User talk:RolandOfGilead13141

How do you know all that skill data for level 18? are you testing it at the proper shrines? Foo 08:34, 9 September 2006 (CDT)
 * Oh ok, now that I see the attributes for the level 18 and 20, it seems you do. gj. Foo 08:36, 9 September 2006 (CDT)
 * No I don't gather hard data for the highest 2 levels. However, skill effects seem to follow a strictly linear curve depending on its linked attribute's level, taking GW's rounding method into account. For levels 0-16 I have tested this theory extensively and have not yet found an exception to it, so I have full confidence that the theory is correct and can be applied to levels 17/18 and 19/20, respectively. RolandOfGilead 08:49, 9 September 2006 (CDT)
 * ok, that sounds better then just guessing numbers. can you say more about it? Foo 09:06, 9 September 2006 (CDT)
 * Sure. The rounding method applied in GW is: x.5 and greater is rounded up, otherwise down. For example, 5/2 = 2.5 = 3 in GW. There are exceptions, but AFAIK they are alwas explicitely stated.


 * The linear curve I mentioned is the basic equation describing a line in a Cartesian coordinate system:


 * y = mx + c, or, for our purposes,


 * effect = effectIncrease*attributeLevel + offset


 * where effectIncrease is the increase per level and offset simply the value at level 0 (but before rounding, so offset can be a fraction, even though a fraction is never displayed to the player).


 * For example, if the skill says "target foe is poisoned for 5-17 seconds", then effectIncrease = 1.0 and offset = 5.0. This is trivial if the increase is a whole number. But what about fractions?


 * Example: Life Siphon effect duration, from level 0 trough 16, and the increase per level below it:


 * 12 13 14 14 15 16 17 18 18 19 20 21 22 22 23 24 25


 * N/A +1 +1 +0 +1 +1 +1 +1 +0 +1 +1 +1 +1 +0 +1 +1 +1


 * Do you see the pattern in the lower line? The increase is always 1/1/0/1/1. You could also say it is 0/1/1/1/1 or 1/0/1/1/1, as long as the pattern returns without fault. Either way, now we know that the effect will increase by 4 every 5 levels. Thus, effectIncrease = 4/5 = 0.8. The only thing left to do is to find out the exact value of offset, which can be determined by what my Maths teacher would've called "a sharp look". In this case, offset is 12.0. Now we can calculate the duration of Life Siphon for an arbitrary level. Hope that helped. RolandOfGilead 10:01, 9 September 2006 (CDT)


 * Hehe, maybe I should have said I'm a mathematician first. it is still a guess, an assumption that the line has nice and round values. how short should a pattern be so that you will consider it reliable? there was a discution about it already, and I think it said that we will only put in values confirmed in-game. (energy cost, after experties and other effects is always rounded down). Foo 11:40, 9 September 2006 (CDT)
 * As long as we don't get to see the source code of the game, every formula and every rule with predictive power the community deduces about game behaviour is an "assumption", unless confirmed by ANet, so "assumption" is not a very useful term to begin with here. But, staying with the term, all assumptions are not alike; there are good ones and bad ones. The good ones are founded on large amounts of data, can explain all data and are not contradicted by even one piece of data.


 * The theory of strictly linear skill progression is a good assumption in the above sense, because it fulfills all these criteria. Feel free to disagree, but to the best of my kowledge (and I have put a lot of time into this), every single one of the hundreds of skills in GW follows the linear curve in levels 0 to 16, which is 80% of the 0-20 range which currently matters. I do not exaggerate when I say that I have checked this behaviour for every (!) unique skill range in Prophecies skills, i.e. 5-17, 20-50, 1-10, etc. etc. ad infinitum. If you want proof, I can upload a csv-file which is readable by the spreadsheet program (excel, OOo calc, etc.) of your choice. As long as there is no shred of evidence contradicting the theory, it would be plain unreasonable to dismiss it in favour of dull data collecting without generalization, just because "it could be different just 1 level further". Yeah, but so what?. All evidence points towards linear progression without any deviation and none speaks against it, so why not go with it? If it should turn out wrong later, ok, let's dump the theory with or without replacement, that's how it goes, no problem at all. But until then, this theory provides an equation which not only correctly describes all known skill values, but also has potential predictive power for non-observed level ranges. Who is afraid of a good assumption which has a non-zero, but imho small, chance of being wrong? RolandOfGilead 13:03, 9 September 2006 (CDT)


 * Progressions are linear through the values for 0 and 15, as in a line through (0, f(0)) and (0, f(15)). Round non-integer results.  Using 0 and 16 doesn't work (Aura of Restoration is one example). See Talk:Skill_details/Archive.  --Fyren 15:46, 9 September 2006 (CDT)
 * Ok, I stand corrected, thanks for discussion. You could've saved me all that by just linking to the formula :) Anyway, at least I was right about the linearity of the function, because the one in Skill details is linear as well, only the coefficients are a ittle different. For example, for Aura of Restoration the equation is y = 152 + x*16.5333333, versus my y = 152 + x*16.5. Obviously the difference is very small but big enough to be noticed, if only in a very few cases. Also, if (v15-v0) is a multiple of 15, then my approach gives the same results, so my method is really a special case and an approximation. RolandOfGilead 17:42, 9 September 2006 (CDT)


 * I'm not Foo (if you're mistaking me for her, heh). I just happened to see your discussion with her so chimed in.  I also have some wiki templating written up that auto-generate progressions, but I don't know if we're going to use it.  See, for example, Sandbox/Symbiosis and the source for it.  The values for 0 and 15 are used to generate the progression table automatically, so hopefully we can ditch all this manual progression data crap.  (We didn't do this from the start since ParserFunctions are needed to do the math, and PF weren't installed till a few weeks ago.)  --Fyren 18:12, 9 September 2006 (CDT)


 * "her"? how did you get to that conclusion? Foo 18:20, 9 September 2006 (CDT)
 * If you insist, I'll use "it" in the future. --Fyren 18:22, 9 September 2006 (CDT)
 * :O Foo 18:25, 9 September 2006 (CDT)
 * Hm guess my brain went into auto-mode when it saw a short name starting with F hehe. Anyway, the Sandbox template looks good, if not complete. What's missing? BTW, yeah I know we didn't have that. I used to be user:Roland of Gilead, but my password got lost after a longer break...Skills have always been my favorite topic :) RolandOfGilead 18:43, 9 September 2006 (CDT)


 * The template is "complete," but see GuildWiki talk:Sandbox/Skill box. The gist is I changed some things to how I personally think it should be but I haven't discussed the changes with anyone, really, so I'm not going just to switch all the skills over.  --Fyren 18:53, 9 September 2006 (CDT)

New window
It's only temporary for the event; it'll be gone till Nightfall is out. So, I dunno. Don't make so many changes that you can't change it back easily. --Fyren 09:49, 22 September 2006 (CDT)
 * Do you know that for certain? Did they announce they will revert the interface changes? Seems pretty pointless to me, but if you got solid info, I'll revert it RolandOfGilead 09:52, 22 September 2006 (CDT)
 * It's in the patch notes. --Fyren 09:53, 22 September 2006 (CDT)
 * Ok, then I would leave the skills and attributes article as it is and let it stand as long as it is correct, i.e. for the duration of the NPE. When the NPE is over, I'll just bend the links back to the skill inventory article as it was 30 minutes ago. Then, when NF goes live, we can again just bend the links to skills and attributes window. Sound like a plan? RolandOfGilead 10:05, 22 September 2006 (CDT)
 * I'm okay with whatever, as long as you're doing the work! --Fyren 10:07, 22 September 2006 (CDT)
 * By the way, Skuld left a 'higher standard' notice on the Skills and Attributes Window page but apparently he doesn't want to tell me what's wrong. Could you comment on what's wrong then? Also see the discussion page. RolandOfGilead 10:16, 22 September 2006 (CDT)

Redirects
GW:REDIRECT says to not make redirects for different cases of article names. --Fyren 12:18, 26 September 2006 (CDT)
 * well sorry, didn't think this is against official policy. Still, it seems such a pointless policy, because searching for an article without capitalization yields unpredictable results. Sometimes it leads directly to the corrrectly capitalized article, which is fine. Usually you get to the search page with the correct article being the #1 hit; still acceptable, but requires another click and page load, and GW isn't the fastest site out there most of the time. Rarely, you won't even get that and the one searching has to assume the article doesn't exist for some reason or whatever else, which is pretty bad and counter-productive imho. At least the last case should be caught. RolandOfGilead 12:28, 26 September 2006 (CDT)


 * I'm just telling you why I deleted it, since you wouldn't have made it if you knew. --Fyren 12:43, 26 September 2006 (CDT)
 * No offense taken. But what do you say about the policy? After skimming GW:REDIRECT (edit:aha, capitalization error :), the discussion there seems to revolve mainly and repeatedly about capitalization, so wouldn't another discussion started by a de-facto newcomer just be annoying? RolandOfGilead 12:46, 26 September 2006 (CDT)


 * In my comment on the talk page there (that I posted while incognito as 68.142.14.19) I say that I'm ambivalent. For all the cases we don't allow, Wikipedia does in order to help searching.  Better searching is obviously good.  But it'll also mean tons of people will link (within the wiki) to those redirects not knowing which is the real article and that's a pain both because they'll use the wrong capitalization and because they should be linking to the real page and not the redirect (it just happens that both would be fixed at the same time with the right link).  --Fyren 13:09, 26 September 2006 (CDT)
 * Is this a pure style question then or am I still missing something? Assume someone links to signet of might instead of Signet of Might. If the former has a redirect to the latter and someone klicking on the former gets to the article, would that not serve the link's purpose, namely to provide a shortcut to the intended article? Obviously, forcing capitalization in wiki links makes the text more readable and is desirable, but it harms the search feature, so I can see a certain conflict of interest there. However, fixing style in articles (i.e. correcting capitalization even though redirects exist) seems easier and more to the wiki spirit than fixing the search function. RolandOfGilead 13:25, 26 September 2006 (CDT)
 * I suppose it's style issue where they decided to protect style and make searching a little more difficult. I didn't follow that discussion and vote closely (or the older ones, which I'm not sure where to find off the top of my head; this has been brought up a few times).  There's GW:ULC which says to use lowercase/normal English case for everything but names of things in game that ANet capitalizes, so "signet of might" should not be used anywhere in an article, whether link or just plain text.  That issue I'm not ambivalent about: ANet's capitalization sucks and it pains me to write things capitalized like they do, whether skill names or whatever. But more than a year ago when I first argued against it I got shot down.  If ULC didn't exist as it does I would want to allow all-lowercase redirects, at least.
 * Specifically about search problems, it only breaks in the case of a title that contains a word starting with a lowercase and a word starting with an uppercase that's not the first word. So if you search for "blacktide den (mission)" you end up at Blacktide Den (Mission), but stuff like "signet of x" breaks since the "x" is capitalized in the article name while "of" isn't.  In most cases, the right article is the first result.  I'm not sure why "signet of might" isn't, I don't know enough about how MW's search works at the moment.  --Fyren 13:39, 26 September 2006 (CDT)
 * Exempting only article titles from ULC for the sole purpose of lower-case redirects is something I would consider a small change but allow for fixing working around the search function misbehaviour. That would be a great benefit for a small compromise. But anyway, if the consensus says otherwise, well, it's still nonsense but I'll get over it. RolandOfGilead 13:56, 26 September 2006 (CDT)