What's new

Discussion The productive way to compare characters

Would you like to see a tool like this implemented in TYM?

  • Yes

    Votes: 27 67.5%
  • No

    Votes: 0 0.0%
  • You could have written a Salt Report in the time it took to write this

    Votes: 13 32.5%

  • Total voters
    40

SneakyTortoise

Official Master of Salt
Didn't m2Dave have a thread similar to this awhile back? Anyways, I'm all for active open discussion about the individual strengths and weaknesses of a character.
I just checked, and you are right. He did create a very similar thread, inviting people to rate their characters on different characteristics, however, this is mainly focusing on the visualisation of something like that as a tool for character discussions and comparisons
 

Infil

Noob
Hi all,

I'm the person who made the ki.infil.net guide. It's a one person show, so I'm responsible for all the content, video, and visualization stuff. I was contacted on twitter because it seems the MKXL community is interested in doing something similar to my visualizations.

I'm not an expert on MK, but I can definitely help you with the coding side of it (meaning, I can provide the code I wrote and answer questions/help you through what you need). The code simply loads in a big JSON file of data, parses it, and then draws stuff to an SVG canvas. Specifically, I use the D3.js library, which is very well known to people who often do web graphics or visualizations. That's just to give any insight in case there are coders reading this.

From reading this thread, it sounds like you guys are interested in something like this, but are unsure exactly how you would make it work for the MK scene (representing different variations, a community vote vs. a small group of people deciding on ratings, etc). I can tell you that I think my KI numbers benefitted from just being one person deciding on them, since I used a mostly consistent metric across the characters. Saying that Jago has 8/10 for offense means something, because I rated all characters offense similarly. If you split the job between different people, you will get cases where two people think 8/10 means different things, which is hard to reconcile (there are ways, but it's definitely more work).

I'm on a trip right now so I won't be able to help with coding for a week or two, but I can check back on this thread every so often when I have wifi access and answer questions or whatever. In the meantime, if you guys are interested in doing it for MK and not waiting for I2, you might want to think about what sub-categories you want to rate the characters on and maybe pick 2 of the top tiers, 2 middle tiers, and 2 bottom tiers and see if you can agree on an out-of-10 rating for each character in each category. That should be an interesting enough exercise to get started.
 

Tanno

The Fantasy is the Reality of the Mind
This is a great idea. It should help with collecting data on the characters. It would give us a good picture of comparing the old and the new characters, as well as helping NRS understand how it should be balanced at the end.
 

Infil

Noob
Oh, I'll also mention that you can rate characters in two different ways (and my visualization does both of these): you can assign everyone an out-of-10 rating for each category, or you could strictly rank them from best to worst with no ties in each category.

Both of these have value. For example, in a game like KI where lots of characters have strong offense, you would probably rate rushdown as no lower than a 7/10 for each character. Just because X has the "worst" offense in the game doesn't actually mean that he struggles to open you up, so you don't want to make it seem like that. But at the same time, if everyone gets an 8/10, then one of those characters is the "worst" of the 8/10s, if that makes sense. Basically, if you rate someone a 2/10 in a category, they have to be absolutely terrible at it. But someone can be the worst in the game at something without actually being all that bad at it.

In my visualizations (at http://ki.infil.net/characters.html), you can toggle the "rank mode" to turn between the out-of-10 ratings and the "everyone from best to worst" ratings. You'll definitely have to make some tough decisions on ratings if you want to do a best to worst, but that's the fun of it, I think. I frequently changed my opinions on ratings as my guide evolved over the years ("wait, did I really think X had worse offense than Y? Guess I'll re-rank this a bit").

The fun part of the "rank mode" graphs is that they are all very pointy, since you can have a character who is 5th best in the game at rushdown, worst in the game at zoning, 14th best at footsies, etc. In many ways, they give you a different look at the character than the standard out-of-10 rating does, because even the slightest of weaknesses in a category means a huge drop, and if they are filled up in a category, you know they are *really* good at it.

I'd recommend you go play with my rank mode visualizations. See if you can find some fun shapes by toggling off the different categories in random orders! It's just a lot of fun to mess around with. For example, if you turn on every category except "total", Thunder's rank mode graph looks a bit like Texas.

And by the way, if you want to see what my definition is of a category like "Rushdown", which is kind of a nebulous thing, just hover your mouse over the category button and a definition pops up. You guys are free to redefine these terms as they apply to MK, of course.
 

STB Shujinkydink

Burning down in flames for kicks
I actually used this when i picked up Soul Calibur V. They have this on the select screen so you know what you're getting into when you pick said character. I think having the developers update this proves true intent of the character. if X character is said to have bad defense but ends up not having it, the community can talk about it and they can update the graph. Giving the community access to fully accessable info is a bad idea. everyone is biased whether they like it or not
 

SneakyTortoise

Official Master of Salt
Hi all,

I'm the person who made the ki.infil.net guide. It's a one person show, so I'm responsible for all the content, video, and visualization stuff. I was contacted on twitter because it seems the MKXL community is interested in doing something similar to my visualizations.

I'm not an expert on MK, but I can definitely help you with the coding side of it (meaning, I can provide the code I wrote and answer questions/help you through what you need). The code simply loads in a big JSON file of data, parses it, and then draws stuff to an SVG canvas. Specifically, I use the D3.js library, which is very well known to people who often do web graphics or visualizations. That's just to give any insight in case there are coders reading this.

From reading this thread, it sounds like you guys are interested in something like this, but are unsure exactly how you would make it work for the MK scene (representing different variations, a community vote vs. a small group of people deciding on ratings, etc). I can tell you that I think my KI numbers benefitted from just being one person deciding on them, since I used a mostly consistent metric across the characters. Saying that Jago has 8/10 for offense means something, because I rated all characters offense similarly. If you split the job between different people, you will get cases where two people think 8/10 means different things, which is hard to reconcile (there are ways, but it's definitely more work).

I'm on a trip right now so I won't be able to help with coding for a week or two, but I can check back on this thread every so often when I have wifi access and answer questions or whatever. In the meantime, if you guys are interested in doing it for MK and not waiting for I2, you might want to think about what sub-categories you want to rate the characters on and maybe pick 2 of the top tiers, 2 middle tiers, and 2 bottom tiers and see if you can agree on an out-of-10 rating for each character in each category. That should be an interesting enough exercise to get started.
Thank you for helping us with this, it is really appreciated.
The only problem I see with having one person do it, is that I think there's only one or two people the NRS community would accept as qualified enough to do it (sonic fox and reo), and I'm not sure they'd want to take the time to do something like this. @YOMI REO @SonicFox5000 would either of you be up for doing something like this?