Wikisource talk:Changing the main page

Criteria mix

Hello, I agree that the Article count is suboptimal. The number of pages in the page mode is a better approximation of work done by contributors, but I would have reservation to use the nomber of uncorrected pages and only that, as it doesn't show the real work done, only how much bots work ;o).

So I see several possibilities:

1. . number of proofread pages;
2. . number of validated pages;
3. . a mix from the 3 numbers, f.e. (number of validated pages X 4) + (number of proofread pages X 2) + total number of pages

Yann 18:46, 11 March 2009 (UTC)

Yes, I seem to support some sort of "mix". Just today there has been made a proposal like this, see Wikisource:Changing the main page#Complex criteria which looks good and I would like to see it - after discussion - as the new criteria. -jkb- 12:56, 12 March 2009 (UTC)

P.S. @ Dovi's edit "moderately complex criteria": yes it seems to me that the criteria (or more criterias) concerning proof read pages play a great role. When we include this criteria so the weight of many subdomains (not only the small ones) will be smaller as they do not have the proof read (and cannot get points). On the other hand if we exclude this criteria (which is the one of the most important quality - not quantitiy - criterias) so such subdomains that do it and that support the quality (de, en, fr, ...) will be strongly disadvantaged. Therefore I think we should have a quality criteria in the mix anyway. -jkb- 18:38, 12 March 2009 (UTC)

The problem with complex sets of criteria is that they go against the keep-it-simple principle. How much weight to give to each criterion gives the opportunity for endless debate.

While I have no problem with a strict alphabetical order of languages, I readily concede that as the number of listed languages grows it becomes impractical to bury the most important languages (by whatever basis for selection) in a long list of options that are only seldom used.

Who is benefited by page views? Does the average non-contributing user care about this? I think that he would be more interested to know how much content is available in his chosen language. This criterion may only be a benefit to those who want to see how their efforts are being received; this is of far more benefit to contributors than to users.

Failing any scheme based on strict alphabetical listing, I very much believe that only some measure of content makes sense, and only in the main namespace at that. This could include the author namespace in recognition of the fact that some projects have merged the two. Proofread and/or validated pages seem unduly artificial and bureaucratic; not everyone makes use of that technique or uses the page namespace. For me it is much easier to proofread a text from a hard copy than by side-scrolling back and forth to proofread. In any event, transcluding material from the page namespace should also tranclude the size of those pages. Total number of pages in the page namespace may indeed reflect bot activity, but pages still remain and intermediate step in building a work.

The deficiencies of the simple article count are clear in that they cannot distinguish between a whole book and a dictionary entry. I can't criticise the people at pt:ws for choosing a separate page for each dictionary entry. (I would be inclined to combine multiple dictionary entries on the same page, but that's a different story.) I would certainly not jump to the conclusion that they are doing this to inflate their page count. Other projects have the opportunity to deal with dictionaries in exactly the same way. They just happened to do it first. A page of dictionary entries linked only by the fact that they all happen to begin with the same letter doesn't seem very useful. A total mainspace+authorspace word count or byte count makes more sense to me. It still does not account for a French text being longer than its English equivalent, or for the compactness of Chinese. A statistically based algorithm could remedy this but that may just bring us back to the problem of complex criteria. Eclecticology 19:05, 12 March 2009 (UTC)

Isn't it usual to have an average figure for how many words are in a page, and then an average figure for how many pages there are in a book? If our formula is well calculated, this formula could become a standard everywhere in the web, couldn't it? The only problem is how to calculate this formula, it will demand a serious work together to take into account all the languages with the specificities of each one; but this very serious work has to be done only once. I'd like that we do it. As soon as we have done it, the formula would be a simple one, very easy to use, once and for all.---Zyephyrus 11:38, 13 March 2009 (UTC)

What about counting how many words are needed in various languages for, say, the Oath of Hippocrates? If there is a ratio in the number of words, can this ratio be used to calculate a formula applicable to all texts in the same language?---Zyephyrus 12:40, 13 March 2009 (UTC)

Number of words on a page won't work because it depends on the size of the typeface. Number of pages in a book is completely variable. War and Peace and The Little Prince are both novels, but their lengths are quite different. The ideas in your second comment make more sense. Biblical texts are some of the most translated works available, so a sample that is statistically significant could easily serve as a basis. I suspect that if one looks hard enough, he will find that some linguist has already worked these out. Eclecticology 22:14, 13 March 2009 (UTC)

Criteria?

I think a possible method is to rely on number of words or size of database of current articles. Number of pages could be given some weight to avoid dumping huge raw texts. Page views and community activity should be disregarded imo. --Obayd 18:53, 4 May 2009 (UTC)

Layout

Here is a proposal concerning layout : Main Page alt.

It is independent of the choice of a ranking criterion.

The idea is to simplify the layout and improve legibility. In addition, only the top 10 languages are sorted.

The idea may be good in my opinion but there is a problem with what the eye perceives visually: an indistinct block of information discourages the reading more than several repartitioned blocks, whatever the repartition is, so I think that some repartition of the block into several pieces would be better. --Zyephyrus 06:52, 5 May 2009 (UTC)