On 21/08/14 13:13, Quim Gil wrote:
On Wednesday, August 20, 2014, Legoktm legoktm.wikipedia@gmail.com wrote:
At Wikimania, we had a good conversation and a proposal (that was basically agreed upon by everyone in the room IIRC!) about creating a "gold standard" for extensions, judging them based on their code quality, compatibility with MediaWiki versions, tests, etc.
I was in that room and I agree that defining and implementing this "gold standard" is a good idea. For what I recall, factors considered included objective facts like compatibility tested with Jenkins, support for LTS, etc. Still, I don't see this as a substitution of user feedack, with its potential drawbacks but also with the uniqueness that such human feedback provides.
Whenever we have this type of discussion, Wordpress is mentioned as a good example of a catalog of plugins with information useful for Wordpress users (meaning Wordpress admins, just like in the case of MediaWiki). Look for instance http://wordpress.org/plugins/feedweb/
Requires: 3.0 or higher Compatible up to: 3.9.2 Last Updated: 2014-8-21 Downloads: 90,268
Something to consider about user ratings. Ratings with stars/whatever are most useful when you're looking at a condensed list of multiple similar <item>s because they add a metric that can be compared and sorted. Suppose you're looking at laptops - you've got a bunch of otherwise similar machines all within the same general hardware specs - so then you look at the higher-rated ones first. More stars? Go check the comments to see why they rated it higher.
But that only applies when you have many, similar items that do the same thing and you need to choose between them, and that just isn't generally the case with extensions. In many cases there is only a single extension for a single purpose, and if you are looking for an extension, you are looking for one that does specifically what you want. So you go find that one, and once you are there, the actual question at this point is 'does it work?', not if it's better than anything else.
So a better model here may be endorsements. Look at game mods as an example - Skyrim has an insane amount of mods you can install, with often surprisingly little overlap of funtionality (yes, there are different lighting mods, but they all do it a little differently, so it comes down to which kind of lighting you prefer more than anything else). Suppose you find a mod on nexusmods that does what you're after, you see it has 100ish endorsements, that means it probably actually works and does what it says on the tin. That's good. You look at the comments, nobody's complaining about it being broken or causing other problems, well, there you go. Good to go.
This seems much more feasible and realistic to our purposes, and I suspect the same backend as for the ratings could well be adapted to do this for extensions overall. Whereas skins, which all tend to be much more similar in their purpose, might actually make better use of the ratings as they are now.
-I