The method to the madness

“‘Data! Data! Data!’ he cried impatiently. ‘I can’t make bricks without clay.'” -Sherlock Holmes in Sir Arthur Conan Doyle’s The Adventure of the Copper Beeches

This is one of my favorite quotes in the Sherlock Holmes library because it perfectly encapsulates what makes him a good detective. He waits for data to make his judgement and he’s amazing at analyzing that data dispassionately. Once he’s looked at that data, he then amends any previous conclusion he’s held based on the new evidence without any angst over having to change his mind.

Humans, by our nature, are kind of terrible at waiting for data to come in, analyzing it dispassionately, and changing our minds from our initial conclusions. This is why the scientific method was developed and I’m going to discuss how YOU can apply it to your Amiibo training and how we as a community can apply it to the metagame overall.

Training Amiibo can be a bit weird at times. They do things that you feel don’t make sense and, a lot of times, we approach the whole thing a bit haphazardly. The Scientific Method can give you focus on how to approach the problems your Amiibo give you. However, let’s start with what the Scientific Method is, broadly speaking.

  1. Ask a question
  2. Construct a Hypothesis
  3. Test with an Experiment
  4. If the testing procedure isn’t working, troubleshoot the procedure
  5. Analyze the data
  6. Draw a conclusion
  7. Share results

With Amiibo training utilizing this method can yield great results. However instead of “Ask a question” the first step is more deciding what you want your specific Amiibo to do. So, let’s run through a little scenario with our newest hypothetical trainer, Johnny Newbie.

Now, Johnny Newbie has a shiny new Amiibo he just bough from Gamestop, his Ridley that he has named Spyro because Johnny Newbie was big into Skylanders back in the day.

Johnny Newbie also feels very clever and is sure no one else has come up with this

Now, Johnny is pretty jazzed and is wanting to make the best darn Ridley Amiibo in the game! He’s sure Alpharad will be making a video about it and everything! So you ask Johnny, “So, you want to be the very best, huh? So what’s your plan? What kind of Amiibo do you want him to be?”

“A great one!” Johnny replies.

“Well, okay, but great how?”

“Well… he wins,” Johnny says, undeterred.

“Well yeah, but good how? Does he fight aggressive off stage? Does he go for meteors? Is he more grounded and goes for throws? Does he like to charge up smash attacks or will you be spamming B-side? Does he like to play in the air but safely above stage? What’s your plan here.”

It’s at this moment, Johnny Newbie realizes he hasn’t really thought this through. He was sort of going to just fight Spyro until he got good enough that the Amiibo was always beating him. You see, Johnny Newbie hadn’t been asking very specific questions. Here are some good ones to maybe ask yourself before you even start training an Amiibo:

-Vanilla or Spirits?
-What 3-4 moves do I want this Amiibo to be good at using?
-Am I playing to win or am I wanting to get this Amiibo to get kills in cool ways?
-If I’m trying to win, what opponents is my Amiibo likely to run into? What strategies will they employ? Will that change what moves or playstyles I want to emphasize?

Notice how asking just a few questions start to narrow your focus on your Amiibo? That’s the power of asking questions. You go from vague concepts to coming up with a plan. We’ve hacked off the rough edges and things are starting to take shape.

So, we run that battery of questions by Johnny Newbie and he decides he wants his Ridley to be an off-stage bully.

“Great.” you say. “So, what moves will he use to get his opponents off-stage?”
“Side-B will be good for that, right?” asks Johnny Newbie.
“Could be. And after has them off-stage?”
“Neutral-air,” Johnny says confidently.

It sounds like Johnny Newbie has a plan… or a hypothesis. This is why you ask questions. Questions get you to drill down and think of specifics. Once you have those specifics nailed down, you have a theory of how your Amiibo will achieve your goal. Now that you have your Hypothesis of how to get your Amiibo there, it’s time to design your experiment.

In Amiibo terms, your Experiment is how you train your Amiibo. Normally this will look like simple 1v1 matches against your Amiibo on a neutral stage. However, depending on what you want your Amiibo to do you might change up your experimental design. Here are a few examples you could try

-Are you trying to make your Amiibo stay very grounded? Perhaps try setting the weight to Heavy so the Amiibo will learn that jumping in the air isn’t advantageous.
-Do you have a specific move you want to train into your Amiibo that THIS is a kill move? You could set up a match where you both start at 300% and thus you can guarantee that certain moves WILL kill.
-Does your Amiibo have a bad habit you want to train out of it? Consider turning the Speed to Slow so that you will have the reaction time necessary to punish your Amiibo for using bad moves.

There are lots of other potential setups you can make to teach your Amiibo certain habits (or break them of bad ones). The point is, you’re free to Experiment here. If, as you train it, it seems like the Amiibo isn’t learning the lessons you want, re-examine your Experiment.

Do you have a good procedure? Maybe you’re doing something wrong? Maybe you have a concept incorrect? Could be! This is where things start to get more difficult. Studies have shown that humans have a hard time changing their minds from the first solution they see. You may get out of the Experiment phase feeling like you did everything perfect. But guess what…

You’re probably wrong.

The reason you’re probably wrong is you haven’t hit the next step yet which is Analyze the Data. And you don’t truly know what the Data will be until you start pitting the Amiibo you train against other Amiibo. The AI for Amiibo is a complicated thing and likely won’t behave 100% like you planned.

But that’s okay. If you enter your Amiibo in a lot of tournaments or pit them against other Amiibo frequently, you’ll soon get a mountain of data to look at and can start to hammer out the quirks in your Amiibo’s behavior… if you’re willing to look at the data dispassionately.

Let’s get back to Johnny Newbie. You know he entered Spyro into a tournament a few days ago but you haven’t heard from him. You reach out and ask how it went and he is NOT pleased.

“It’s so stupid! He went up against an Incineroar! That thing is so busted! It’s not fair! Why would the tournament organizer let something in that’s so broken!?”

Uh oh! Johnny clearly has let the loss get to him. His pride has been wounded. And fair enough, he put a lot of time into that Ridley’s training! If something you put so much work into got bounced in the first round it’s only natural to feel some shame. However he’s not looking at the problem dispassionately. He’s let his emotion cloud his judgement!

Is he wrong? Maybe. Maybe not. Perhaps the Incineroar Amiibo IS broken (note i use an Amiibo that is nowhere close to being published so we can all view this argument without any concern over current events). Maybe there is some quirk in his programming that makes it just plain unfair to have it fight other Amiibos. However Johnny is snapping to that decision after one match. It’s not exactly a thorough though process.

Johnny has, as the Scientific Method would tell you to do, Drawn a Conclusion. However, Johnny didn’t view the data dispassionately so he’s not approaching this with a clear head. Also, you had to go talk to him. He’s not sharing his results. We are lucky and have a very active community of Amiibo trainers who would love to chime in and answer his questions. Johnny is just ignoring that resource completely.

Did Johnny watch the replay the tournament organizer was kind enough to post? What moves were giving his Ridley trouble? Is this maybe just a bad match-up? How did the Incineroar fare later in the tournament? Was your Ridley using some moves that weren’t ideal for that match-up?

Johnny isn’t asking himself these questions. Maybe it’s because he’s a newbie and doesn’t know to ask these questions. Maybe it’s because he’s let his emotions get the better of him. These are all things that can be addressed though going over the data, sharing what he’s learned, coming up with new questions to ask, forming new hypotheses, and then creating new experiments by trying to tweek Spyro’s behavior.

However, if Johnny Newbie just insists on the fact that Incineroar is broken, he’ll never improve his Spyro because insisting something is broken takes all the control away from Johnny and gives it to the programmers at Nintendo and, I hate to break it to Johnny, they don’t much care how well his dear Spyro does in an online Amiibo tournament. They still have to fix Olimar or online play or something. Spyro is VERY low on their priority list for the programmers.

Never give your hopes and dreams for your Amiibo over to someone else. Take a hard look at their performance! Figure out what they are doing wrong to figure out why they aren’t achieving your goals. Note, I said, “achieving your goals.” Not, “Winning.” The point of playing with Amiibos is to have fun, however you determine that to be. If all you want your Amiibo to do is dunk on other Amiibo for glory kills, by all means, go for it! Maximize your fun! That’s what this is supposed to be. Fun.

So now this is where I segue-way from talking about how people can scientific methodology to maximize fun into how we can use the same process to figure out our current metagame. If that doesn’t sound like barrel of fun to you then I don’t know what is!

Right now we’re looking at a different problem in the two of the three main meta-games (for those curious, the main metagames are 1-on-1 Vanilla, 1-on-1 Spirits, and Doubles). I say “problem” hesitantly as we are barely over a month into the game’s life so right now they are more observations that are hinting at potential issues but that’s kind of a mouthful so… you know what? Let’s just stick with “problems.”

The state of the meta right now was very well put in Leaf’s recent article, “A Month Later-What We’ve Discovered Thus Far”. If you haven’t read it yet, I highly recommend it. One of the issues it talks about is how Links are doing very well in the Vanilla metagame.

Link’s overall game is very versatile. He has good projectiles that allow him to keep space from his opponent if he wants to but also has some strong smash attacks for when the opponent gets to close. What more, a lot of the other Amiibo that we currently consider to be strong in the early Vanilla meta (Bowser, Ganondorf, Isabelle, Yoshi) all have bad match-ups against Link.

We are looking at a situation where Link is sort of the rock to the rest of the A-list’s scissors. Link does have bad match-ups but a lot of those match-ups either haven’t been entered a lot in these tournaments or are getting cut up by the metaphorical scissors. With all the papers cut up, it just paves the way for rocks to smash scissors, giving Links a lot of victories early on.

Now, does that mean we should ban Link from Vanilla? Well, let’s apply the scientific method…
Question: Is the A-list that Link is strong against written in stone?
Hypothesis: No, it is not. We’ve only been playing a month and a lot of characters who have theoretical range-based advantages against Link (Mega Man & Samus) haven’t been entered into a lot of these tournaments. Additionally, Link had a bad match-up against Marth and Lucina in Smash 4 and people aren’t entering Marth & Lucina into a lot of tournaments because… people are probably tired of Marth and Lucina since they dominated in Smash 4.
Possible Experiment: People start entering those characters and other less utilized characters into tournaments and see how they fare against Link.

I myself will start to train a Samus and a MegaMan specifically for this purpose. Additionally, we have no idea how Link will match up against some of the Amiibos that are coming out in February: Ice Climbers, King K. Rool, and… Piranha Plant.

I for one welcome our new floral overlord

Even if the Plant can’t save us, we also may just have a sample size bias. Everyone and their dog has a Link Amiibo. He got one of the coolest updates to his kit. People have been experimenting with him a lot. Another possible explanation is we’ve just figured out Link’s strengths faster than other Amiibo because we, as a community, have put more time into him.

The point is, I still feel we’re a good bit away from being able to reliably state Link is truly a meta-wrecking issue. There are too many reasonable theories on why he is currently doing well in tournaments and how we might be able to counter him to do anything drastic just yet.

The same cannot be said for the Spirits metagame. It may only be a month in but a quick issue has been found. It’s two particular skills that you can attach to any Amiibo: Super Armor and Slow Super Armor.

In the Spirits metagame, these two skills are proving to be very effective at winning matches, especially when equipped onto Amiibos who have slow, powerful smash attacks that can benefit from the flinch resistance (Bowser, Ganondorf, and Ike to name a few). This has lead to several Spirits tournaments where the final 4 Amiibo are all ones that have equipped some form of Super Armor.

That, in and of itself, might not be a problem… if it weren’t for the fact that Super Armor vs. Super Armor fights can be a little on the boring side as two Amiibos slowly walk up to one another and just charge up big attacks. It doesn’t tend to allow for a lot of interplay or exciting matches.

If it were just one Amiibo dominating that would be one thing but the skill is proving to be quite versatile and with another heavyweight on the way (King K. Rool) we’re likely going to be adding to the list of Amiibo that can take advantage of it. So, should we start banning it?

NO!

We are scientists! We don’t just knee jerk react. We apply a scientific methodology and see if we can find potential solutions! May it turn out Super Armor is just too strong and is warping the entire Spirits metagame around it? Certainly is a possibilty we need to consider given how things look right now. But banning it too soon would be a mistake as we’d be limiting the options people have without fully investigating other avenues of attack. So let’s pick some good questions to start and get to hypothesizing!

Question: What makes Super Armor and Slow Super Armor so strong?
Hypothesis: It allows Amiibo that normally get punished for standing there with long wind-up attacks to avoid said punishment. Furthermore when Amiibo run up to try and interrupt the attack like you normally would, THEY get punished.
Counter-Question: Could you train an Amiibo to be throw-based to punish the Super Armor style? Have people tried that?
Answer: Some have and it hasn’t gone well. However, not a lot have. The number of SA and SSA Amiibo that have been entered greatly outnumbers the number of anti-Super Armor Amiibo that have been entered. But even if you did the SSA still have another slot and could start running Impoved Escapability to bypass it. Most haven’t yet though because there hasn’t been a need to yet.
Hypothesis: There might exist a type of Amiibo with grabs, throws, and command-grabs that could counter Super Armor and we haven’t experimented with that enough.
Experiment: Create a tournament where the express purpose is to test this exact concept and see if anyone can find an Amiibo style that can dominate Super Armor-based Amiibo

So, my friends, this is where I put my tourney where my article is and announce I’m creating exactly that tournament. It will actually be the first of a series of one day tournaments that I call…
D.A.T.A.S.T.R.E.A.M.- Diagnostic Amiibo Tournament Analyzing Strategy Through Regular Experimentation And Matches

This is also the first time you’ve seen an Amiibo tournament with an overly complicated acronym, let me just say, “Hi. I’m Splice. I’m the acronym guy. I have a B.S. in Physics and every Physics graduate is a mad scientist at heart and, as Saturday Morning Cartoons have taught us, mad scientists like their acronyms to spell stuff out.”

DATASTREAM tournaments will be designed to figure something out or work on specific concepts within the Amiibo metagames. This first tournament will be a two-stage Round Robin tournament where I’m asking everyone to send one Amiibo that utilizes Super Armor or Slow Super Armor-based strategies and one that is designed to count them. In theory there is a healthy Spirits metagame out there that can include the Super Armor mechanic but only if we find a reliable counter first.

And the best part? YOU can help! Enter the tournament by filling out this form and attaching your .bin files to it. Do you not know how to extract your Amiibos’ .bin files to submit to online tournaments? Well, then you should read this handy article that will start walking you through that process!

Can’t quite figure it out? Well, then let me point you back to the Amiibo Dojo Discord channel where there are lots of people who have done it before that would be willing to help troubleshoot your issues! Additionally, you can find out about other tournaments people are throwing all the time in the #tournaments section of the Discord! It’s seriously a good resource for Amiibo-related shenanigans. If you haven’t joined it yet, what are you waiting for? Jump on in, the water is… fine. It’s weird but it’s also fine.

Even if you can’t join (deadline for submission is Thursday night at 6:30PM Pacific/9:30 PM Eastern) you can definitely help the discussion by showing up for the tournament when I broadcast it on twitch.tv/splicestream at 9am Pacific/Noon Eastern on Saturday, January 19th. There in chat we can share our observations and enjoy the comradery that comes with watching Esports together… d’awww. Isn’t that sappy?

That’s not the only tournament I’ll be hosting where we’re testing things out. I have another tournament that will be a bit of an extended format (happening over multiple days) that will be exploring the Squad Strike Format. We’re doing this partially to herald the new feature and as an overall celebration of the new Smash Bros. It’s called G.R.A.S.S.C.U.T.T.E.R.

Grand Round-robin Amiibo Squad Strikes Celebrating Ultimate’s Truly Thrilling, Exciting Release… I did mention I’m an aspiring mad scientist, right?

I’m excited to see what interesting viewing and competitive opportunities the Squad Strike format could bring to Amiibos. It’s the fun of Team combat with the strategy of 1v1! The only thing I ask is you send a good variety of Amiibo personalities and types as we’re hoping to get a lot of data from the tournament. We are especially looking for Versatile, Lighting Fast, Cautious, Realistic, Offensive, Technician, Entertainer, or Tricky personality Amiibos so we can get a better feel for what they do.

So, while I hope you all help out and join DATASTREAM #01- Finding Cracks in the Armor, it’s no big deal if you can’t. I’m hoping you can apply scientific methodoligy to your Amiibo training and towards helping us figure out the problems in the metagame. We’ll never find good solutions if we cut off experimentation too quickly. And we’ll never get good experimentation without a wide variety of data. That data will come from wonderful trainers like yourself looking into avenues you hadn’t considered before.

Don’t be afraid to get crazy. After all, the worst thing that can happen is it doesn’t work and we learn something. As far as experiments go, seems pretty low risk. And who knows, you mind discover something we had never considered or, at the very least, something incredibly hilarious. But, until next time…

Stay fit.
Keep sharp.
Make good decisions.
Splice out!

Advertisements

Post a Comment