On Mon, 10 Mar 2003, Jimmy Wales wrote:
Tomos at Wikipedia wrote:
- English-centrism
I would say that if at all possible, within the parameters of trying to come up with a better and more reasonable counting system, that any changes should benefit (if possible!) the non-English wikipedias more than the English wikipedia.
I disagree. We should try to get the greatest benefit for ALL Wikipedias. What we should try to avoid is changes that benefit the English one but are disadvantageous to (some of) the others, but if we have to options, one of which benefits the English a lot and the others somewhat, and the other the English a little bit and the others somewhat, then the first one is preferable.
In this particular case, Brion's motivation is that the French has seen people adding commas for no good reason, right? I absolutely do NOT want the French to get the feeling that we're changing the rules in order to penalize them.
If I remember correctly, the French that are present at this list were just as negative about this action as the others were.
If you want another reason, the current comma-count underestimates the Japanese Wikipedia quite a bit because Japanese uses less commas than most western languages.
Regarding other issues mentioned here: 1. If we measure from some minimum size, I would like to put my vote at 100 bytes. 2. Voting seems like a good idea. My proposal would be to use two questions, the first what kind of counting system to use (keep the comma count, count all non-redirect main namespace articles, set a minimum size, anything else that gets proposed, do not count at all), the second what minimum size to use (taking the median answer, and counting people who do not fill in something here but say 'all articles' to question 1 as voting for 0). 3. Some languages have taken over the 'Wikipedia is not a dictionary' attitude from the English, others tend to welcome very stubby articles with only a dictionary definition. Both ways of working are valid, but I do think that the 'dictionary' encyclopedias tend to appear 'too large' if we use article count - especially if we do it to all articles, or with a cutoff at only 6 or 10 or even 100 bytes. I would therefore like to propose to extend the existing statistics by _also_ making a list of the sizes by total article _size_ rather than total article _number_.
Andre Engels