Wikipedia:Graphics Lab/Resources/Research and Development

Reading the above comments about standards gave me an idea of one area where we could easily standardize (And I didn't see this already dealt with on fR, but I don't know French): Representative distributions. Currently this is a mess, EG:

• 
  The Canadian seating chart idea
• Japanese election style, 1 dot / one deputy system. Legend include in the SVG.
  Japanese election style, 1 dot / one deputy system. Legend include in the SVG.
• A seating chart form
  A seating chart form
• 
  This more common version (May be worth getting a copy of the software that generated it)
• 
  Japanese circle style.
• 
  French presidential election, 2007 1st.svg
• 
  Alaska RaceComposition2005 PieChart.svg, 3D view.

Does anyone think that it would be worth picking a standard of display of representation and slowly converting everything else to it (As well as, hopefully, having the new electoral images follow it)? Thanx, 68.39.174.238 (talk) 18:56, 11 January 2008 (UTC)

Yes, but the French already working on standards for pie charts. Another is that such standards should be the same for 3 labs. A Babelfish [1] translation of the page will give you an idea of what they're talking about. We should at least try to get an English version of the existing French standards. Some would likely be adopted, others, we will want to set the standard. Sagredo^{⊙☿♀♁♂♃♄} 19:16, 11 January 2008 (UTC)

That's the thing though, do we want electoral representation images to be pie charts? The Japanese idea is interesting, especially since it can be used to show what parties have the relevant (super)-majorities. Granted, we could do that with a semicircle as well, Do you think we could do that? Have, say, a SVG chart like the Czech one, but with the constitutionally specified majorities and supermajorities indicated on it? I'm thinking that something like the United States Senate would end up like: A 100 member house, with the divisions at 49, 51, and 100 (Democrats, Democrats + caucusing Independents, Republicans), and divisions indicated at 50, 60 and 67 members, counting from the current majority outward. Also, I suggest that the majortarian party/coalition be on the left and the various majorities (½, 2/3, etc) all be based on that side. I wish I could make images to show you what I'm thinking of. 68.39.174.238 (talk) 07:39, 12 January 2008 (UTC)

Current Proposal

Note

This standard's proposal is now at "90% set up", your comments are welcome in the "Comments" section. Then, set the last 10% will need discussions/critics, images creations, and true tests which will ask lot of work and time too. Yug   111: Liberal Democrat Party.   24: Clean Government Party.   82: Democratic Party.   6: Socialist Party.   4: People's New Party.   4: Independants.   2: Independants.

Here is a draft proposal, listing the mains parameters of this kind of charts, and giving my opinion about each of them.

Semi-circle and circle system : I don't really like semi-circle or circle. Why ? Because despite to be really clear, they have to show roughly two informations : the right number of deputy by color + the area of each color (in % of the circle). And I think that this last points is difficult to edit on a SVG, difficult to check, and so may be falsified (i.e. a pro-republican may display the republican area bigger than is true %).

refused. Not convenient untill we don't have an easy way to make and check that each color area is conform to it's pourcentage.

One deputy/One colored point system: I like more this system. Because : 1/ editable by every SVG newbie, 2/ quickly checkable by everyone, 3/ easily expandable, 4/ the % can perfectly be remove to the Image:page.svg and in the article.

Y proposition supported : seems to me to be the best way to be easily makable, clear, and checkable.

Displaying the Assembly shape system : seem useless. That's more difficult to provide/draw, is different for every countries, and make then like that will ask us lot of work without comparative gain in the message.

refused.

3D view VS 2D view (circle pie charts) : don't seems interesting. It look like one more difficulty adding to already existing difficulties link to 2D pie chart.

refused.

Colors ? : I prefer soft colors such display in the 1st and 4th images. An SVG displaying the list of the 12 (?) recommended soft colors should be create.

Y proposition valided : don't seems controversial. Stay to create an SVG listing the 12 (?) soft colors recommended, please wait the first attempt.

Include in the Image : country name or province name, years of vote ; total number of seats ; number of seat for each group/color ; % of each group (this is it really need here ?), exact day when the data was pick up such as "YYYY/MM/DD" (where to put this) . Do do not include (not forget) : groups' names (need on the image ? may be display on the legend using {{legend}}), explanations (should clearly be in the article)

Y proposition valided : seems to me the best optimization. To see clearly where put what, please wait the first attempt.

Exclude from the Image, include in the article : Legend explaining what means each color using the template {{legend}} : {{legend|#0000ff|Republican}} =

  Republicans (127 seats; 43%)

, with the number of seats and % of each groups.

Y proposition valided : seems to me the best optimization.

Naming convention : the name should contain all the following data : Country name, election name, year, language code (? not need if we agree to always produce free-of-text charts), and round number.
A proposition of recommendation, underlining the Country :
Countryenglishname_YYYY's_ElectionName -languagecode.svg , i.e. : Japan_2007's_House of Councilors election_1st-en.svg.
If you have a doubt for these names just check Wikipedia's article, i.e. Wikipedia use fluent names such as "French legislative election, 2007". But this should be put into basic english such as : France 2007's legislative election 1st-fr.svg (Taiwanese -> Taiwan ; Kenyanese ? Kenyan ? -> Kenya 2007's ...).

? proposition : seems to me the best optimization for international use -> validated, except if this name seems really ugly to native english speakers' ears (then -> French legislative election, 2007 1st-fr.svg, based on article's name.).

For standard creation, we have to keep in mind: that the image should be perfectly clear, able to say quickly and clearly the facts. In Wikipedia : the illustration will be accompanied by a legend, an article, etc , but the image will also be print by high school students, and so the image file need to bring the basis to be a good start point of a presentation. 1/ the filename provide the context Japan 2007 House of Councilors election-en.svg , 2/ the image provide the data : year ; total number of seat ; seats by group ; % by group , 3/ yes, we expect from our "high school student" to be able to read file name and to remember the topic of his speech... 4/ but what provide the groups names ? I think we should also expect from our high school student to remember the image's legend and articles comments.

Yug 13:19, 12 January 2008 (UTC)

Comments

Pie chart, or list of colored dot ?

This current proposal finally encourage to avoid pie charts. Despite their good looking, they don't offer enough guarantees of accuracy to be raise to the level of a recommendation. I'm myself a little disapointed by this conclusion, but I had to make a choice between "Good looking" and "Easy to do, to check", I chosen the second one. If you know a way to color disk on Inkscape into pie charts by seting the % of each part, they please tell it. Yug. 17:55, 12 January 2008 (UTC)

Could I ask why some were arbitrarily marked "rejected" and how this is "90%" done? I think 3 people (Myself, you and User:Sagredo) have commented on this: far to few to be able to arbitrarily declare a standard and certainly not enough to "reject" proposals out of hand which obvious have some merit (They're being used in at least one articel). 68.39.174.238 (talk) 21:26, 12 January 2008 (UTC)

How to set up an optimized recommendation ?: I use here my experience of Wiki-fr. In summary : talks don't match action. Vote are often huge lots of time for those who vote, or for the one who make the proposal. Also, despite they are not official rules, the French GLab de facto work since 3 years on the following way :

Stage 1: a new need appears ;

Stage 2: several users start to notice that an optimized recommendation may be welcome ;

Stage 3: an user who understand the needs and the difficulties and who feel skilled enough to manage perfectly it take the issue in charge ;

Stage 4: he study and discute -alone- the issue : Listing the sub-points, self-discussion of each sub-points such as "interest + difficulties -> comparative gain ? -> propose to reject/keep this or this solution ; This can be visible to everyone or not ; In this stage, he can also receive comments, but he keep the lead.

Stage 4: this user improve his proposal as much as possible -alone- to then be able to explain why he made each choice ;

Stage 5: but it always stay some doubts (comparative gain of 2 possible solutions unclear), so he show the proposal to the community, received and listen fairly their comments, and compare them ;

Stage 6: choice have to be done, he state which solutions he have finally keep : A new recommendation is ready. Other graphists (who were looking at the development of the recommendation) can now refer to this recommendation to make their works.

You can try to use an other systems, as I unsuccessfully tried, but votes says nothing clearly, talks are most of times far longer for not more efficiency, and if I should wait vote or endless talks for such technical choices to do, then I guarantee to you that I will never do a second recommendation for the Graphic Lab. I'm not crazy, I prefer drink a good "Chocolat chaud" than choice a way which is wasting my time.

The issue have to be study deeply, efficient choices have to be done, add-on can be submit in the Stage 5 : that's for me the best way to work.

Yug 14:00, 13 January 2008 (UTC)

I'm not (for obvious reasons) really involved with the means of consensus here, but I think that's a French system and not a universal one. The rough and ready blanket-primary-of-systems we have here seems to be working (Especially since this doesn't involve major wiki politics like pagemoving, etc.). Once a standard is adopted, it should be fairly easy for everyone (here) to follow it. 68.39.174.238 (talk) 16:27, 13 January 2008 (UTC)

1. It's certainly time to start the page Editing Wikipedia:Graphic Lab/Research and Development and move this there. Perhaps put a note on the appropriate German and French pages, even if it's in English and on the images to improve page, everyone should be welcomed to make comments on the English page any language. I think English will prove to be the most universal, but with any luck we'll get translators as well. Perhaps we'll end up with a trilingual page.
2. Slashme is a chemical engineer, and he has/knows about software which will generate all sorts of charts from raw data. Fvasconcellos may very well will know something as well. I left notes on their talk pages.
3. About all I can put in is the layman's point of view. And that just for the U. S. I think the public is very used to circular pie charts for things like budgets. Seating charts for things like legislatures. I really don't have the technical knowlege to keep up with this. I did study engineering, but that, too was a while back. [2] Sagredo^{⊙☿♀♁♂♃♄} 23:53, 12 January 2008 (UTC)

I have had a look at the alternatives, and I strongly support the one-representative one-dot system. It gives more information at a glance than the pie chart: The pie chart shows you the balance of representation, but the dot-chart also shows you which parties only have one rep or two reps in sitting. The seating plan is cute, but I agree that it's not worth the added effort, and is inefficient in terms of pixels used per amount of information presented. --Slashme (talk) 06:52, 13 January 2008 (UTC)

Ok, I also made this choice because it's the most efficient. It stay to find a way to be nice and efficient in the same time : delete space between square ? use dot ? etc. Yug 15:09, 13 January 2008 (UTC)

As for the soft colours, nice idea to keep a standard SVG for this purpose. --Slashme (talk) 06:52, 13 January 2008 (UTC)

Ok. Yug 15:09, 13 January 2008 (UTC)

How about giving the option of making it an HTML table (wikitable?) or an SVG at the user's discretion? It could in principle look exactly the same, and has some technical advantages when you have a grid, and might be easier to fix for people without graphic design experience or software when a by-election changes one seat. --Slashme (talk) 06:52, 13 January 2008 (UTC)

I'll look into methods of automatically generating the SVG of the table. I'm sure it can't be too hard, in fact it might be possible to do by template! --Slashme (talk) 06:52, 13 January 2008 (UTC)

Interesting idea, we are here to test new ways... so go ahead if you think that can produce a gain. Yug 15:09, 13 January 2008 (UTC)

I suggest it be Wikipedia:Graphics Lab/Standards and/or Wikipedia:Graphics Lab/Standards/New, as that seems to be our main intent. As to the colors, I recommend that we just follow the party color templates already existing. If someone would be willing to create those pages, I will write up a formal standard based on what's been going on here, and you people (the ones who will actually be making these graphs) can correct and discuss it (Or rewrite it if I screw it up), etc. 68.39.174.238 (talk) 13:04, 13 January 2008 (UTC)

Research and development page to create ?: I think that's is not a good idea to create now 3 pages (/Research and development ; /Standards ; /Standard new) when the first recommendation set up in the English Lab is not even finish. But... Action is better than words, so if you are able to manage them 3 [successfully] : go head, create them, edit them, etc. Yug 13:36, 13 January 2008 (UTC) —Preceding unsigned comment added by 210.203.61.15 (talk)

I reject (To use your language) "Research and development". I intend "Standards" to be for the established standards, and "/New" for discussion (EG. If they existed, this discussion would be under /New, and when finalized would be summarized and placed under /Standards. 68.39.174.238 (talk) 16:21, 13 January 2008 (UTC)