I have looked at the validation option on test, and want to make the following remarks:
* Can the categories be changed/extended? Outside WikipediaEN (maybe de), suitability for 1.0 is not an interesting point, and legality is I think not an issue to do via page validation (most pages are fully legal anyway). On the other hand, I would like to add options as "neutrality" and "balance" (although perhaps only 'neutrality' might also be enough); also, the possibility to specify the validator's own knowledge of the subject would be good (if an expert considers an article "solid as a rock" or "extensive" that means more than when someone off the street does so).
* Putting suitability for 1.0 in a single binary choice is too simplistic. There are different reasons to consider an article unsuitable, and various in-between possibilities. I can think of at least: ** Suitable ** Subject unsuitable ** Subject suitable but article unsuitable ** Suitable in shortened form ** Suitable after improving style/completeness/factuality/neutrality
* Comments given with the validation seem to be discarded; at least I cannot find them back
* Would it not be better to change the validation statistics. Those percentages and numbers seem to be more confusing than helping, whereas getting the separate validations (instead of just the totals) would be interesting. I am thinking of something like (I hope it looks a bit ok in everyone's favorite font...)
version blabla | validation 1 | 4 | 3 | 4 | 2 | 0 | 1 | | validation 2 | 3 | - | 2 | 4 | 1 | 1 | | validation 3 | 4 | - | - | - | - | 0 | | average | 3.7 | 3.0 | 3.0 | 3.0 | 0.5 | 0.7|
* Nicks of validators are not given; is this on purpose? I can see both advantages and disadvantages. Advantage is that one will be less inclined to give a better judgement because of personal reasons, or to attack the messenger. Disadvantage is that one cannot check whether a high or low rating is real or just created by sockpuppets or such.
Andre Engels
Andre Engels wrote:
I have looked at the validation option on test, and want to make the following remarks:
Disclaimer: I wrote the validation feature some time ago, and haven't touched it since. I'm surprised it is still working at all...
- Can the categories be changed/extended?
Yes. There should be no categories at all when the feature is turned on at a wiki(pedia). Adding categories requires SQL, IIRC, but is otherwise quite easy. Deleting categories, or changing the "value range", not so much.
- Putting suitability for 1.0 in a single binary choice is too
simplistic.
My though was to have a machine-readable yes/no decision, so a snapshot can be taken automatically at any point. The *reasons* for the yes/no decision can be set with the other categories. But, it was only a demonstration, and when/if we turn on the validation feature, that whole schema is likely to go to heck(TM) anyway.
- Comments given with the validation seem to be discarded; at least I
cannot find them back
As said above, I didn't keep it up-to-date, and AFAIK noone else did either. I'm too busy with other things to invest large amounts of time in that right now, too :-(
- Would it not be better to change the validation statistics. Those
percentages and numbers seem to be more confusing than helping, whereas getting the separate validations (instead of just the totals) would be interesting. I am thinking of something like (I hope it looks a bit ok in everyone's favorite font...)
version blabla | validation 1 | 4 | 3 | 4 | 2 | 0 | 1 | | validation 2 | 3 | - | 2 | 4 | 1 | 1 | | validation 3 | 4 | - | - | - | - | 0 | | average | 3.7 | 3.0 | 3.0 | 3.0 | 0.5 | 0.7|
Of course that can be done, but it would just hide the percentages and show lots of detail data instead.
- Nicks of validators are not given; is this on purpose? I can see
both advantages and disadvantages. Advantage is that one will be less inclined to give a better judgement because of personal reasons, or to attack the messenger. Disadvantage is that one cannot check whether a high or low rating is real or just created by sockpuppets or such.
Originally, I planned to give access to the "detail view" to sysops only (privacy vs. sockpuppet fighting, as you said). I can't even remember if I coded that or not. Probably not.
Magnus
P.S.: I forget a lot these days. What's my name again? ;-)
wikitech-l@lists.wikimedia.org