PvXwiki: How it works

Xenomortis

Xenomortis

Tea Powered

Join Date: May 2008

UK

N/

16 Jul 2012 at 22:20 - 41
Quote:
Originally Posted by Relyk View Post
This is a discussion of average, weighted scores, so speaking about integers can be somewhat of a misnomer. The threshold to keep a build is 3.75, it's completely possible for scores to be below three with enough votes. It only becomes that "two point" system for the minimum 5 vote requirement. The wiki previously allowed for scores where builds that were considered average were stored, the Other section. The problem is that builds that get vetted as 'average' don't get used. The only thing worse than trashing a build is storing a build that people theorycraft as average. There are plenty of great and awesome builds available on every profession, if someone comes to the site looking for builds, why would a choose one the a vetting process regards as average and there are builds considered superior in quality? That's what happened with the Other section, where builds there weren't used for years. A build carries with it a connotation that if it's good or great, that people will like it and use it. Does that mean that happens? Not all the time. But that's how consensus was reached for the build and I think gives a lot of weight to builds that actually get vetted. It's also a way to reduce systemic bias for individuals actually testing builds. Issues such as vote bombing are simply an abuse of the system and are more of an issue of users themselves than the actually system.
When I vote and I consider a build to be decent and above average, I essentially have two options: I vote 5 for Effectiveness or 4 for Effectiveness. If I vote 3, it's a trash (and a lot of the stuff that works is, technically, considerably better than average, but ignore that). Universality counts for bugger all and is typically high anyway since most decent stuff works almost everywhere.
You ask me to distinguish between the top tier stuff, the AP builds: the SoS Rit hero, the PR Derv build, etc, from the still good but not top tier builds, like the Panic Mesmer or the FD Mes/Nec, and the stuff that's not as good, but still substantial and more than functional (and can often get onto hero teams); like Elemental Attunement builds or non-MM Necro builds.
Even inside each of these crude tiers, there's a hierarchy.
For me, a lot falls above the 6/10 mark (a lot I still don't use myself), but I can't rate 7/10 and I'm stuck choosing 8/10 and 10/10.

I bring this up because I've received a fair amount of ire over this; with an admin simply telling me I'm wrong for not 4-5/5ing something.

Kunder

Desert Nomad

Join Date: Nov 2010

16 Jul 2012 at 23:29 - 42
Ideally there would be multiple categories of ratings. The final score would be a composite of the categories compared with the average score of the other builds for the class. e.g.:

Single target DPS
AoE DPS
Single player Healing
Party Healing
Protection
Utility
Resilience

Ideally we could have some handy benchmarks, either with absolute numbers or relative ratings compared to other well-known builds.

The problem at the moment is that just giving a score to a build requires no thought at all, which presents a very low barrier to retards who think their shit doesn't stink because they beat NM. I have a fanciful hope that in some way requiring players to score a build based on these attributes would force some critical thinking, and perhaps lead to a more constructive debate. A good number of the best builds will be hard to rate individually since synergy between builds is the ultimate goal, but it's a good start.

Unfortunately I have no idea how to improve team builds. The whole concept just seems to be a failure. As far as the player base is concerned the only metric of a team build's effectiveness is how dumb you can play and still survive. I don't see that ever changing, sadly.

Relyk

Lion's Arch Merchant

Join Date: Feb 2008

W/

17 Jul 2012 at 02:16 - 43
Quote:
Originally Posted by Xenomortis View Post
When I vote and I consider a build to be decent and above average, I essentially have two options: I vote 5 for Effectiveness or 4 for Effectiveness. If I vote 3, it's a trash (and a lot of the stuff that works is, technically, considerably better than average, but ignore that). Universality counts for bugger all and is typically high anyway since most decent stuff works almost everywhere.
You ask me to distinguish between the top tier stuff, the AP builds: the SoS Rit hero, the PR Derv build, etc, from the still good but not top tier builds, like the Panic Mesmer or the FD Mes/Nec, and the stuff that's not as good, but still substantial and more than functional (and can often get onto hero teams); like Elemental Attunement builds or non-MM Necro builds.
Even inside each of these crude tiers, there's a hierarchy.
For me, a lot falls above the 6/10 mark (a lot I still don't use myself), but I can't rate 7/10 and I'm stuck choosing 8/10 and 10/10.

I bring this up because I've received a fair amount of ire over this; with an admin simply telling me I'm wrong for not 4-5/5ing something.
People have different criteria for what's considered top tier. I consider Panic Mesmer top tier for instance. The rating system is arbitrary at best, people will still be unsatisfied even if you gain precision on a 10 point scale because other votes will vary significantly. I'm not sure how much more accurate then the current 5 point scale it would be really. I think a point you sort of bring up is that people tend to use universality to bump the score up and down already for that 7/10, which is more likely to irk you since it's "minor" vote balancing than actual voting. There was favorable consensus for doing so but nothing came out of it. The wiki currently doesn't really have the capability of modifying the rating system either. If it did change, all the builds would have to be re-vetted.

The one idea I would explore is lowering the requirement to 3.5 from 3.75 as criteria for a good build as the 5 point scale tends to deviate from the critical 7/10 greatly thanks to small sample size.