On the fourth day of December, Scott's Thoughts gave to me: A tiny turtle's bubbles The fire starter digging A garden lizard's leech seed And those three weirdo mons punching
I have a confession to make. I obviously was too young, and I didn't know WHY moves with 100% accuracy sometimes missed in Gen 1. Now, I know that it's because of an oversight that is exclusive to Gen RBY. I didn't know that it wasn't a thing that was still in the games until I started watching these solo-challenges in 2020. I played 7 generations of Pokemon, I played Smogon competitive pretty heavily during Gen 4, I dabbled in VGC during Gen 6. That entire time I always thought to myself, while using a 100% accurate move, and not having my accuracy lowered, or my opponents evasion raised in any way, I still would always think to myself "I hope that this doesn't miss." I assumed that whole time, that no move, aside from Swift, Aerial Ace, etc, was actually 100% accurate. I never once verbalized this to someone, so that they might correct me. I never once looked it up to see if it was still in the game. I never noticed that it hadn't happened to me in like... 20 years. Nope, I just assumed that it was an intended mechanic designed to "spice things up" just like tripping in Brawl or something. From 1999 all the way until 2020. That's my story.
Rest assured that you were not alone, I also fell victim to thinking no move was actually 100% accurate unless it explicitly said "does not miss", like Aerial Ace. It started in Pokemon Blue and it wasn't until a year or two ago, watching some random YT video that explained the gen 1 spaghetti code
Charmander lover here! I think its fair for squirtle to be ranked higher, just as you said its more consistent. That being said, this is way closer than i thought! Thanks for the amazing runs!
I was so scared that charmander would be ranked higher than my favorite little turtle. But I'm definitely happy that all 3 starters put in good performances. The charmander line gets so much love everywhere, so it's nice to see the squirtle squad get small wins over it every now and then.
@@TheWispGuy I mean, Eevee is more of a legitimate starter than Pikachu, lore-wise. Pikachu's only appearances as a starter are all due to your rival getting impatient and taking the Eevee that you were originally meant to have, whereas Eevee was always going to be your starter in XD Gale of Darkness.
For anyone wondering, the odds of that reset in Erica's gym (at 18:50) are very low, and Scott got incredibly unlucky. For Bellsprout to win the battle, Ice Beam needs to miss (1/256). 0.3% Then, Stun Spore needs to hit. Stun Spore has 75% accuracy, and the enemy AI in Gen 1 has a 75% debuff on hitting status moves, so this is 75% * 75%, which is 56.25%. 0.3% * 56.25% = 0.22% Wrap does 2 damage per hit and Squirtle has 120 HP. Wrap has a 15% chance to hit 0 times, a 37.5% chance each to hit 2 or 3 times, and a 12.5% chance each to hit 4 or 5 times. On average, Wrap will hit 2.55 times per accurracy roll and deal 5.1 damage to Squirtle, again, on average. So it needs to hit 24 times in a row on average, or for Paralyse to proc when it misses (25%), or for Ice Beam to miss when it misses (0.3%). Since Wrap hitting is more likely than those other outcomes, I'll estimate an upper bound and just go with that probability. 85% chance for Wrap to hit, 24 times in a row, is (17/20)^24 = 2.02%. Multiply that by the 0.22% chance for Ice Beam to miss and Stun Spore to hit, and my upper bound estimate for how likely Bellsprout to win this battle is 0.0045%, less than half a pecent of a percent. One in over 22,000. Note: Bellsprout has a 7.81% chance to crit in Gen 1. Crits do less than double damage but as Scott always notes, multi-hit moves continue to crit if any of the hits crit, which compounds the additional damage. I didn't factor crits in because I'm not that smart but I imagine I'm less than an order of magnitude off. The real shocker for me is that there's a 2% chance to hit 85% odds 24 times in a row. That's far more likely than I would have guessed.
Damn. JRose recently had a string of crazy luck in his favour on a one of his runs. I forget the sequence but it was also astronomical. This has to beat that out though. My benchmark for improbability will always be when someone pulled four Bombs in a row as Peach (1/256 x 1/256 x 1/256 x 1/256) but it's crazy just what kind of insane events that pokemon players also generate. I guess it's like 1000 monkeys writing at 1000 typewriters.
@@josephnation9063True, I neglected this. I'll amend my calculations. 15% chance to hit zero times, 37.5% to hit twice, 37.% to hit three times, 12.5% four hits, 12.5% five hits. This averages out to 2.55 hits per accuracy roll - given that each hit is 2 damage, each accurracy roll deals an average of 5.1 damage, so it only needs to hit 24 times on average. 85% 24 times is just over 2%. Multiply by 0.22% and that's 0.004451197844%, less than half a percent of a percent. One in over 22,000. Rather a large difference that makes 😅
As a charmander main i think you did great, and the fact that chamander beat either of the other pokemon in any metric is something to be celebrated! I feel like online discourse makes it sound like charmander is the worst and that your gonna have a terrible time if you use it him. turns out, nah. if you use him, sure your gym fights will be hard, but you're gonna make up for that with all the the trainers that you can burn you way through. the fact that you only really struggled on one pokemon says A LOT about charmander's viability. third place is third place and we all knew that we were gonna be here, but we put up a great fight.
So when talking about nuzlocking it's horrible which is probably a third of the community. The other 2 thirds are people that say just have fun but if you don't pick what they think is best you are going to hell.
The times between the 3 turned out pretty much how I expected, Sleep is just to busted in Gen 1 especially when you have setup moves. Also feels good for Bulba to win since it's my favorite mon.
I'm so happy to see how these runs turned out!! The optimization was fantastic and it was great to watch. Seeing first stage Pokemon able to push into these ranges is really cool to see.
18:3618:46 shout out to the bellsprout missing Wrap only 2 times, and both of those turns squirtle gets fully paralyzed, what an absolute trooper of a plant they are 😅
I'm happy for the turtoise to get it's video and benchmark, would've love for it to be the best around, but sometimes just beating the Char line is good enough for me. Also, kudos to the Saur line, they deserve to be number 1 at something!, also Ivysaur is the best middle stage, but we are not ready to get that convo started.
I really enjoy the widmanstatten pattern in the bottom right, it’s a special crystalization that happens to iron minerals in the absence of oxygen, exclusively in meteorites. An excelent way to tell if a rock you’ve found formed in space
As a charmander fan, I definitely don’t have an ill will towards your ranking it lower. In fact, I’m super impressed you got it to do as well as it did. And I’m super proud of charmander too ❤
Sraring at squirtle's back sprite throughout this video, the all blue palette makes its shell look so much like crystal and making me imagine it as an ice type.
As a Charmander stan, I have to admit that Squirtle is way more consistent. If anything, I was surprised the final times were so close. I weighed Withdraw too heavily and didn't really account for Sleep Powder or Swords Dance in my Squirtle>>Charmander>Bulbasaur prediction. I really enjoyed this series! Hopefully the middle evolutions and reruns of final forms are coming soon!
That fight in Erika's gym was so painful to watch. Poor Squirtle deserved better for its first run, there was so much awful luck for it. At least it got a good final time. =)
Out of all the starters, i don't actually have a preference I assumed Squirtle or Bulbasaur would be the best. Looks like Bulbasaur took it. Which makes sense in the spirit of the game. So many playthroughs though, I gotta give you props for that.
A fan favorite. For good reason too, such an incredible pokemon. And fun to use, too! I think it's gonna do a pretty good job. I could see it beating charmander
I think that a good way to weigh in the attempts per Pokémon would be to take them as a time average. i.e. (sum of attempt times)/(number of attempts), which would show the Pokémon's consistency in achieving a given time, good luck outliers would balance out with poor luck outliers. Anyways, love your videos a ton and your data collection, I am truly excited for this month of thoughts.
Can we just talk about the fact that freakin Squirtle can learn DIG but Bulbasaur a Pokémon literally meant to be bulb as in plant and dinosaur a Pokémon all about being buried in the ground cannot learn DIG. Good old Pokémon logic
Somewhere in the time of Squirtle being paralyzed and wrapped to death by that Bellsprout, I noticed that the Bellsprout actually does miss at least 1 wrap, it just so happened that on the same turn, Squirtle was immobilized due to paralysis. Just another pain-layer on the giant pain-onion.
Loved the starter mini series! No matter the results I'm here for squirtle. Also I predict that ursaring will destroy donphan in the vs video tomorrow.
I am a charmander die hard. That said I don't care that he underperformed. I don't love him because he is the best, I love him because he is awesome. Plus C tier is green and that's my favorite color. He wins in my eyes.
Damn straight. Imo, it has the perfect mix of cool and cute, can learn cool moves and it is orange. I love all three starters, but I will always be a Charmander fan!
Bulbasaur was exactly as expected ; slow to win fights but never really struggles and just consistently gets through the game. Charmander surprised me; with a decent amount of extra training it can muscle through everything up to Lance, which is just a roadblock and all that extra training hobbles it. Squirtle just fucking wrecks the whole game.
Incoming Wall of Text: For incorporating the number of runs, at the baseline, my first thought is some kind of penalty function with the runs and time as input which generates a _third_ tier list. Two ways I can see the argument going: 1. An exponential or logarithmic penalty function, something like... `adjusted_time = final_time_seconds * f(runs - 2); // Subtract because there's ALWAYS at least two runs` ...where `f(0)` is 1.0 and by the time you get to `f(10)` it's something like 1.25 or even 1.5. You might argue it should be `f(run - percentile(25, all_run_counts[]))` or even `f(runs - median(all_run_counts[]))` under the assumption that maybe, say, Bulbasaur shouldn't be punished harshly because it took four whole runs. That's not implausible; do tests until you think it matches how you feel. You know your dataset and what you're trying to make it confess to! (Clamp your lower bound to 1.0 in that case unless you want things that only do 2-3 runs to get _bonuses...)_ 2. Bucket it with more granularity and apply flat penalties. Time or Tier or whatever your gut says. This would require a lot more "gut feeling" and is harder to justify than misusing algebra to throw a fit, but it'd still probably be interesting. 2a. Actually, there are some possible interesting visualisations you can generate relatively simply to start getting a better understanding of the data: - Bucket run count by the the number of tiers, then order by time within that tier as a very simple example. - Then, yet another tier list can be generated by penalising the original final time by half of the tier range as calculated by the run count tier (e.g. S-tier time but D-tier attempts? That's a five minute penalty.) These kinds of _ad hoc_ analysis can sometimes be helpful clues for finding useful trends. Come to think of it, I wonder what the distribution of attempt counts is? Is it normal? Flat? Bimodal? Something weird? That can help inform your approach. Another thing to keep in mind is how you figure rerankings into it and the simple matter of your own skill improving. Your evolution as a player has pushed times lower on replays, which shifts all of the above metrics in weird ways. So you'd be forced to either make arbitrary decisions about cutoffs for pentalty thresholds or come up with some kind of metric for your own skill level. Actually, thinking about it from that angle... I suspect the _"proper"_ solution would involve going *much* deeper into the actual run data. See, the number of runs to achieve a given result is a measurement that arises from the compound probabiity of all output randomness across the run. There are intangibles, of course -- hesitation, menuing, bonks, Churro, _et al._ -- but your raw play logs can probably be leveraged to to get you close to something like a "luck/consistency factor". *In theory,* this is general enough that it doesn't matter what your skill level is at the time because the amount of luck you need for a time is reduced by better stats/moves and better decisions around e.g. XP routing and move learning. Another neat thing is the calculation can be applied to _every_ run, which can give inroads to a better sense of "luck-per-time" crossover points, what consistency looks like numerically, and even more concrete effects of your own improvement. I realise it'd be a *lot* of time and effort to implement (I suspect there a fair bit of opportunity to trade complexity for accuracy, but even still), but I think it's something worth _thinking about_ even if it can't happen/get farmed out right now.
Super interesting. My knowledge of stats is limited (bio b.s. stats and one semester of graduate level ecological stats) but I'd love to see some more in-depth analyses of these runs by someone who's more well-versed
Interesting thoughts. Penalizing by run count is difficult or impossible to do well because the best run may only be slightly better than the second best but with way more attempts. Both your suggestions suffer from this problem. You need to take into account (1) the raw run count and (2) the actual value generated by the extra runs.
@@fangiscool1 Yeah, that's part of why I came to the conclusion I did at the end. It's one of those situations where it IS an easily-collected metric that exists and it can, at times, somewhat tell you something about reality, but only weakly and often deceptively. (And good luck convincing middle managers of the need to invest in better... ヘ（。□°）ヘ)
It would be really interesting to see what time Bulbasaur (and others) would have without sleep. Would love to see a comparison as you did with Butterfree a really long time ago.
This is interesting when compared to the final evolution run. Perhaps the evolution stat changes have a bit to do with it. Am just guessing. I had some thoughts for naming trainers in yellow, so hope you like them Misty’s gym; the mandatory trainer with Goldeen could be named Meena, which is a Sanskrit name for “fish” and “precious stone”. Nina and Ninette are both Russian names meaning “fish” and “little girl” (also “fire”, but 2/3 ain’t bad). Personally, I like Ninette for her. Brock’s gym; I found a good name for the Lightyears trainer; Braxton, which literally means “Brock’s Town”. I found this while looking for a name for the self-destructing hiker. There are a lot of common and unusual names meaning rock, so couldn’t really choose. Hope you like.
I've done speedruns with all of them (although with evolutions), and I'd come to a similar conclusion. Venusaur has the best potential, but RNG can make it just a nightmare. Charizard was my favorite to use, although early/early-mid game is boring with such weak (Ember's 40 BP for so long is bad) or inaccurate (Mega Punch) moves for so long, but it's the de facto best late game IMO, Swords Dance with those stats is just incredible. Blastoise... it's consistent, but the game goes to a glacial pace against Water-type trainers like Lorelei. Also it's just stat-starved so badly. Great movepool though. Awesome runs.
Charizard was always my least favorite to use, mostly just because of how painful Lance is. But there are new ideas around to solve him, so I’m excited to try the Fire-type again.
@@ScottsThoughtsPokemon Lance is a hard obstacle for any Pokemon not able to tank Hydro Pump well sadly. In Yellow he's the best E4 member I think, unless you're weak to Ground-type attacks.
This is wild, i feel like we never see scott buy ice beam from the little girl. Really pissed me off that she only gives out useless moves in lets go Pikachu/Eevee
Nice to see you figured out a way to make Koga easier in the 2nd Playthrough with Mimic. I wonder though with that strategy, would Sleep Powder be better so you can get by Koga at a lower level?
I'm a little sad squirtle didn't place first, but I'm going to say sleep powder gave bulbasaur an advantage. As unevolved pomemon they have a harder time outspeeding or getting one hits so cheese strats make a big difference and squirtle's moves are just too straightforward.
Skull Bash is so disappointing, because I remember using that in Gen 1, and I was like "Why wouldn't I just use Strength of Double-Edge instead of this?", not to mention that this was the only time Skull Bash was ever a TM, so I just always wrote it off as a less-useful version of Strength.
I really enjoyed this comparison. Your result mimics the advice from a game guide I had as a child :). Question though - have you started mixing your audio differently recently? I watch UA-cam at 1.5x and your voice has recently gained some audio clipping (within the last couple of months). I know you aren't mixing for people playing at 1.5x but I was just curious if it was a change you'd chosen to make. I remember you had some vocal health issues recently, so maybe it's because your voice is lower when you talk in more recent videos. I'm a big fan of your videos either way :) thanks for making such great run videos.
Very interesting. I'll let Scott answer when he gets to this comment, but I think you're right. I'm pretty sure I heard him mention on a stream that he's had to be more intentional with the way he speaks and it's resulting in a much calmer sounding voice, which I'll take is probably lower, too.
I think, it didn't matter but is there a reason why you used 7 rare candies before Koga to Level 57 and right after 2 for Blaine so the Exp from Koga didn't count. Did you want to prevent a possible Mid battle level Up against Sabrina after that or was there another or just no reason to not use all 9 rare candies at once?
I was surprised that Squirtle didn't win because speedrunners used to use Blastoise before Nidoking became the new tech, but then I realized it's because they use items. Btw, you mentioned fissure could ko Squirtle, but it can't. Fissure can't hit mons higher level than it.
Talking about the dig glitch when fighting Agatha made me think of something. Does removing the paralysis remove that state if you were to heal with a full restore?
On the fourth day of December, Scott's Thoughts gave to me:
A tiny turtle's bubbles
The fire starter digging
A garden lizard's leech seed
And those three weirdo mons punching
5 Legendaries!?!
Gotta find this after every video!
I will be looking forward to this comment in every video all month
if you don't comment on tomorrow's video we riot
I have a confession to make. I obviously was too young, and I didn't know WHY moves with 100% accuracy sometimes missed in Gen 1. Now, I know that it's because of an oversight that is exclusive to Gen RBY. I didn't know that it wasn't a thing that was still in the games until I started watching these solo-challenges in 2020. I played 7 generations of Pokemon, I played Smogon competitive pretty heavily during Gen 4, I dabbled in VGC during Gen 6. That entire time I always thought to myself, while using a 100% accurate move, and not having my accuracy lowered, or my opponents evasion raised in any way, I still would always think to myself "I hope that this doesn't miss." I assumed that whole time, that no move, aside from Swift, Aerial Ace, etc, was actually 100% accurate.
I never once verbalized this to someone, so that they might correct me. I never once looked it up to see if it was still in the game. I never noticed that it hadn't happened to me in like... 20 years. Nope, I just assumed that it was an intended mechanic designed to "spice things up" just like tripping in Brawl or something. From 1999 all the way until 2020. That's my story.
Most likely was added to spice things up as you said, since its still present in stadium, just even rarer.
Good thing they backpedalled on it
I promise you weren't the only one 😂
Rest assured that you were not alone, I also fell victim to thinking no move was actually 100% accurate unless it explicitly said "does not miss", like Aerial Ace.
It started in Pokemon Blue and it wasn't until a year or two ago, watching some random YT video that explained the gen 1 spaghetti code
I was exactly the same until I got into VGC last year. 23 years of being completely wrong. RIP me
I didn't realize this either... What's actually the point of the no miss moves then? Did the devs think sand attack would be everywhere??
Squirtle went super saiyan against Oak. Its +6 defense stat was its lowest non-hp stat and I think that's just hilarious.
Squirtle (against Oak): "And this... is... to go further... BEYOND! SQUIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRR...!"
Charmander lover here!
I think its fair for squirtle to be ranked higher, just as you said its more consistent. That being said, this is way closer than i thought! Thanks for the amazing runs!
I was so scared that charmander would be ranked higher than my favorite little turtle. But I'm definitely happy that all 3 starters put in good performances. The charmander line gets so much love everywhere, so it's nice to see the squirtle squad get small wins over it every now and then.
Scott: "Squirtle, the final starter."
Pikachu: "Hold my light ball."
My thoughts exactly
Don't forget Eevee as well.
@@TheWispGuy
I mean, Eevee is more of a legitimate starter than Pikachu, lore-wise. Pikachu's only appearances as a starter are all due to your rival getting impatient and taking the Eevee that you were originally meant to have, whereas Eevee was always going to be your starter in XD Gale of Darkness.
For anyone wondering, the odds of that reset in Erica's gym (at 18:50) are very low, and Scott got incredibly unlucky.
For Bellsprout to win the battle, Ice Beam needs to miss (1/256). 0.3%
Then, Stun Spore needs to hit. Stun Spore has 75% accuracy, and the enemy AI in Gen 1 has a 75% debuff on hitting status moves, so this is 75% * 75%, which is 56.25%. 0.3% * 56.25% = 0.22%
Wrap does 2 damage per hit and Squirtle has 120 HP. Wrap has a 15% chance to hit 0 times, a 37.5% chance each to hit 2 or 3 times, and a 12.5% chance each to hit 4 or 5 times. On average, Wrap will hit 2.55 times per accurracy roll and deal 5.1 damage to Squirtle, again, on average. So it needs to hit 24 times in a row on average, or for Paralyse to proc when it misses (25%), or for Ice Beam to miss when it misses (0.3%). Since Wrap hitting is more likely than those other outcomes, I'll estimate an upper bound and just go with that probability.
85% chance for Wrap to hit, 24 times in a row, is (17/20)^24 = 2.02%. Multiply that by the 0.22% chance for Ice Beam to miss and Stun Spore to hit, and my upper bound estimate for how likely Bellsprout to win this battle is 0.0045%, less than half a pecent of a percent. One in over 22,000.
Note: Bellsprout has a 7.81% chance to crit in Gen 1. Crits do less than double damage but as Scott always notes, multi-hit moves continue to crit if any of the hits crit, which compounds the additional damage. I didn't factor crits in because I'm not that smart but I imagine I'm less than an order of magnitude off.
The real shocker for me is that there's a 2% chance to hit 85% odds 24 times in a row. That's far more likely than I would have guessed.
Wrap didn't have to roll accuracy 60 times.
Damn. JRose recently had a string of crazy luck in his favour on a one of his runs. I forget the sequence but it was also astronomical. This has to beat that out though. My benchmark for improbability will always be when someone pulled four Bombs in a row as Peach (1/256 x 1/256 x 1/256 x 1/256) but it's crazy just what kind of insane events that pokemon players also generate.
I guess it's like 1000 monkeys writing at 1000 typewriters.
@@josephnation9063True, I neglected this. I'll amend my calculations.
15% chance to hit zero times, 37.5% to hit twice, 37.% to hit three times, 12.5% four hits, 12.5% five hits. This averages out to 2.55 hits per accuracy roll - given that each hit is 2 damage, each accurracy roll deals an average of 5.1 damage, so it only needs to hit 24 times on average. 85% 24 times is just over 2%. Multiply by 0.22% and that's 0.004451197844%, less than half a percent of a percent. One in over 22,000. Rather a large difference that makes 😅
As a charmander main i think you did great, and the fact that chamander beat either of the other pokemon in any metric is something to be celebrated! I feel like online discourse makes it sound like charmander is the worst and that your gonna have a terrible time if you use it him. turns out, nah. if you use him, sure your gym fights will be hard, but you're gonna make up for that with all the the trainers that you can burn you way through. the fact that you only really struggled on one pokemon says A LOT about charmander's viability. third place is third place and we all knew that we were gonna be here, but we put up a great fight.
So when talking about nuzlocking it's horrible which is probably a third of the community. The other 2 thirds are people that say just have fun but if you don't pick what they think is best you are going to hell.
@@itsthekid9815 it's funny, but it's also frustrating
The times between the 3 turned out pretty much how I expected, Sleep is just to busted in Gen 1 especially when you have setup moves. Also feels good for Bulba to win since it's my favorite mon.
I'm so happy to see how these runs turned out!! The optimization was fantastic and it was great to watch. Seeing first stage Pokemon able to push into these ranges is really cool to see.
I'm really looking forward to the Donphand vs Ursaring video ^^
18:36 18:46 shout out to the bellsprout missing Wrap only 2 times, and both of those turns squirtle gets fully paralyzed, what an absolute trooper of a plant they are 😅
I'm happy for the turtoise to get it's video and benchmark, would've love for it to be the best around, but sometimes just beating the Char line is good enough for me. Also, kudos to the Saur line, they deserve to be number 1 at something!, also Ivysaur is the best middle stage, but we are not ready to get that convo started.
I mean, at least Bulbasaur is #1 in the dex 😅
Wartortle has entered the chat.
@@wompus_kingWartortle has left the chat in fear 😂
My guess is the best midstage is weepingbell, but it could be polywhirl or ivysaur
@@ZealousPawn oh, I meant Ivy was the best middle evo of the starters
Excellent performance for Squirtle, but we all know that deep down Bulbasaur would weed out the competition.
I really enjoy the widmanstatten pattern in the bottom right, it’s a special crystalization that happens to iron minerals in the absence of oxygen, exclusively in meteorites. An excelent way to tell if a rock you’ve found formed in space
Ok
Jeeeeez that Squirtle vs Oak battle was insane: 999 in every non HP stat except a +6 defence!
As a charmander fan, I definitely don’t have an ill will towards your ranking it lower. In fact, I’m super impressed you got it to do as well as it did. And I’m super proud of charmander too ❤
Sraring at squirtle's back sprite throughout this video, the all blue palette makes its shell look so much like crystal and making me imagine it as an ice type.
I can't unsee that now.
I think I prefer the Squirtle line the best from gen 1, Bulba 2nd Charmander 3rd as it's always a fan favourite
Bulbasaur: Bulba
Charmander: Char
Squirtle: Bubble
Love it 😂
That Professor Oak battle was absolutely insane 😂
Using dig while paralyzed has a chance to turn you invulnerable!
Edit: ah, you knew that
I really enjoyed this mini series a lot and I'm excited to see Crystal come back tomorrow, with the return of the dapper gent Pryce, as well!
Excellent video as always
As a Charmander stan, I have to admit that Squirtle is way more consistent. If anything, I was surprised the final times were so close. I weighed Withdraw too heavily and didn't really account for Sleep Powder or Swords Dance in my Squirtle>>Charmander>Bulbasaur prediction. I really enjoyed this series! Hopefully the middle evolutions and reruns of final forms are coming soon!
Squirtle’s back sprite always looked like an old man to me
That fight in Erika's gym was so painful to watch. Poor Squirtle deserved better for its first run, there was so much awful luck for it. At least it got a good final time. =)
Bulbasaur Run: Bulba
Charmander Run: Char
Squirtle Run: _Oh, come on!_
Out of all the starters, i don't actually have a preference I assumed Squirtle or Bulbasaur would be the best. Looks like Bulbasaur took it. Which makes sense in the spirit of the game. So many playthroughs though, I gotta give you props for that.
The proper order of the Kanto starters is the finish. Good stuff! Two of my favorite in a showdown tomorrow! Scott’s bringing it this month!
A fan favorite. For good reason too, such an incredible pokemon. And fun to use, too! I think it's gonna do a pretty good job. I could see it beating charmander
1 minute in and I feel a need to write a comment out of appreciation. I like the added text boxes of info about the things being mentioned.
Kind of fits in line with the old ideal, easy = Bulbasaur, normal = Squirtle, hard = Charmander
I think that a good way to weigh in the attempts per Pokémon would be to take them as a time average. i.e. (sum of attempt times)/(number of attempts), which would show the Pokémon's consistency in achieving a given time, good luck outliers would balance out with poor luck outliers. Anyways, love your videos a ton and your data collection, I am truly excited for this month of thoughts.
that loss to the bellsprout wrap cheese made me physically wince in pain, yeesh
Really impressed by the optimization in the second run. Great vid
Can we just talk about the fact that freakin Squirtle can learn DIG but Bulbasaur a Pokémon literally meant to be bulb as in plant and dinosaur a Pokémon all about being buried in the ground cannot learn DIG. Good old Pokémon logic
Look at those little arms on Bulbasaur. I think that gamefreak was on point with that one...
Squirtle looks so happy in the thumbnail 🥺🥰
Somewhere in the time of Squirtle being paralyzed and wrapped to death by that Bellsprout, I noticed that the Bellsprout actually does miss at least 1 wrap, it just so happened that on the same turn, Squirtle was immobilized due to paralysis. Just another pain-layer on the giant pain-onion.
Can't wait for the results, Squirtle is probably gonna end up between the other two starters but i wonder how is it going to perform overall anyway
Diehard Squirtle fan.
I decided to make a gen 1 playthrough with just water types... man, it's crazy how many great Pokemon are water!
I just noticed the options for the trainer's name are Scott, Shawn and Sarina. Such a nice little detail :D
Loved the starter mini series! No matter the results I'm here for squirtle. Also I predict that ursaring will destroy donphan in the vs video tomorrow.
I am a charmander die hard. That said I don't care that he underperformed. I don't love him because he is the best, I love him because he is awesome. Plus C tier is green and that's my favorite color. He wins in my eyes.
Damn straight. Imo, it has the perfect mix of cool and cute, can learn cool moves and it is orange. I love all three starters, but I will always be a Charmander fan!
Looking forward to see their mid stage evolutions. Love the vids, keep it up! :)
Oh Look. its the actual best Starter in the entirety of pokemon. Let's goooooo
I don't see Totodile in this video
I'm sorry Chikorita is not in this video
Based
I'm sorry this isn't a Totodile run? 🤔
@kannavz3494 this is the worst opinion on the internet. undebatedly
It doesn't feel like a Scott's thoughts video if there isn't an instance of going for ice beam trying for a freeze.
I didn't expect charmander to have the fastest brock split.
I am loud and proud a Bulbasaur fan so I am quite pleased with these results. They are exactly as I would expect. Bulbasaur supremacy for the win!
Squirtle for the win! Definitely the best starter
20:19 You can't freeze Venomoth, because it's an ice-type. :)
I love your content.
Don't be scared of change. Nothing exists without change.
Bulbasaur was exactly as expected ; slow to win fights but never really struggles and just consistently gets through the game.
Charmander surprised me; with a decent amount of extra training it can muscle through everything up to Lance, which is just a roadblock and all that extra training hobbles it.
Squirtle just fucking wrecks the whole game.
Getting rid of Ice Beam for Mimic?! You absolute mad lad
This was expected.
Can't wait for tomorrow!
Charmander fan here. Would love to see this battle take place in Gen 2. Kanto is set up to hurt Charmander in almost every way.
Incoming Wall of Text:
For incorporating the number of runs, at the baseline, my first thought is some kind of penalty function with the runs and time as input which generates a _third_ tier list. Two ways I can see the argument going:
1. An exponential or logarithmic penalty function, something like...
`adjusted_time = final_time_seconds * f(runs - 2); // Subtract because there's ALWAYS at least two runs`
...where `f(0)` is 1.0 and by the time you get to `f(10)` it's something like 1.25 or even 1.5. You might argue it should be `f(run - percentile(25, all_run_counts[]))` or even `f(runs - median(all_run_counts[]))` under the assumption that maybe, say, Bulbasaur shouldn't be punished harshly because it took four whole runs. That's not implausible; do tests until you think it matches how you feel. You know your dataset and what you're trying to make it confess to! (Clamp your lower bound to 1.0 in that case unless you want things that only do 2-3 runs to get _bonuses...)_
2. Bucket it with more granularity and apply flat penalties. Time or Tier or whatever your gut says. This would require a lot more "gut feeling" and is harder to justify than misusing algebra to throw a fit, but it'd still probably be interesting.
2a. Actually, there are some possible interesting visualisations you can generate relatively simply to start getting a better understanding of the data:
- Bucket run count by the the number of tiers, then order by time within that tier as a very simple example.
- Then, yet another tier list can be generated by penalising the original final time by half of the tier range as calculated by the run count tier (e.g. S-tier time but D-tier attempts? That's a five minute penalty.)
These kinds of _ad hoc_ analysis can sometimes be helpful clues for finding useful trends.
Come to think of it, I wonder what the distribution of attempt counts is? Is it normal? Flat? Bimodal? Something weird? That can help inform your approach.
Another thing to keep in mind is how you figure rerankings into it and the simple matter of your own skill improving. Your evolution as a player has pushed times lower on replays, which shifts all of the above metrics in weird ways. So you'd be forced to either make arbitrary decisions about cutoffs for pentalty thresholds or come up with some kind of metric for your own skill level.
Actually, thinking about it from that angle... I suspect the _"proper"_ solution would involve going *much* deeper into the actual run data.
See, the number of runs to achieve a given result is a measurement that arises from the compound probabiity of all output randomness across the run. There are intangibles, of course -- hesitation, menuing, bonks, Churro, _et al._ -- but your raw play logs can probably be leveraged to to get you close to something like a "luck/consistency factor".
*In theory,* this is general enough that it doesn't matter what your skill level is at the time because the amount of luck you need for a time is reduced by better stats/moves and better decisions around e.g. XP routing and move learning. Another neat thing is the calculation can be applied to _every_ run, which can give inroads to a better sense of "luck-per-time" crossover points, what consistency looks like numerically, and even more concrete effects of your own improvement.
I realise it'd be a *lot* of time and effort to implement (I suspect there a fair bit of opportunity to trade complexity for accuracy, but even still), but I think it's something worth _thinking about_ even if it can't happen/get farmed out right now.
Super interesting. My knowledge of stats is limited (bio b.s. stats and one semester of graduate level ecological stats) but I'd love to see some more in-depth analyses of these runs by someone who's more well-versed
Interesting thoughts. Penalizing by run count is difficult or impossible to do well because the best run may only be slightly better than the second best but with way more attempts. Both your suggestions suffer from this problem. You need to take into account (1) the raw run count and (2) the actual value generated by the extra runs.
@@fangiscool1 Yeah, that's part of why I came to the conclusion I did at the end.
It's one of those situations where it IS an easily-collected metric that exists and it can, at times, somewhat tell you something about reality, but only weakly and often deceptively. (And good luck convincing middle managers of the need to invest in better... ヘ（。□°）ヘ)
I just love how bulky Squirtle is.
I had an eerily similar moveset when I did my solo Squirtle line playthrough
Squirtle in leafgreen was good as bite became a dark type move...
This algorithm boost was sponsored by turtle pokemon gang 😎
It would be really interesting to see what time Bulbasaur (and others) would have without sleep. Would love to see a comparison as you did with Butterfree a really long time ago.
Daily December is kicking my rear having to play catchup!
I feel like we missed an opportunity to have Squirtle with sunglasses in the art.
Squirtle is my boy, stats be darned
Thank you for the great videos and happy new year
That wrap did miss once, but you got fully paralyzed
The Teenage Mutant Ninja Squirtle
❤ 🍕
This is interesting when compared to the final evolution run. Perhaps the evolution stat changes have a bit to do with it. Am just guessing.
I had some thoughts for naming trainers in yellow, so hope you like them
Misty’s gym; the mandatory trainer with Goldeen could be named Meena, which is a Sanskrit name for “fish” and “precious stone”. Nina and Ninette are both Russian names meaning “fish” and “little girl” (also “fire”, but 2/3 ain’t bad). Personally, I like Ninette for her.
Brock’s gym; I found a good name for the Lightyears trainer; Braxton, which literally means “Brock’s Town”. I found this while looking for a name for the self-destructing hiker. There are a lot of common and unusual names meaning rock, so couldn’t really choose.
Hope you like.
SD hiker is Dudley tho!
thanks to gen 9 i am a major donphan simp in support of great tusks so im saying it now team donphan taking the W tomorrow
I've done speedruns with all of them (although with evolutions), and I'd come to a similar conclusion. Venusaur has the best potential, but RNG can make it just a nightmare. Charizard was my favorite to use, although early/early-mid game is boring with such weak (Ember's 40 BP for so long is bad) or inaccurate (Mega Punch) moves for so long, but it's the de facto best late game IMO, Swords Dance with those stats is just incredible. Blastoise... it's consistent, but the game goes to a glacial pace against Water-type trainers like Lorelei. Also it's just stat-starved so badly. Great movepool though.
Awesome runs.
Charizard was always my least favorite to use, mostly just because of how painful Lance is. But there are new ideas around to solve him, so I’m excited to try the Fire-type again.
@@ScottsThoughtsPokemon Lance is a hard obstacle for any Pokemon not able to tank Hydro Pump well sadly. In Yellow he's the best E4 member I think, unless you're weak to Ground-type attacks.
Guessing its second best it can safely badge boost vs physical attackers with withdraw also has good coverage.
Go bulba! You can fend these imposters off!
Scott, love the videos keep it up !!
This is wild, i feel like we never see scott buy ice beam from the little girl. Really pissed me off that she only gives out useless moves in lets go Pikachu/Eevee
Just imagine if Skull Bash had it's defense boost effect in Gen 1. Badge boosts alone...
no one:
absolutely no one:
scott: “BulbaSARRRR”
Okay and here i thought it was just me hearing it wrong
He's Canadian... give him a break
Hi, Charmander lover here.
I think you did right by the fire boi :)
I'd like if the overlay indicated opponents' Types (and the Venomoth would display random Types).
Hahaha that sounds hilarious!
Finally my favorite gen1 starter
The champions team makes me curious about a Magneton run.
I think missing Mega Punch in Mt. Moon slowed down some of the early game.
Nice to see you figured out a way to make Koga easier in the 2nd Playthrough with Mimic. I wonder though with that strategy, would Sleep Powder be better so you can get by Koga at a lower level?
There! Bulbasaur is easy mode, Squirtle is medium, and Charmander is hard.
I'm a little sad squirtle didn't place first, but I'm going to say sleep powder gave bulbasaur an advantage. As unevolved pomemon they have a harder time outspeeding or getting one hits so cheese strats make a big difference and squirtle's moves are just too straightforward.
squirtle squad rise up
Skull Bash is so disappointing, because I remember using that in Gen 1, and I was like "Why wouldn't I just use Strength of Double-Edge instead of this?", not to mention that this was the only time Skull Bash was ever a TM, so I just always wrote it off as a less-useful version of Strength.
I really enjoyed this comparison. Your result mimics the advice from a game guide I had as a child :).
Question though - have you started mixing your audio differently recently? I watch UA-cam at 1.5x and your voice has recently gained some audio clipping (within the last couple of months). I know you aren't mixing for people playing at 1.5x but I was just curious if it was a change you'd chosen to make.
I remember you had some vocal health issues recently, so maybe it's because your voice is lower when you talk in more recent videos.
I'm a big fan of your videos either way :) thanks for making such great run videos.
Very interesting. I'll let Scott answer when he gets to this comment, but I think you're right. I'm pretty sure I heard him mention on a stream that he's had to be more intentional with the way he speaks and it's resulting in a much calmer sounding voice, which I'll take is probably lower, too.
Mimic mid way through the route 😮 that’s new
❤ amazing
amazing ^^
amazing @@cueneytalpan3386
256 miss with icebeam...gets wrapped for eternity...
hahahahaBAHAHAHAHAHAHAHAHAHAHAHAHAHA.
Clearly Squirtle is a Sea tier Pokemon
My favorite!! Woot!!!!
Charmander is my fav but I fully expect it to be the worst of the 3. Though I thought Squirtle would have been the best.
You think of bulbasaur as a frog? I always thought of it as a sort of dinoSAUR.
I think, it didn't matter but is there a reason why you used 7 rare candies before Koga to Level 57 and right after 2 for Blaine so the Exp from Koga didn't count. Did you want to prevent a possible Mid battle level Up against Sabrina after that or was there another or just no reason to not use all 9 rare candies at once?
18:20 i m,ean it sucks - but its kinda hilarious all the same!
I was surprised that Squirtle didn't win because speedrunners used to use Blastoise before Nidoking became the new tech, but then I realized it's because they use items.
Btw, you mentioned fissure could ko Squirtle, but it can't. Fissure can't hit mons higher level than it.
Actually, in gen 1, Fissure will automatically fail if the opponent has a higher Speed stat. The level based system wasn't implemented until Gen 2.
Squirtle was always my go to in Blue because my older brother had Red and always chose Charmander. Rooting for this little guy!
Baby get the candles, Scott just dropped another one.
Talking about the dig glitch when fighting Agatha made me think of something. Does removing the paralysis remove that state if you were to heal with a full restore?