Wiktionary talk:Votes/2010-02/ToC format

Web traffic analysis

cumulative page views per number of headers

The number of headers on a page influences how likely someone viewing the page with LHS ToC will be able to see a definition without scrolling. I analyzed the traffic stats for 2009-08 which included around 880 of the most common main namespace pages. From that I graphed the cumulative distribution of page views per number of page headers. This indicates the percentage of page views on pages with at most a certain number of headers. For instance, on a common setup (1280x800 monitor with Firefox) the ToC for a page with about 12 headers will push the definitions off the first screen. Looking at the graph we can see that 75% of views are of pages with have at most 11 headers. This means that about 25% of views are of pages that, with high probability, will not have a definition on the initial screen. Using the graph one can make determinations about modified viewer experiences. I think it's unacceptably high that around a quarter of page views will not have a definition on the "landing" screen and is the main reason why almost any of the options are preferable to the default. This is especially true since as time goes on, pages get more and more headers. --Bequw → τ 04:32, 24 February 2010 (UTC)

Option 1.5

Why isn't there an option for "table of contents being above the entries, uncollapsed, on the top-left, showing only language headings, with the numbers being hidden"? --Vahagn Petrosyan 07:49, 25 February 2010 (UTC)

I didn't put that in simply because I didn't think it had a realistic possibility of winning. Do you want to add it in as option 7? We're only on the first day of the vote and notifying PK and Bequw of the change wouldn't be too difficult. --Yair rand 07:56, 25 February 2010 (UTC)

Yes, can you add it as option 7? I can't make screenshots. There is no need to notify PK and Bequw, they'll see it. --Vahagn Petrosyan 08:07, 25 February 2010 (UTC)

page-dependent

Using CSS, there seems to be know way to make the choice of option page-dependent (within the main namespace). Am I correct? (Something with many languages, each with only one part of speech, might benefit from a languages-only TOC, while something with a lot of English and nothing else might benefit from a TOC showing POS.)​—msh210℠ 17:00, 25 February 2010 (UTC)

Something with javascript could surely be done. We might want to show etymologies when there's more than one (to distinguish homonyms). --Bequw → τ 18:11, 25 February 2010 (UTC)

On further thought, this might be a popular option. Do we want to add voting options for this type of ToC (left non-floating, left floating, and right floating)? If it is popular, I guess it's better to add it sooner rather than later. We could leave it a bit general (something like "ToC with page-dependent depth/detail") and work out specifics if it's the most popular. Linking to sub-language headers is not stable with our ELE so I don't think we need to worry about the fact that the displayed ToC for a page could change as the page grows in size. --Bequw → τ 23:19, 25 February 2010 (UTC)

option 6

Does option 6 mean "keep TOCs as they are now, but collapsed by default"? It doesn't specify.​—msh210℠ 18:56, 25 February 2010 (UTC)

Yes, that's what it means. I don't think there's really much possibility for confusion on that though. What else could it be assumed to mean? --Yair rand 20:06, 25 February 2010 (UTC)

Right, I was just making sure.​—msh210℠ 20:08, 25 February 2010 (UTC)

Result

To get an idea of the result of the vote, I have used the following online calculator: Ranked-ballot voting calculator, Washington University, St. Louis.

Without the use of a tiebreaker, Schulze method gives a tie: "a, d, e and g lose no beatpath comparisons" (a=1,d=4,e=5,g=7). Given "the ranking d>f>e>a>g>c>b was used as a random-ballot tiebreaker", the calculator has determined that the winner according Schulze method is option 4, labeled "d" in the data below.

The data I have fed into the calculator (11 votes):

1:c>e>f>g>d>b>a
1:e>d>c>g>b>f>a
1:b>a>f>d>c>g>e
1:a>g>d>e>c>b>f
1:a>b=d>f>g>c=e
1:g>e>d>f>a>c>b
1:d>f>e>a>g>c>b
1:a>c>g>b=d=e=f
1:d>e>c>b>g>a>f
1:e>a=b=c=d=e=g
1:f>a=g>c=e>b=d

I have obtained this data by mapping 1 to a, 2 to b, ..., 7 to g, from

3>5>6>7>4>2>1
5>4>3>7>2>6>1
2>1>6>4>3>7>5
1>7>4>5>3>2>6
1>2=4>6>7>3=5
7>5>4>6>1>3>2
4>6>5>1>7>3>2
1>3>7>2=4=5=6
4>5>3>2>7>1>6
5>1=2=3=4=5=7
6>1=7>3=5>2=4

The pairwise matrix:

		against
		a	b	c	d	e	f	g
for	a		6½	7½	5½	5	6	6
	b	4½		2½	3	2½	6½	3½
	c	3½	8½		3½	4	6	5½
	d	5½	8	7½		5½	7½	5½
	e	6	8½	7	5½		6½	5
	f	5	4½	5	3½	4½		5
	g	5	7½	5½	5½	6	6

--Dan Polansky 14:30, 6 February 2011 (UTC)