Protagonist Structure - This classic format centers around one central character, with the story primarily focused on their journey. Side characters may have their episodes, but the narrative revolves around the main protagonist. Examples include One Piece and Dragon Ball.
2. "Ensemble" Protagonist Structure - Here, the story balances its attention across a group of characters. While individual episodes or arcs might spotlight specific characters, no single protagonist dominates the narrative. Examples include Macross and the more recent Ishura.
3. "Shifting" Protagonist Structure - In this format, the main character changes between seasons or arcs. While each phase has a central figure, the overarching story accommodates multiple protagonists over time. Notable examples are JoJo's Bizarre Adventure and Uma Musume.
Of course, hybrid models blur these lines. Durarara!! is largely ensemble-based but also highlights three key characters who drive the plot. Similarly, Teen Titans (the original animated series) operated as an ensemble but gave each season a shifting focus on a particular character. Or the shifting ensemble of Digimon. You might get a different curve with an ensemble show or with one of hybrid styles.
Bleach is more like 1 but with a bit of 2. It is very centered around the MC group, but the other groups do get a lot of spotlight without the main characters, and get their own narratives. Heck in the last arc the MCs are not even present in half of it
isn't one piece basically 1 and 2?
@@diablo.the.cheater I find very common that very common in late stage shonen action anime/manga. The world has become so big, the main character loses primary focus. That might be an interesting thing to research, the exact moment when the MC stops being the MC.
@@Zoro-gk5ow I agree, One Piece has a strong ensemble. But the story generally bounces between Luffy and someone else. Maybe a better example of an MC focus structure.. maybe like Future Boy Conan
@@bennypmo but the thing is that the story of one piece doesn't necessarily revolve around Luffy...sure he's an integral part but the world keeps on moving and changing without much influence from Luffy....soo I guess the first one is correct but it's also not kinda?
I LOVE VISUALIZED DATA ANALYSIS!
More like visualized BETA analysis HAHAHAHAHAGAHAHAHAGA NERDS
This video felt absolutely pointless to me
My favorite part of JoJo is the blood lineage structure as opposed to the continuous Journey and/or tournament.
It's kinda nice to see one story end for the next family member to continue it for them.
Hunter×Hunter would (probably) turn up a very different looking graph because of how each arc introduces new characters that sometimes become POVs, in occasions they even become regulars, and some will pop out of nowhere in a future episode after appearing very briefly earlier. And that's just talking about the anime, since the manga basically changed protagonists in the most recent arc.
Also worth mentioning would probably be World Trigger, which seems to add characters without that much of an introduction and make them regulars (to the point that it's kinda difficult for me to follow the story sometimes because of the massive cast).
Jojo’s is kind of a villain of the week show. In part 3 and 4 enemy stand users usually get like 1-2 episodes before losing and never being seen again. Part 4 has characters show up more, because it’s centered in a town and not a journey. One Piece typically has even minor antagonists appear in several episodes. I think that has a lot to do with why they’re so different.
from part 1 to 6 Dio is still the overarching villain of the series
Jojo 🤝 the pokemon manga
- Manga with weird-ass characters appearances data graph
I think the reason for jojos abnormality here is that jojo is like several different stories that all take place in a shared universe.
characters from previous jojo parts can show up in later ones, but fundimwntally its more a bunch of mid length shows all wrapped under one umbrella.
to put it simply, its an entirely different category than a short anime where all the characters are shown all the time or a long anime where a core group is the focus in a world of characters.
I imagine if you found a series that is also made of multiple stories in one shared universe it would follow a similar trend
try the "neon genisis evangellion", "code geas", or "a certain (magical index, scientific railgun, etc.)" series as they all fit this criteria. maybe they would align with jojo more than the others
2 universes
@@someonehavinganidentitycrisisa bit more if you count the one created by made in heaven, the one re-created after Pucci died and the various that Funny Valentine visits while trying to escape the infinite spin, also 3 more that time they shot Johnny.
@@pepe-zj6cnAlso Bites the Dust works by going to a new timeline, so if you want you can count those too
Is Toaru worth getting into? Watch order looks confusing but the franchise looks kinda fun to pick up on
@@Glatier the light novel is written very smart, but the series as a whole can be chopped up to an action harem with a lot of "lolis". grossed me out.
apparently it gets uber good when the odin arc happens and stays that way, but thats not animated and I dropped it before that.
As someone who started watching/reading One piece when Water7 was getting written and started reading jojo after part 3 was animated...
I really cant decide which is my favourite anime, but they are both in my N1 spot.
They are completely different but for me its like Reading the Odissey and going to see the Parthenon, they are both works of Art but in very different categories and they each give completely different vibes.
I wonder how EXTREMELY long animes would look, even longer than OP, Conan for example idk.
I actually did not know there was a longer anime. I never even thought to check. That's such a good idea, thank you!
Definitely Conan, I don’t think that one has ended
@@notdavidsecondary9958 yeah, OP is only the 16th longest by episode count, with Conan coming in at number 15. Doraemon is either number 1 or numbers 12 (~1407 eps) and 5 (1787) depending on if you want to treat all the adaptations as one long franchise or 3 separate series (the first isn't worth assigning a number to due to being only ~50 segments long).
The symbol of constancy and persistence though is Sazae-San, which has been running weekly since '69 with only around 9 weeks of hiatus (4 in 1975 due to an oil crisis and 5 in 2020 due to obvious reasons). The episode count is ~2771. If it keeps going you'll eventually need to watch 3 eps a week to catch up to it in your lifetime...
Stats taken from Wikipedia.
Conan has to have thousands of characters that show up less than 5 times, right? Can't imagine there would be many mysteries that last more than five episodes (usually it's like two)
@@hiurro I mean, there's a little shy of 700 pages in the "characters" category on the (non-fandom) wiki, so I think it's safe to say that that's probably a good approximation of how many characters there are. Which makes sense if you have a few recurring suspects that reduce the number of new characters per case.
If we assume there's some frequently returning patsies, it's easily possible to have case with no new characters, and then there's the various case-less eps (filler). With only "a little"* over 1000 eps, it makes sense that they're not even three quarters of the way to their first thousand characters.
*If you can call almost 150 eps (or roughly 11-12 cours) "a little"...
A chart to show the logarithmic progression of JoJo characters being drawn in an episode? Like some sort of Log Pose? The rabbit hole gets deeper.
this tool is actually really interesting. i hope i can structurally make models of character relations as well in relation to the plot or arcs. thanks! great vid and analysis.
I... I don't get what the point of this video was.
People love wasting time these days
I think it's just a thought-provoking research based on how important characters to the show really are.
The point isn't exactly informative, but you know its just a silly diagram.
It's up to you if you find it pointless or not.
I think he just likes anime and graphs
Interesting stats
@ eh, that's subjective
That's an insane level of dedication, props to this guy
you should do thiw with manga, since these stories are planned in chpaters not eps. Few of these long anime restructure the story to match the eps.
really nice
1:32 Yea I want to see more of Soge King. A lot of RUclipsrs ignore him and its so weird that nobody is talking about him. Thanks for doing it!
That’s precisely why I love JoJo, it’s not everyday you make 9 unique main characters, have each of them get their bro or sis that fits them and have a story that improves along the way with it
Most shows have a main cast that stays throughout, but JoJo has a new main character every season.
1:22 Oh I was thinking about the Whole Cake to Wano crew splitup, Chopper was with the crew on Whole Cake while Usopp stayed on Wano and sold toad oil lmao
Same deal with Nami and Zoro, thats why he ranks below her
It's certinaly possible, or maybe its a combination of things. I mainly just felt compelled to talk about Sogeking (who ever he is)
@notdavidsecondary9958 Easily the most interesting character, and I'm really saddened that Oda sort of just forgot about him.
I really do wonder WHERE sniper island is though.
1:30 REALEST OBSERVATION EVER
I didn't expect to come across anime Data Mining but it was great to find this video on my recomended
JoJo is super unique when it comes to everything, despite being one of the most commonly immitated or even plagerized anime ever. One thing that makes it unique is the parts structure, which is the main reason it looks so different than the other shows. Even the protagonist, main cast or main villain don't stay constant. Plus, while Naruto, One Piece and Bleach have some villain of the week, it's not as heavily emphasized as in JoJo.
Jojos is immitated? what?? Totally disagree
@javierslytherin9898 how lol there's a reason the everything being a jojo reference is such a common meme
@@javierslytherin9898 Have you never watched another shonen in your life? Without JoJo, all other battle shonen that came after it wouldn't exist. They all copy hamon and stands to some level. There is a reason why so many people said the phrase "is that a JoJo reference?" after seeing JoJo to the point of it becoming a meme.
The parts structure and the main cast changing isn't unique at all, it's been around since Kamen Rider (1971) and Super Sentai (1975). There was also this little-known Super Sentai ripoff called 'Power Rangers' that was really popular in the 90s, way before Jojo's became mainstream in the West.
Edit: Ultraman did it since 1966
Jojo's has basically a timeskip after each part, most returning characters become secundary or mention only to give way to a new roster. No character has appeared in every single part, and part 7-9 are a completely new continuity with close to no recurring characters
When I saw the thumbnail, I legit thought you were going to frame Kolmogorov-Smirnov as a divergence and coin it the "One Piece JoJo Divergence"
originally that is what I was going to do, but I realized it would be a much more interesting video to do that for a larger variety of anime and see if you could do some sort of clustering based off pair-wise distances. This was still KS-distance-esque so I left it as is.
I have read papers that KS distance isn't that great for heavy tail statistics though, will have to think about that
I love when a creator talks about the video already being to long, and it's less than 10 minutes
Think this happens because Jojo has different parts and like only 4 characters actually meaningfully appear in more than one part and never more than 3(yes I know Jotaro also makes a cameo in part 5 and part 2 which would make his count 5 but it's too minor to count) parts so it makes sense it would be drastically different from the others.
The Part 2 Jotaro foreshadowing is anime-only
Also, the fact that JJBA's characters can actually die, unlike in One Piece
@@geschnitztekiste4111 it's actually part of the the manga's volumes. Part 2's final volume has Part 3's first chapter
To me this divergence suggests that any given part of JoJo's operates as a story with a smaller central cast, like the 4 protagonists of Diamond is Unbreakable vs the 10 straw hats
Interesting stuff, further research is a must!
I wouldnt say "its not ideal" that so many long running anime/manga have such a significant overlap in characters vs appearances. I'd argue that's a really interesting quirk of storytelling especially at length. I feel since the other anime you pulled were "similar" long running shows, that the amount of and length of arcs is probably the reason they are so close. I predict that if you tried other "Monster of the week" style anime that they would follow a pattern similar to that of jojos, with a much quicker falloff with a shorter graph.
This series of videos are so cool. Genuinely good work and great finds!
This channel is a rare gem, I love your videos!
This is so cool I love seeing stats visualized
I think the structure of the story is most important for character relationships. Most often we only follow one main cast, which travels to many different locations always meeting new people. These location specific groups rarely interact with each other. But they are often only one or two main character relationships away to any other person in the world.
You should include Legends of the Galactic Heroes (Ginga Eiyu Densetsu), and the reason for Jojo's abnormality is because effectively each 'part' of Jojo with a few exceptions is essentially a new series with a whole new cast of both protagonists and villains, often with only one or two ties to previous parts.
Aside from what a lot of people already said about Jojo's story being separated into very clearly different parts that feel like individual stories, there's also the fact that out of all the other anime you compared, Jojo is the only one to have been adapted when the full manga story was already done (up until what is already adapted, that is) and in the current seasonal format instead of an on going series throughout the whole year. This helps to take it even further away from these other shows, specially One Piece and Pokemon, that were made not only while the manga (games in Pokemon's case) was still going, but a lot of times became almost caught up to it, which lead to either a lot of fillers and/or dragging and pacing issues that Jojo did not had. A Jojo character that appeared in 5 episodes could appear in 20+ episodes if it had from the start the same adaptation format that these other anime had (just an example, but there are a lot of Jojo episodes that adapt about 4 manga chapters, so the math kinda checks out in most cases).
And an unironic fun fact is that even though I just said that One Piece and Pokemon are the most further away from Jojo in this comparison, they also have the the closest story structure to Jojo, in which an arc lasts for a couple years and the next one feels almost like a soft soft reboot of the story (2 softs because they still have more implications and connections to the main overarching story than Jojo, that feel more "rebooty" in this aspect), with the introduction of an entire new cast of characters to that specific setting and plot line and maybe an addition/switch of one of the main crew.
0:36 One Piece Opening 12 on the Tv, my beloved :)
I've just started my research, but I have a theory that many of the big writing issues in the most popular shonen are due to their veneration for and inspiration from Dragon Ball, which was further solidified by the popularity of the 'big three,' either in general or in the western market specifically. It was very cool to see a graph that visualizes the similarities between the character appearances of One piece, Naruto, and Bleach. Thank you!
I think the trend your seeing may come from the average number of characters per episode being relatively stable. A histogram of the characters per episode would show how true this is, along with a moving average of the number of characters per episode over time.
If the number of characters per episode is roughly fixed then we can use that to establish an exchange rate between characters based on the number of episodes they'll appear in. For example, ten characters who appear in ten episodes each would be worth a hundred characters that appear in just one episode each. Put more generally, the number of characters and the number of episodes they appear in should be inversely proportional.
oh that is a super interesting idea. I'll have to think about that tomorrow. I think for the most part if you look at the graph at 1:03 the number of character per episode is mostly constant on the whole (given it spans approx 1100 episodes). I havn't checked it for the other shows but I like that idea!
I like how on the jojo stats you can just tell which one is Jotaro
“Only 50% of a one piece character appearing more than once”
So there is a 50% chance that Batchee is coming back?
The bigger problem here is that it's using data from the anime and not the manga
why is that a problem
@notdavidsecondary9958 I was partly making a joke 😂 but actually, I wouldn't create this graph using the One Piece anime since the pacing is all over the place and all the extended recaps at the beginning of each episode would mean some characters would technically show up in episodes in which the chapter counter part doesn't include them. Also, the manga has "cover stories" that the anime doesn't adapt however, the normal/colour spread cover pages might also confuse the data collection because they often use characters that are not in the current story.
But yeah TLDR: Anime repeats/adds scenes that are not in the equivalent manga chapters which are more concise. But also I was joking 🤣
This difference is likely due to the pretty strict segmentation of Jojo. With only a couple rare exceptions most Jojo characters are only seen within their part/arc, whereas you're much more likely to see a random previous character from an early episode in other shows.
6:32 yes, I do enjoy it
Jojo's curve is evidence of the power of smaller scale, focused stories. When I think about the casts of the other stories, vast quantities of those 1 or 2 appearance characters are forgettable and unimportant. They are mostly background. Jojo being divided into smaller parts that follow different core casts on interconnected, yet smaller stories, make for a long running series where even the characters who only appear once or twice EXTREMELY memorable and impactful. there are many Jojo villains and side characters who I can vividly remember, quote, and recall important scenes and direct impact those 'smaller roles' still had in a massive, long running story. Example: When you (presumably a Jojo fan) hear "N'doul", you probably immediately remember how impactful he was to the plot and how his stand worked, etc. He barely had any screentime, but he MATTERED. The problem with so many long running stories' bloated casts is that there are interesting, potentially compelling characters who will unfortunately fall to the wayside. There are so many characters who barely appear in Jojo but they are so interesting and special that there are fans who would cite those characters as being among their favorites
What you can clearly guess from this visualisation is how big the main group of protagonists is and how well they hold together. Basically the longer the graph needs to drop the bigger the main group. (higher number of characters appear more often)
2:00 We were THIS close… we were on the verge of greatness!
EXTREMELY , Well made video! Thank you and subscribed!
vid looks great bro all these graphs and stuff... wow
That's interesting!! I do think that JoJo having very distinct parts with almost no shared characters is what makes its graph so different. All the other shows tend to have a main cast that occasionally gets expanded, with a bunch of side characters that only appear for a season or less, so that's probably why they look so similar.
I think it would be interesting to quantify what I think makes One Piece special, which is how characters can become relevant again hundreds of chapters after their first arc ends. With your current graph that information is removed: it doesn't matter if a character appears 20 chapters in a row, 10 chapters in a row and then 10 more four arcs later or a single chapter in 20 different arcs. That's shown in the raster, but it's hard to read there, so having a graph showing the maximum/average period of episodes between character appearances after they are introduced would be interesting. I'd expect One Piece to be significantly different there, but I would like to see what the numbers actually say.
All I know is both have an adventure story just done in different ways, and I love them.
Hey this video just popped up randomly for me and while I didnt understand anything you were actually saying (not your fault Im just smoothbrain) it was really cool to look at the graphs for some reason keep it up buddy
I probably would never think about this (despite never actually watching either, but if I did wanted to get into one piece for some reason the newly animated series could make it a short watch)
Nice job.
I find it interesting that Pokémon follow roughly with the other anime, but it makes some sense considering it's nature as a long running anime.
It's interesting in that it has three characters who appear in almost every episode, being Ash, Jessie and James, and if the Pokémon are counter Pikachu and Meowth, Brock would follow in not far behind considering his prominence in the Anime since Indigo, and apparent from Orange Islands, being a major character in every season up till the end of Diamond and Pearl.
It's interesting in that much like One Piece, beloved characters make intermittent reappearances, but for the most part the seasons take a general JoJo like approach in that they have new small groups being the main characters when they switch over too a new season.
You should add fairytail to the list. There is an abundance of characters that make their way in and out of different episodes. I think the graph for it would turn out prety interesting
5:05
Maybe you could normalize the X axis, as % of episodes. since 200 vs 1000+ is quite a difference even in a loglog plot.
I would also be interested in seeing a semi-Log plot where only the y axis is a log plot. I know that would use all the space on characters that appears a lot. But I think that makes sense since that is also how watching the show feels. Characters that appear in nearly every episode is a big part of the show. So fair that they take up a lot of space on the right side of the plot 🤷♂.
4:09 I don't understand the Y axis 10^0=1 that much I get. But 7 appearances 50% that would be about 10^(-0.3)~0.5. I guess that could be -0.3. So the Y axis is % of characters if I got it right. So the Y axis is already normalized to a value between 0 and 1.
It would be nice with ticks, minor ticks and maybe even a grid. It would make the plot easier to read at least for me.
0:27
This plot mush be normalized somehow. Cause one piece have many more episodes than JoJo. I also would have thought one piece had way more characters, but on the plot it looks like each character take op the same width. Maybe it is just the first episodes of of one piece on top of all the episodes of jojo. (am I confused?)
It is interesting that jojo stands out. I have not watched the show. But it looks like the show changes main characters as the show goes on.
I guess the others looks very close due to the loglog plot 🤷♂.
Or maybe shows have a budget for a number of voice actors thus they limit the number of new characters in an episode and the number of recurring characters in a season. And that is why all the other show look so similar.
It would interesting to not count episodes but minuts on screen, although that data would be much harder to obtain. Guess that could be a job for a AI in the future.
Or maybe a system like the once they use for sports data like football where they record every contact with the ball, every meter every player ran etc.
great points (as always) I'll reply point by point lol
1) I had thought of this, and I think it is something I'll think about more but the issue that I had was that it makes the graphs even more confusing to read. Here they all start at 1, but if you normalize by the total number of episodes, each graph starts at 1/NumberOfEpisodes which is not the same for each show. This made the graph extremely confusing without more explination. However, nothing fundamentally changed.
2) semi-log could have been good to check. Admittedly that was sort of force-of-habit. In my IRL work I use log-log all the time and so here it was just the first thing that came to mind.
3) Your're correct on the y axis stuff. The lack of tick marks was an annoying thing that happened when I tried to import the .svgs from python to blender. I need to look into how to do that better in the future. I tried to mark them using those little horizontal and vertical bars to compensate, but I agree some tick marks would have been better.
4) Oh yeah here the actual vertical height is technically meaningless. I mainly just wanted to be able to show the overall structure, but if I was to plot them so that the dots were the same in both shows, you'd be barely able to see Jojo because of how large one piece is. I could have done a compromise to animate jojo from small to large to make that explicit.
5)regarding your last point -- thats something I've been thinking about quite a lot. Maybe there could be some feature recognition one could do but im not sure. I know in some arcs too the characters in one piece also change their clothing which would make it hard for an AI to track them. I think its a super interesting problem but looking into it would also be wayyy more effort than I can afford right now haha. I'm just having a lot of fun doing these little side videos right now during the holiday break lol
@@notdavidsecondary9958I agree with OP's first point, since if you have an anime with say only 12 episodes, its plot would dive down earlier than the plot for JoJo, if the horizontal axis is the number of episodes. If you divide by the number of episodes, the horizontal axis will simply become the proportion (or percentage) of episodes the character appears in. The leftmost point of the plot would be different depending on the anime, but that doesn't matter (or you could connect the leftmost end of the plot to the vertical axis with a horizontal segment, which might be clearer; just a thought though)
@@notdavidsecondary9958 As always on this channel I get a much better answer that I had any right to get. 🙏
1)
It is true that for the first one it might be more confusing. And it might also end up just showing differences in show length even more, I can't really imagine how it will look. But I think it might make it easier to see if a show have many repeating characters or many new characters compared to another show. No matter the length of the show.
2)
I always just get straight lines when I am using loglog plots. And I am addicted to plots. So if I already have the data in python I often make both linear, semi-log's and loglog. Just to see the data from different angles. I often also look at the derivative or a cumulative plot. They are all easy and fast to make. So I might as well 😂.
3)
That is a good excuse. I have not tried importing matplotlib plots into blender. I would probably just have save the plot as a high-res PNG pic. (Cause that is what I normally do). And then just imported the PNG on a surface. But now I want to try to import a svg file. (I guess you cannot import a pdf into blender, can also be vector based). I have never tried importing vector graphics into blender, but I have often had problems with importing vector graphics in other programs. 😬
5)
Yes tracking characters from video would be a huge undertaking. Maybe it is possible to use the subtitles... But often the subtitles does not state who is talking. That would have been super helpful. But still not perfect of anything. But manually going through every scene would be quite an undertaking. 😂
But well this is always the problem with data, there are always some details missing. 😂
@@notdavidsecondary9958 Also, the algorithm loves this video. It took less than 24 hours to overtake the "Comparing Massive Anime Character (and Brain) Networks" 🤷♂
@@Petch85 Good questions are hard to write, they deserve good answers. Just to comment on your point 3 -- the reason to use an svg over a transparent png is that you can animate the datapoints in it (e.g., when the curves appear, or when the dots are made bigger and stuff like that). Some of that is possible with a png with some clever cheats, but it limits you a lot. The big issue is that it imports everything as a curve without a profile, and curves without profiles don't render. So for the tick marks I would have to go to each tick mark and assign them a profile, and as far as I can tell you can't link curve profiles like you can material (though now that I think about it I should double check that...).
But also yeah, I agree its been interesting seeing the algorithm do its thing so differently. I'm actually a little worried because this was genuinely meant to be just as a side channel where I can upload random quick-to-make stuff, but now the audience is very different to the main channel. That does give me an idea for a future follow up to the 'youtube subscriber network' video on the main channel though...
dude, this is awesome bro,
yeah i wonder who sogeking is, but it doesn't matter, he is the hero SHP need and always inside your heart
Definitely interested. My theory was initially that it came down to the continuous arcs vs villain of the week structure you mentioned but Pokemon as I remember it was more a of the week sort of show than a shonen arc type show and it is pretty close to One Piece. Then I thought about it more and you mentioned the generational divide structure of JoJo also makes a compartmentalized raster and network diagram, so maybe it has to do with JoJo's kind of unique generational structure. But Pokemon also has generations though not as harshly divided and once again it's not even the closest to JoJo. As for anime I'd suggest, maybe YuYu Hakusho or Yugioh. Not an anime but I'd love to see it and it'd be as big or bigger than Pokemon and One Piece, I'd love to see this done for Monday Night RAW.
Actually you hit on something I want to talk about for pokemon and this would be a great opportunity to test something -- how much of the pokemon anime did watch? Like mainly early pokemon, or more the newer stuff? Or the whole thing perhaps?
@notdavidsecondary9958 I watched through the Gens 1 - 3 stuff as a kid. I would then peek in every now and then to see what's up with new moves. The structure seemed to be Ash traveling across the region encountering new places, people, and regions until Alola where the structure shifts to him staying at the school with a larger cast of consistent characters with little excursions.
@@jarekguevara6563 That's perfect, thank you so much for replying and letting me know :)
Yeah my experience was more or less the same (I think I fell off around the start of gen3). And yeah exactly, there is this very clear shift from the original one-off characters format to a larger more consistent cast in later seasons, and that's why it follows one piece so well. I'm hoping to do a video on that in the relatively near future.
A hypothesis: if you were to perform the conversion of each individual JoJo arc to this appearance table it would also follow the same trend albeit at a smaller scale as total appearances would be lower. I would also posit that this trend would even exist outside of anime, perhaps it would be worth pulling samples from popular western shows with similar run lengths like House, Supernatural, etc. with a diverse enough data set we may see an overarching trend amongst all shows that have a core group with supporting characters that may only ever be in a few episodes.
very much agree, I suspect the same would happen. I think House would be an interesting one to compare against. I wasn't a devout watcher but I've seen some here and there and from what I remmeber it's got a core cast that appears almost every episode and then the patients which are almost always one-offs. That's actually why I included pokemon thinking it'd be the same but in the later seasons they really changed up that format, at least from what the raster shows.
You should use the "physics" method. First make a list of all the variables that are important, for example "number of chapters" "number of characters" and "total appearances". And then analyze separately how each of these things should affect the final value. Should the number of chapters positively or negatively affect the final value? Should the number of characters make the final number bigger or smaller?
Finally, you multiply all the values that should increase the final value, and divide them by those that should decrease the final value. This gives you a basic formula with a high probability of being precise (although not exact).
1:42
Sogeking was legendary...
I miss this fella.
we gotta compare "villain of the week" anime to long-running arc-based ones.
This video is a pretty good example of how easy it is to misrepresent things with statistics, just by organizing a graph in a way that muddies our understanding of the data
@@Lazypackmule i’d love if you could elaborate- in what way do you feel the data is being misrepresented?
@@notdavidsecondary9958 I'm referring to how many iterations you went through to represent things in a way that made the contrast apparent
@@Lazypackmule The number of itterations… as in… one? Changing the histogram to a cumulative distribution function to address the ways in which a histogram misrepresents heavy tailed data?
I guess I’m curious what your remedy to this would be then - Leave it as the misleading thing?
@@notdavidsecondary9958 I have no idea why you're so insistent on interpreting this as some kind of criticism against you
Apologies, the tone of my comment was perhaps not the best reading it back. I am genuinely asking out of a sense of wanting to understand how it is a misrepresentation. If your comment had been "this was not presented well" then I'd be like, okay yeah I can see that, CDFs can be unintuitive and I'll work to fix that for the future. However, your comment was that this was actively misrepresenting the data, and that is what I was hoping to understand. Typically the step (iteration?) of transforming a histogram to a CDF is not viewed as a misrepresentation of the data (in fact, often the reverse would be more true because of how poorly the histogram represents heavy-tailed data).
yeah changing the main caste every other volume along with supporting cast will change how many appearances characters get
Caste System?
Jojo is like 9 diffrent series in a threnchcoat. hell starting with part 7 the series isnt even set in the same canon anymore. Get the number of apperances of Baruto characters and Combine this list into a Baruto+Naruto Set and youll see something closer to the JoJo line then the standard trend. Love this video. Try adding smaller series like Death note to see if the data changes depending on series length. It might also be a shonen trend exsclusively, sailor moon would be a good example to test that against.
Im curious why you think my data is "all weird"? I purposefully chose it for the contrast so I'm not sure I understand the comment. Also regarding the shonen comment -- that's a great hypothesis and that's what I thought too, but that is why Pokemon was included
@@notdavidsecondary9958if jojo is different because it is a combination of multiple stories featuring their own characters, then if you combine a few unrelated animes, would you get similar results as jojo?
@@koktszfung id say a better comparison would be a whole bunch of stories from a shared universe that has some mild crossover
@@notdavidsecondary9958 im not critisizing you, i like your data. Im just trying to hypothesise on the data. Sorry if that wasnt clear. This is a great video.
@@inverted_paradox4170 haha no sorry, I did not in any way read your comment as criticism, tone just doesn't carry over well over youtube comments. It was very clear you were giving genuine ideas/ feedback :). I was just curious for clarification as maybe I had missed something that was obvious.
I mean one thing that sticks out to me is how the graphs dont have the same max number of appearances, with jojo having the least, so of course it ends earlier
you could probably just do something along the lines of what you did for the number of characters and theyd be even more similar
Nice video! Though I wonder how did you count the Pokemon themselves in the Pokemon character list? Do all appearences of a species count as the same character even though they may or may not be different? If not: do you treat legendary Pokémon differently (as lore may imply that they are unique)? Looking forward to future work!
Yeah that's a great question. The way the code is currently structured, I believe every pokemon is just counted as the 'generic' version of it. i.e., 'ash's pikachu' would just count as every 'pikachu'. Bulbapedia does often denote unique versions in parenthesis, so I believe it should be fixable and if I do make that follow up looking at the pokemon raster that is something I'd fix. For the question about the legendaries, that one im not sure -- it really depends on how the fan wiki stores that information.
@@notdavidsecondary9958 Interesting, thanks for the answer. Sorry if this is already explained in another video, but how do you actually get the data? Webscraping from the wiki?
effectively yeah, I use the beautifulsoup python module. Some wiki's (e.g., wikipedia) let you download the entire thing, in which case I'll do that so as to not stress the host server, though if that isn't an option I use plenty of delay timers to slow the program down.
Amazing work! Really want to see more of this. Bell ringend
I think what this shows is that one piece, more or less, established the normal distribution for anime given its length and age. Most anime will begin and end in the time One piece has existed, so it sets the bar for Shonen anime and Manga to "follow".
Honestly, since you hinted at it, I'm now really curious about how different genres compare, especially since you hypothesized that it may be the nature of this genre/demographic/longrunning nature of the shows you compared
seeing which other genres' graphs JoJo's is more similar to might highlight something about it
Dude this is awesome
Glad you enjoyed it :)
So what you are saying is that jojo is…… Bizarre
50% of characters appearing for only 1% of the show to me does kinda show that maybe One Piece has a bit of a problem with drawing things out too much to the point where there's enough time for some characters to be introduced only to be never seen again almost immediately afterwards.
mimics real life though, you can meet someone on the street and have a lovely conversation and never see them again, i think it's a service to enrich the world, also this data comes from the anime which it shouldnt but too late i guess
@@ultratronger it's a fictitious story though. If I wanted a story that gave me the experience of real life I'd go outside or read a biography or something.
@sarbe6625 what story gives you the feel that the world is alive and breathes and feels so full of life and it's always moving even when you're not looking like One Piece? Certainly none in manga. It's easier to compare it to something like A Song of Ice and Fire.
jotaro and joseph are the only ones who show up in 3+ parts so i think it makes sense why the graph is so different
I just started cheering out loud when the KL divergence appeared on screen oh my GOD!!!!
The Jojo divergence probably comes from the fact that fewer of the main and supporting characters consistently across the show due to the "Part" structure
cool analysis 👍
It might be useful to take this idea and use it to compare genres of shows. Something like Detective Conan, while it would have more episodes, is a lot less serial and plot driven iirc. Doraemon might be a good comparison with it.
I wonder what would be revealed for other shows out there, like sitcoms? Seinfeld versus Friends we might see a similar curve, but what about something like Days of Our Lives the longest running daytime TV drama? The soap operas? Those things run long enough on human life scales that characters literally grow older and have children through the series as a staple for the longest running ones.
:Turns abstract understanding of anime structure into data analysis question:
"I have carefully set this up to be sure I'm asking a question that makes sense."
:Sees data analysis:
"I have no idea what this means."
Been there, buddy. In this case, it might be interesting to take one step back to the log/log plot and compare each show on that. The CDP approach is useful for *finding* these kinds of divergences, but not especially helpful in drilling down.
My theory is that there aren't any characters in Jojo who persist through the entire narrative, so you get this stack of "casts" from each arc that are there for that entire part of the story and then just disappear forever. The other shows slowly expand the main cast so you get a more gradual termination, lifting the entire line up a bit and extending closer to zero. You can at least see *that* in your comparison.
You know, I really hope sogeking shows up again. Maybe he'll be a major player in the Elbaf arc?
fingers crossed. Could you imagine if Sogeking ends up joining the strawhats??
Legend of the Galactic Heroes(1988) might be a good point of comparison because the cast is pretty evenly split between both sides over 110 episodes. So it might be interesting to see whether there's a bias between either side.
Ranma 1/2 or Avatar the last Airbender could also be interesting pics to see whether other genres fit the shounen trend.
Another anime that might be worth checking out is Gundam, especially the first Gundam series - while a lot of villain of the week stuff does exist at the start of the very first run of Gundam, even that early, Char and a lot of the higher Zion generals are introduced early, though the show does use them wisely and sparingly. I'd be very interested to see what its stats look like
JoJo's is, more or less, a world, and each part within is it's own story, tied loosely to another by a couple strings and the place it is charted upon, unlike most other series, JoJo's is practically an anthology, explaining it's divergence for the norm, especially with SBR and ahead being essentially on a whole different plane.
I'd really love to see this become a series
Instead of having total number of appearances, why not try percentage of episodes the character shows up in, this will make the list more standardised instead of having lines that end in different places. It would also mean you don't have to log the x axis and it will work better for shows with varying numbers of episodes.
I did try this, it doesn't change anything. Actually it just makes the plots much harder to read as they no longer allign anywhere. The smallest value would no longer be 1 episode, but rather 1/N where N is the length of the anime, which is different for each. This wouldn't be a problem if, say, each anime had a character who appeared in every episode, but that isn't the case for any of them, and especially not the case for jojo. Moreover, since this is just a uniform scaling, it doesn't prevent the need for log axis, it just shifts it from the domain.
You gained a new subscriber! Very interesting!
A long running anime that is different (and underrated) is monster a psychological thriller.
1:38 if he didn't show up the series will be completely different thank you Sogeking wherever you are
So your saying that Usopp is Sogeking? What a wild theory buddy.
It would be interesting to see how other genres of anime stack up. Like adding ranma 1/2, sailor moon, and card captor sakura would be interesting shoujo picks. Detective Conan or Sgt. Frog would be good for a childrens anime. You could also compare it to western anime-inspired shows like ben 10 and avatar
I connject that Jojo looks so different because it regularly ditches it's cast. Luffy and Goku and Ash never go away, but . I predict that Digimon will closely match Jojo, if not be more extreme because across the 12 seasons, there are 8 separate continuities, that exactly 2 characters have crossed (In the show, more have in the manga and games)
Ash is actually gone, there are new protagonists even if only for a year yet.
Visualizing stats are always so cool
now this is quality content
The most interesting would be to see what transformations would let jojo look like one piece
3:36 Why not? That perfectly exemplifies Zipf's law, a linear regression with slope ≈ -1 !
It also predicts word frequency, city sizes, popularity of chess openings, etc. There is a Vsauce video about Zipf's law mysterious appearances, you should look into that
Apologies I'm kind of confused by the "why not?", I give a few reasons for why it is not a good representation. Was there something there you disagree with?
Regardless though, Zipf distributions are really cool! and I was going to talk about them in this video but they are also very tricky and you need to be careful in seeing Zipf where there actually isn't any. If you're interested to read it, I'll also include my reply to another similar comment here:
I’ll caution against jumping to conclusions and thinking its zipf (actually this is not zipf but rather something called power law), especially when plotted in that form (what you want to do is logarithmic binning).
To conclude that it is "zipf", you need to perform maximum likelihood estimation to get the exponent (caution: Linear regression to get the slope is *really bad* for this do not use it), calculate its pvalue and then alternate hypothesis testing against other distributions (see Clauset 2007). I haven’t done any of these so it's premature to tell, but keep in mind that the cdf of a powerlaw is itself a powerlaw and our CDFs are… doubioisly powerlaw (which is not to say they cant be powerlaw, cdfs can over exaggerate differences especially when in log log scale).
@@notdavidsecondary9958 Yeah, I agree it is too early to declare Zipf's law; any straight line on a log-log graph boils down to a power law. But extrapolating a line through the first ~20 points (where the bulk of sporadic characters are) crosses axis at around ~(1, 10^2.5) and ~(10^2.5,1) i.e. slope -1, which is suspiciously Zipf-like.
I don't have any problems with using a CDF, as integration does help smooth out noise. But using the cumulative obscured a bit that connection.
Also I was half expecting for no one to read my comment, I'm glad to be heard
It makes perfect sense why it is that way. Luffy, Naruto, Ichigo, Goku and Ash all are THE MC of the show. They have a few characters that stick around, some longer some shorter, but the core of the show's crew rarely changes. Jojo's on the other hand has Parts, meaning that there really is not singular MC like that (tho the Joestar Bloodline does connect the Parts, as well as Dio who has "something to do" with every single villain/antagonist)
Id say the show thats probably the most opposite to this (i guess) is Cowboy Bebop. That being said the Show is also very short lol