Spiral Knights - Community created gear class lists: discussion

Community created gear class lists: discussion

112 replies [Last post]

Mon, 05/19/2014 - 03:11

#101

Kwibble

So I am back...

I have finally arrived back at the forums after a week of work experience (which I thoroughly enjoyed. Programming is awesome fun).

[EDIT] I must apologize for this extremely lengthy post. This post is about a formula used for calculating ranks for swords. If you don't want to read that sort of stuff, or just don't care, feel free to skip this post. Bopp, I need you to read through it and give pointers/and also to read some things I pointed at you...

@Bopp Yes, I agree that the system of Midnight-Dj and Military-Lupin came up with was lacking in the specialised damage department - especially after I followed your train of thought. And so I then took into account what Krakob said about Acheron only ranking C over all (which is definitely wrong).

I have now started thinking about it mathematically in those sorts of terms. If we scaled down the rankings to 3, we would still have the same problem - if we increased the ranking to ten, still the same problem. So what if we implemented a 3/5 ranking system. That would require everyone ranking things out of 5, but then we turn that into A B C respectively. This poses the problem of not being as precise (which we already agreed upon as important [I think]).

So what would happen if we combined the two processes (Midnight-Dj/Military-Lupin's and Bopp's)?

That would end up being something with 3 categories which would be:
Damage against all monster groups averaged (which gives the 'overall utility')
DPS (this gives the damage output)
SwordSwingSize (This is to account for say crowd control. And yes, this is a sword specific formula)

This would result in a formula along the lines of:
(((FiendScore + UndeadScore + BeastScore + SlimeScore + ConstructScore + GremlinScore) / 6) + (((DPSInvulnerable / 1000) + (DPSNormal / 1000) + (DPSVulnerable / 1000)) / 3) + SwordSwingSizeScore) / 4

Now before you go all out and yell at me for not making sense, look at the examples below. And before you go yelling at me about that being too much to vote on, people only vote on those first six (damage vs monster families) scores - the rest are pulled from say Bopp's sword guide. Bopp's guide has data that is not really dependent on opinion - it is pure fact. So that means that it doesn't have to be voted upon. And once again, before you go yelling at me about the fact that such a formula takes too much time to compute - just send me the voted upon data, and I will write up a program/script that can give us the answers quickly and efficiently. Just give me the data from the voting, I am happy to calculate it all.

[NOTE: In these examples, I can't find data about DPS against invulnerable/normal enemies. So for the sake of these examples, I am going to use the DPS from Bopp's sword guide. I will need to test with other data before we can actually use this formula though. Yes, there is Lancer Knightz (guild) data, but I have yet to read through it and interpret it. I will get to that for my formula. Bopp? Mind helping me with that?]

And so here is an example of the Acheron put through my formula:
(((5+5+4+3+2+2) / 6) + (716 / 1000) + 0.5) / 4
= (21 + 0.7 + 0.5) / 4
= 22.2 / 4
= 5.55 (This is called Rank S! :P)
= 5
= Rank A

And here is the Leviathan Blade:
(((3+3+3+3+3+3) / 6) + (432 / 1000) + 0.5)) / 4
= (18 + 0.4 + 0.5) / 4
= 4.7
= Rank B

Or, if you think that it should be lower than that (honestly, what do people have against the Levi Blade aside from the fact that it can't deal massive damage against two monster groups? Someone explain that to me?), here is with the damage vs monster families with a D score:
(((2+2+2+2+2+2) / 6) + (432 / 1000) + 0.5) / 4
= (12 + 0.4 + 0.5) / 4
= 3.2
= 3
= Rank C (Oh look, still ranked C. Is this still better than people are expecting?)

So please, for further testing of my formula, give me some weapons that you think wouldn't be in their right spots. I will run them through my formula and give the results.
And remember, I am that math junky guy who can actually be bothered to boring repetitious work (considering I can't vote).

Mon, 05/19/2014 - 04:04

#102

Bopp

response

Kwibble, thanks for thinking all of that through. It's not a bad idea.

There is a problem: The bonus damage that you get from using a damage type against a vulnerable monster is smaller than the penalty that you get from using a damage type against a resistant monster. In other words, WRH and the piercing swords perform extremely badly against their resistant monsters. Thus the DPS part of your calculation will cause them to end up weaker than normal swords.

But WRH and piercing swords are generally preferred over normal swords by expert players. And that's because expert players don't think about weapons in terms of their average utility. They don't need one sword to handle all monsters, because they can choose their equipment based on the upcoming threats.

For example, one of my alts owns only two weapons: Combuster and Final Flourish. (And Proto stuff.) Do I care that Final Flourish is weak against slimes? No, because I would never use it on slimes. I would use Combuster. And this isn't some kind of extremely tricked-out knight. This knight owns only two weapons.

Also, isn't the three-part DPS calculation kind of redundant with the six monster family rankings? If we get rid of the former, then we return to my system. If we get rid of the latter, then we could make some kind of hybrid of your system with Midnight-Djs/Military-Lupin's --- player votes mixed with data. I recommend that you try to develop your idea in that direction more. It could end up better than my idea.

Mon, 05/19/2014 - 07:47

#103

Plancker

I still think if you just

I still think if you just count damage you are underrating and overrating some weapons. for example the rocket hammer is great against greavers, not because of its damage, but because of the constant interruption around you. You can slide through a group of greavers that surround you (like in arenas) and kill them all with minimal damage taken. on the other hand using a combuster against a group of 15 mecha knights will be safer than using the hammer. thats why I dont think we need some sort of universal formula, but just let people vote based on their gut, with some guidelines that keep in mind the advantages each weapon has over certain enemies. While voting like that is more subjective than using just a formula, it may be more accurate because one formula cant list all the variables. And I think with a high enough number of serious voters, the subjectiveness shouldnt be a problem.

Mon, 05/19/2014 - 10:52

#104

Bopp

great

Great. I agree with letting people vote according to their subjective view, which takes into account many factors that may be too complicated to quantify. Kwibble was just talking about supplementing the votes with some data. I could take it or leave it.

Plancker, how many aspects are they voting on, and what are those aspects? For example, is there a single rating of "overall quality"? Or is there "overall quality against fiends", "overall quality against constructs", etc? Or is there "overall quality against greavers", "overall quality against devilites", etc? Or what? That's what the argument here is about.

Mon, 05/19/2014 - 13:36

#105

Plancker

My idea would be to create a

My idea would be to create a general guideline with questions to help people make their opinions. some examples that come up for me are:

-How does the weapon perform against mechaknights/gremlin thwackers/almirian crusaders (basically all elite monsters that are not your average punching bags)
-How does the weapon perform against greavers
-against devilites
-against a room of turrets
-against large enemies with shields like trojans
-against gremlin menders maybe?
-How well does it perform in danger missions (combuster is excellent in LoA and GiTM, works in HoI; GF is only great at C42 => 2-3 vs 1)
-How well does it perform in shadow lairs
-How good is this weapon for farming the typical crown-farm missions like FSC/jelly king

(only taking the appropriate monster-related ones into account, for example a pure elemental sword shouldnt get negative points because its not good at killing gremlin thwackers or menders; also I only listed gremlin menders as healers because of their shielding aura, silkwings are like jellies and zombies, easy kills)

I think we dont really need to give it a score on each question, but just use them to base your opinion on. I think thats why it might also be best to not try and give every weapon a score, but try to have everyone create their own personal list based on their subjective view on the item, that uses relative position as a measurement of the weapons rank, instead of an absolute score. But to do that players would have to rate all the swords at the same time. I think the best way to do that would be to make a list of all the swords, then move a sword up if its better than the one above it, and keep doing that until you dont feel like you can move anything anymore. (Bubble sort on wikipedia explains this method) Then maybe give each sword their position on the list as the score, then add up all the scores from all players' lists. for example if 10 people rate acheron as the best sword and 5 people rate it as 2nd best out of 20 (?) acheron would get 10x20+5x19 points from those players. That system may be hard to do because it requires everyone to rate every weapon, and I would personally find it hard to rate weapons I dont have. Even though I own a divine avenger I dont think I can rate a GF as accurately, even though ive seen it used by party members countless times and know its attack patterns etc. All I can do at most is make a good guess of how strong it is, but I would still not know how strong the curse on enemies actually is or how easy/hard it is to land enough hits on dodging gremlins to actually kill them. I also think people would tend to overrate some of the weapons in their own arsenal, which is another problem.

Mon, 05/19/2014 - 13:32

#106

Bopp

this is the 19-point rating system

So, to summarize, you want to rate all sword on a single scale. The scale would have 19 ratings. You assign the best sword a rating of 1, the second best sword a rating of 2, the worst sword a rating of 19, etc.

I mentioned this 19-point scale idea in post #32 above. But the consensus in this thread has tended toward a 5-point scale (A-E or 1-5). I myself don't feel strongly about it.

We don't have consensus on the number of scales yet. I think that a single scale is too simplistic, no matter how many points are available for it. Midnight-Dj/Military-Lupin think that three scales suffice. I think that six scales suffice.

Deciding things by consensus/committee is often difficult.

Mon, 05/19/2014 - 13:48

#107

Plancker

Well I like the 5 ranks

Well I like the 5 ranks scale, I think its more appealing to new players. But I think in order for us to make a list with 5 ranks, we would probably need a 19-point scale, which can be transformed into a 5rank one later. I think that is safer because it forces us to compare each weapon and evaluate all its (dis)advantages over certain other weapons. If we try to put them in ranks from the start I think there will be a lot of confusion as to what exactly is for example A rank or C rank. However rating them on a 19-point scale is easy, a sword with rating 5 is better than a rating 6 sword. Then after we get all the votes from everyone, we can look how many full lists we get (= how many players actually voted). Then depending on that number (for example 24 players voted) we can decide the rating treshold for each rank: for instance A-rank has all weapons that have less than 75 rating (which would mean it averages around the top 3rd weapon on players votes). I think it is important though to decide on the tresholds with care and set them before we add all of the scores together, as to avoid arguements about borderline weapons.

Mon, 05/19/2014 - 19:55

#108

Kwibble

So...

I have been experimenting with refining my formula and doing quite a bit of number crunching. But to fully test the scope of my new improved formula, I need to run a 'bad' sword through. So can someone tell me this:

What would you rank Winmillion's effectiveness against neutral targets? (E - A)
What is Winmillion's DPS? (ASI+6, damage+0, heat level 10, depth 28, vulnerable enemies)
What is Winmillion's stroke width and length?

Basically, if Bopp was going to add Winnillion to his sword guide, what would the stats be?

Tue, 05/20/2014 - 04:43

#109

Bopp

good question

Good question. I don't have Winmillion, or even consider it to exist, so I can't answer. (I suppose I should get it.)

Sat, 05/24/2014 - 16:23

#110

Kwibble

So... Anything of use for me? Or we just chillin'

Sun, 06/15/2014 - 07:07

#111

Krakob

Laziness, being forgetful, and being busy is a wonderful combination, don't you think?

This thread has been around for about two months but to be plain honest, I don't think I fancy trying to make it proceed further in the direction it has taken earlier. I get the feeling that interest diminished for others and honestly, it kinda did for me, too. While a big reason that this has slipped away is that I've been kinda busy, it has also been because I just haven't felt like working more on it.
I would by all means be happy if someone else keeps this discussion thread going but if not, I might just enstablish my own rules and make a vote without any discussion to speed up the process and get done with it.

Sun, 06/15/2014 - 07:54

#112

Bopp

hard to accomplish anything by committee

It's hard to accomplish anything, if you need to get a bunch of people on board. It's much easier just to unilaterally do the task yourself, before your interest goes away. Sorry it didn't work out. For what it's worth, this thread has inspired me to perhaps reorganize my sword guide based on bad-good-better-best categories for each family.

Search

Community created gear class lists: discussion