Research:Judge

From Discworld MUD Wiki
Revision as of 20:32, 3 September 2011 by Frazyl (Talk | contribs) (Speed anomalies: typo, rephrase)

Jump to: navigation, search
Filing cabinet.gif Old content for this page has been archived, and can be found at the following locations.
Judge/Archive 1Judge/Information thresholdsJudge/Quality
Judge/Speed

Old content

Since judge changed with this announcement, all the old research data is invalid, and has therefore been consigned to the archives. --Chat 20:02, 10 January 2011 (UTC)

New judge research

So, new judge research is needed!

Relevant stuff to know:

  • Your skills don't seem to affect the average result returned anymore - just how much information you can see, and the variance of the results returned.
  • You have a 'judge info rating' (JIR) for each weapon class, which is derived from your fi.me.x bonus in that weapon class and your ad.ev.we bonus (see here for the formula).
  • The difficulty of judging a weapon correctly is a test of its inherent difficulty versus your JIR bonus.
  • If you judge a weapon successfully, then your JIR determines how much data you see out of:
    • The overall weapon quality
    • How easy it is to attack with
    • How easy it is to parry with
    • The maximum damage with the weapon
    • The average damage with the weapon
    • How fast the weapon is to use.
  • Once you are able to see speed info, you can see all the data, and the variance of the results goes to zero (ie. you consistently get the same results with judge each time).

Therefore, the way forward seems clear now:

  • Get a JIR high enough that you can see speed info for your preferred weapon class.
  • Judge weapons in that class.
  • Put the results on that weapon's page

I've got a high enough JIR in sword (but not anywhere near in anything else), so I'll start on swords.

--Chat 19:30, 19 January 2011 (UTC)

Another heads up: if you don't have the command at your disposal, judge will not let you see that the weapon can perform this command. I fell for this trap for the tiger fang, and after learning slash I could see it.

--E 1 April, 2011 (Not a joke)

The affects of Stats on Judge

This is what is known so far:

  • Strength is the only thing shown to affect the results of judge. This is regardless of melee bonus.
  • The only things so far changed by varying Strength are ease of attack and parrying capability.
  • It varies from weapon to weapon what is changed. See Fine sabre and Rose-hilted long sword for a very good example of this.
  • The variations are not standard to a specific weight. A fine sabre and rose-hilted long sword have equal weights yet are affected differently by strength variations.
  • The variations are not standard to the length of the weapon. Two different weapons with the same dimensions and weight will vary differing amounts.
  • There seems to be a "recommended strength" point at which the parry goes up 1 level for each point of strength. This is consistent on almost all weapons. For a long sword, for example, this is around the 16, 17, and 18 strength points. Otherwise, the drop is far more gradual, decreasing 1 point for every 3 or 4 of strength.
  • There is not a "recommended strength" point for ease of attack. For all weapons so far, it has been the same level for every 7 points of strength.
  • Changes in dexterity have shown no impact on judge, from going as low as 11 to as high as 23.
  • Going from 10 constitution to 5 showed no impact on judge. It is very likely that there is none.

What needs to be done:

  • All weapons should probably be judged at every strength value. Different values can be obtained by use of dex gloves, flat caps and other stat items.
  • Add as many strength results as possible to the weapon infobox for each weapon.
  • Make sure that all results for the main listing of each weapon type are done using the same strength.
  •  ?

Information thresholds

See Research:Judge/Information thresholds for research on how much information is presented when the judge command is successfully used.

Quality

See Research:Judge/Quality for research on the characteristics returned by the judge command.

Relationships

Some interesting, and rather conclusive graphs:
Att-vs-wgt.gif
Par-vs-wgt.gif
Spd-vs-wgt.gif
As you can see, there are direct relationships between weight and hands used for:

  • Speed
  • Ease of parry (parametric rating)
  • Ease of attack (parametric rating).

The following equations fit the current data to +/-1:

  • speed (rating) = floor(15.5 - 0.1 * (9 * weight * hands)), capped at 13.
  • attack (parametric) = floor(146.67 - 2.955 * (9 * weight / (1 + hands))), no cap (yet).
  • parry (parametric) = floor(32.5 - 2 * (9 * weight / (2 + hands))), capped at 23.

(You can combine the parametric results with strength to get the ratings - see {{judge-attack-var}} and {{judge-parry-var}})

I'd suggest the caps above are specific to the judge command, not the real weapon's performance.

--Chat 15:58, 23 July 2011 (UTC)


Wait, if we can determine speed, ease of attack and ease of parry from the weight and the overall judge as a function of the other judge info could we avoid having to do all the judges at different strength? We'd just need the precise weight, max and average damage and everything else could in theory be calculated from those.

More basically we could copy the data from another weapon of the same weight and fix the overall judge (I'm too tired to try to figure it out but it's probably some kind of weighted average of the others).

Even simpler, a weapon with the same weight, max and average damage should judge exactly the same (as observed I think). --Frazyl 08:25, 7 August 2011 (UTC)

Eventually, yes, but I think we need to sort out the following first:
  • Accounting for the +/- 1 discrepancy - the formulae above are approximate.
  • Sorting out a formula for 'overall' - I know it's dependent on more than weight/strength/hands (probably involving max/ave), but I've not been able to come up with a sensible formula yet.
We don't want to prematurely move to a single-judge system, as we'll lose all the useful data that could be used to work out the above formulae.
More generally, though, I wonder whether we ought to be strictly tieing weapon info to the output of the judge command at all:
  • We can produce a more accurate 'attack' rating than judge can.
  • Judge's 'parry' rating is, frankly, wrong. I've now done some testing on parrying and I can confirm that heavier weapons are better on the whole, in line with the theory (see Research:Parry for the theoretical parrying relationships).
  • If parry's wrong, then that probably invalidates 'overall' too.
--Chat 10:18, 7 August 2011 (UTC)
On the parry thing, I think the main problem there is that you can't know what weapon your opponent is using so that's why judge is the way it is. I do know that if something says its worse than another for parry and they're both the same weight or close, the one that is worse for parry does parry worse.
I've also been talking with some people who are heavy users of GFR and it seems that the length of the weapons may play a factor in parrying as well since with GFR the same weight can be obtained for two different weapons of different lengths and parrying capability and according to those people that use GFR, longer weapons seem to parry better for the same weight than shorter weapons.
--Baldarov 11:09, 7 August 2011 (UTC)

Speed anomalies

Arg. I tried to clean up {{wr-speed}} by removing unseen ratings but I found another unused one after that. Then I made Research:Judge/Speed to try to see if the levels really correlated to weight*hands.

Right now there seems to be 9 levels that have weapons in them, including the "fast" level which is entirely made of anomalous weapons.

There's lots of other anomalous weapons but excluding those and the "fast" level the other levels map gradually from 0 to 35 lb.

string rating weight*hands range
extremely slow 1 22 2/9 - 35 5/9
very slow 2 18 6/9 - 20
slow 3 15 1/9 - 17 7/9
slightly slow 4 11 1/9 - 14
rather quick 5 8 - 10
quick 6 5 5/9 - 7 7/9
very quick 7 4 - 5 3/9
very fast 9 2 7/9 - 3 8/9
extremely fast 10 1/9 - 2 6/9
fast 8 2 2/9 - 3 3/9

I checked Tachibo (without long blade) and the chisel to see if the data was correct but it is.

So, not sure what to make of this. Maybe some weapons have been tweaked to have a speed different than the weight would make it, but the "fast" level doesn't make much sense to me. --Frazyl 00:04, 4 September 2011 (UTC)

Low weight weapons

There seems to be a point at which weapons under a specific weight (mostly 6/9 or lower, maybe 7/9) give some odd results for parry. A good example of this is knife (custom weapon). As can be seen, it doesn't follow the usual pattern for parrying (14 13 12 12 12 11 ...) and is instead just (13 13 13 13). This has been bugrepped but I can't say that anything will come of it any time soon.

It may not even be a bug but another issue with something else in the MUD. --Baldarov 20:27, 23 July 2011 (UTC)

Ok, it may not be weight related but instead an issue with Judge as twisted wooden staff also has the same issue. Or it may be some property of the weapon too. --Baldarov 09:59, 5 August 2011 (UTC)