Wikipedia talk:Manual of Style/Accessibility/Archive 8

Archive 5

Archive 6

Archive 7

Archive 8

Archive 9

Archive 10

→

Archive 15

Equilibrium Revisited

Hi everyone!
The equilibrium discussion has sat long enough that I think consensus has been decided. So I downloaded AWB and currently have it running, finding all instances of <math>\rightleftharpoons</math> and replace it with {{Eqm}}, which contains the same bit of TeX with alt text added. Unfortunately, I've run into a bit of an obstacle- that string almost never appears outside a larger <math></math> string that contains the rest of the equation, i.e. <math>A + B \rightleftharpoons C + D</math>. It seems to me then, that instead of bulk replacing all the equilibrium symbols, it would be better to simply use alt text for the entire equation. The question that I have, is in doing this would it be preferable to use chemical names or chemical equations? For example, if my equation was $CH_{3}CO_{2}H+H_{2}O\rightleftharpoons CH_{3}CO_{2}^{-}+H_{3}O^{+}$, for the alt text should I write "Acetic acid plus water is in equilibrium with..." or "CH₃CO₂H plus H₂O is in equilibrium with..."? ~ L'Aquatique! [^talk/_stats] 21:28, 3 November 2008 (UTC)

It'd be better to write out the chemical symbols. That's what sighted users get in all chemical equations, so that's what blind people should hear. Graham87 01:07, 5 November 2008 (UTC)

Great! Thanks. ~ L'Aquatique! [^talk/_stats] 08:54, 6 November 2008 (UTC)

Yes, chemical equations should not be set in TeX. There are occasions where editors have set them in TeX, but these should be changed for plain text: it looks better, it's easier to edit and it's more accessible for screen readers etc. Physchim62 (talk) 08:58, 6 November 2008 (UTC)

What do blind people hear? "Cee aych three cee ow two aych plus aych two ow ...", "Carbon-trihydrogen-Carbon-dioxide-hydrogen added di-hydrogen-mono-oxide ...", or "Acetic acid and water is in equilibrium with Acetate and hydrogen ion"? The software I've listened to can generally handles math equations half decently and will try to render chemistry equations as math equations, so it'll sound like the first one. I personally like it written out with chemical names but worry about those using braille displays. — Dispenser 06:36, 7 November 2008 (UTC)

The vast majority of blind people will hear the first example. Use of abbreviations is the best way to write chemistry equations for the blind, because they don't take long to read with a speech synthesizer. Graham87 07:10, 7 November 2008 (UTC)

Wow, Dispenser, it took me a disturbingly long amount of time to figure out what you had written in the first example... Kinda looks like Welsh or something. *sigh* Ah, for a phonetic alphabet! ~ L'Aquatique! [^talk/_stats] 08:42, 7 November 2008 (UTC)

Okay, so I have another question. As it happens, TeX does not allow use of superscript and subscript in alt text. However, it seems like you sort of need them for differentiation purposes. For example, ${\mbox{rate}}=k_{-}{S}^{\sigma }{T}^{\tau }\,\!$ does not mean the same thing as ${\mbox{rate}}=k{-}{S}^{\sigma }{T}^{\tau }\,\!$. Is there another way we can designate superscript and subscript 1.) without making an unholy mess and 2.) in a way that people who haven't been privy to this conversation will readily recognize? ~ L'Aquatique! [^talk/_stats] 09:07, 7 November 2008 (UTC)

←

Hmm, tricky one, as k₊ and k_– are pronounced "k-plus" and "k-minus"! So you two sample equations would read "k-minus S-to-the-sigma T-to-the-tau" and "k minus S-to-the-sigma…"! Physchim62 (talk) 10:13, 7 November 2008 (UTC)

Ahh! Mine head hurts. Those two would presumably be read the same by a screen reader. Indeed, it took me longer than I care to admit to notice the difference.

Writing "to the" in place of superscript seems somewhat reasonable, although how would you designate the end of the superscript? For example, by that formula a^b+c would read the same as a^b+c, (both would be "a to the b plus c") even though they're quite obviously not the same ~ L'Aquatique! [^talk/_stats] 10:27, 7 November 2008 (UTC)

There's a reason mathematicians prefer equations to talk! The only way of disambiguating that in talk is to add some parentheses to the equation, or to use inflection and rhythm to try to convey the order of association. --Itub (talk) 10:42, 7 November 2008 (UTC)

This is an interesting discussion! How do you plan on changing the ALT text of math-mode equations — perhaps by a script?

Anyway, here are some ideas that might help. You could use "sub" in the ALT text to indicate subscripts, as in "k sub m" to indicate k_m; that's used among scientists, although it's less common and slightly longer. Secondly, you could state multiplication and other operators explicitly; for example, k₋ [A] [B] could be read as "k-sub-minus times A times B" or better yet, "k-sub-minus times the concentration of A times the concentration of B". For well-defined special cases such as chemical equations, you could write a template that produces the correctly formatted text and ALT text automatically. Such a template would allow you to change the formatting consistently across the wiki as technology progressed or fashions changed. Braces or parentheses do seem like the simplest way to distinguish a^b+c and a^b+c, the former being "a to the open paren b plus c closed paren" and the latter "a to the b plus c".

Two more ideas, just off-the-cuff. A clever script could produce your "acetic acid plus water" translation from the chemical-symbol ALT text, which might be helpful for many students. However, to allow for more sophisticated chemical equations, you might consider stashing the SMILES coding (or something equivalent) as an attribute in the image tag. Please let me know if I can be helpful in the coding, Proteins (talk) 13:42, 7 November 2008 (UTC)

I would say that it is best to try to keep as close to the way these equations are actually read, rather than creating a new translation. Unfortunately, that would mean a new HTML tag along the lines of <vcomma /> (for "virtual comma") to tell voice readers when to insert a short pause, just as natural speakers would do when reading out loud a potentially ambiguous equation. Physchim62 (talk) 16:07, 7 November 2008 (UTC)

Proteins: the <math> tag, which codes equations using TeX, has an alt= attribute.

Physchim: writing them to be read the way they would be read in actuality makes sense, but then we get back to the fact that a^b+c and a^b+c are read the same. Parethensis are an interesting idea (a^(b+c) -> a to the (b plus c) ) are they actually red in screan readers or does the speech synthesizer merely alter the intonation, like you would when reading a parenthetical statement in real life? ~ L'Aquatique! [^talk/_stats] 19:25, 7 November 2008 (UTC)

I made a sandbox in which to try a few equations, and tested them with Fire Vox and with a demo version of JAWS. I should say right away that I'm a newbie to both programs, and don't know the best settings; Graham could give you much better information — perhaps subscripts and superscripts are a solved problem for screen readers?

Here was my experience at least. Fire Vox skips over parentheses altogether, but JAWS reads opening and closing parentheses as "left parenthesis" and "right parenthesis". However, JAWS read k_m ("k sub m") and g_m ("g sub m") as "kilometer" and "gram", respectively; it seems that JAWS is a little too smart for its own good. Fire Vox did make a different pause between a^b+c and a^b+c, so there's no need for a virtual comma there; however, JAWS did not, at least for me. For example, JAWS made no distinction between "a b + c + d" (juxtaposed a and b plus c plus d) and "a^b+c + d" (a to the left paren b plus c right paren + d). FWIW, Proteins (talk) 22:15, 7 November 2008 (UTC)

Where did you get that demo version of Jaws? Fiddling with settings is what I do best. ~ L'Aquatique! [^talk/_stats] 06:19, 8 November 2008 (UTC)

Hi L'Aquatique!, Graham pointed me to the demo version. I downloaded version 9, but version 10 has just been released. If it's like version 9, it operates under "40-minute mode", meaning that you have 40 minutes to try it out. However, whenever you reboot your computer, you get another 40 minutes; I believe you can do that indefinitely. Some of its options seem to be disabled in the demo, but even so, it has lots of settings to fiddle with. Have fun and please share with us what you learn! :) Proteins (talk) 13:35, 8 November 2008 (UTC)

Thanks. I have some hearing problems, but we'll see what I can do. Too bad it probably doesn't come with subtitles! L'Aquatique[talk] 19:50, 8 November 2008 (UTC)

It is funny how the makers of JAWS didn't think of subtitles, but I suppose they had their hands full in getting the rest of the program to work well; it seems a gargantuan task to me.

It's nice to meet a fellow scientist, L'Aquatique! With Graham's help, I've been trying to write some scripts to help visually impaired Wikipedians, but if you had anything you wanted to improve for better accessibility, you should be bold; we might be able to find a good solution. My dad is also hearing-impaired, and I just finished my first course in American Sign Language, which I loved. For a while, I was getting addicted to the ASL videos on YouTube.

I've been thinking about your problem of making chemical reactions accessible. I drafted a script to do that and some basic calculations, although please be forewarned that it's still rudimentary. You can try the script out by adding "importScript('User:Proteins/chemicalreactions.js');" to your monobook.js page. A small tab labelled "rxn" will appear at the top of the page, in the line with the "history" and "move" tabs; clicking on the "rxn" will invoke the script. For now, the script will work only on this sandbox. I'm a little embarrassed about mentioning such a basic script, but I think it could become much better with input from other people. By the way, everyone here should feel free to as critical as they want. It's OK if you think the script is not very useful; I'm just tinkering and would appreciate the guidance.

In its present form, the script reads all the chemical reactions on a page, and describes them one by one, checking that mass is conserved and the reaction is balanced. You'll see right away at the sandbox that the script isn't savvy about nuclear reactions, since it thinks that "H + n → D" is unbalanced. The last two equations in the sandbox are deliberately unbalanced, for testing purposes. I was thinking of adding buttons to each reaction so that a reader could check each one separately for balancing and mass/charge conservation: does that sound like a better approach?

By looking at the sandbox contents, you'll see that much of the information is encoded in SPAN elements. These could be generated automatically from a well-designed set of templates. Such templates would also provide a common interface for chemical and nuclear reactions across Wikipedia.

A general property that I like about this script is that the content (in this case, the chemical reaction) can become more interactive and instructive at the request of the reader. The article has "hidden depths" that are unfolded by the script(s). A similar approach could be extended to other types of Wikipedia content, such as mathematical equations, peptide sequences, taxonomies, etc. You can imagine adding buttons to give explanations or to do almost any sort of calculation or cross-check or even to plot things. Proteins (talk) 04:47, 9 November 2008 (UTC)

Haha, well, I'm not the cool kind of scientist, unfortunately. I can't do math to save my life, so I had to go into Biology and leave all the bigger questions to others. I have a fairly firm grasp of Chemistry, but it's not as good as I'd like.

I've spent a few forty minute sessions with JAWS now, and I have to say, it's quite confounding! Subtitles aren't really necessary since everything that it reads appears visually as well, but it seems like it gets easily confused. For example, I tried to get it to read me a short story I wrote... everything was going fine until it suddenly decided to skip a paragraph, and then things really started getting whacky and it was randomly leaving out words, skipping around... Not a pretty picture. I tried it on Wikipedia as well with a marginal amount more of success- I do like how it pronounces my username "el akwateek" instead of "lakwateek"- gotta give it some kudos for trying! Anyway, I downloaded the basic training manual and I'm slowly reading through it. I think if I can learn at least some of the shortcuts things will go much smoother- my attempting to use the mouse probably confused it. I do have to say, though, if I can get it to work it will be the realization of a childhood dream- I remember late nights wanting more than anything to stay up and read one more chapter of a book, but I just couldn't keep my eyelids open! L'Aquatique[talk] 05:10, 9 November 2008 (UTC)

Coolness is relative and the grass is always greener; you'd be surprised how much "bio-envy" that quantum physicists have. Quite a few of my Ivy-League colleagues have confided in me something like, "Oh, I am so sick of flogging Nature to eke out one more tidbit of information. I wish I could go into bio-sciences and do something that would have an impact! Plus, that's where the real money is." Watch out, a superstring theorist might join you at the fishery. ;)

I've also had trouble learning how to use JAWS, but I have to admit, I haven't been practicing as much as I should. It's probably like a fine violin that screetches until you master it. Everyone else seems to use it, and if we care about improving accessibility, then we must go to the mountain; it's unreasonable to expect that the mountain will come to us. Shortcuts have worked for me, although I do find it a little scary how JAWS takes over your computer and won't let you type things into text boxes if it's in a bad mood; have you had that happen? I also wish JAWS or Fire Vox could make MP3's or audio files, so that I could listen to them without the screen reader running. Then you could listen to your favorite articles as you fall asleep — I know the feeling.

If you get a chance, please let me know what you think of the chemistry script, or the other scripts listed on my user page. They're definitely works in progress, and could use some help. Proteins (talk) 07:14, 9 November 2008 (UTC)

Hi Proteins, JAWS has a thing called forms mode, which lets you type text into a textbox. It's activated by pressing enter when you're on the bit that says "edit". When forms mode is off, the letter keys and numbers 1-6 are used for quick navigation keys. JAWS 10 tries to activate forms mode automatically when it thinks you want it activated. I haven't played with JAWS 10 much, so I don't know how well that works; upgrading is (a) expensive and (b) is a *huge* deal for me because I have to redo most of my JAWS customisations. BTW I didn't know that JAWS 10 was released until I read this discussion ... so I had to update the JAWS (screen reader) article which was a whole week out of date. As for playing Wikipedia articles on MP3 players, Pediaphon can convert Wikipedia articles into MP3 files; I wish I'd remembered about the program earlier. There is also Wikipedia:WikiProject Spoken Wikipedia; one of my favourite audio files produced by that project is the spoken version of acetylseryltyrosylseryliso...serine, which might, or might not, be a good antidote to insomnia for you. ;) Graham87 11:59, 9 November 2008 (UTC)

That's helpful, Graham, thank you! It was markable how quickly Freedom Scientific moved from beta-testing to release of JAWS 10. I haven't tried the JAWS 10 demo; do you think there's been enough change/improvement that I should upgrade?

If we wanted a script to produce lullabies such as acetylseryltyrosylseryliso...serine from arbitrary protein sequences, I could do that in an eye-blink. That protein sequence is only 158 residues long, which is relatively small for a protein. Some proteins are thirty times longer and may be decorated with many additional chemical groups, such as sugars. Different proteins might be better lullabies for different people; they might say, "Really? I prefer to fall asleep to the potassium transporter protein." Some peptides are cyclic, too, so the lullaby could repeat endlessly... Proteins (talk) 14:36, 9 November 2008 (UTC)

There aren't many new features in JAWS 10 that are important for testing JAWS on websites, besides the auto-forms mode thing. The beta cycle is a normal length for JAWS: the first JAWS 10 beta was released in late August, and they've been updating it since then. IIRC the last time the beta testing process took longer was in JAWS 5.0, but that version was basically a complete rewrite of JAWS. Graham87 15:03, 9 November 2008 (UTC)

JAWS reads superscripts and subscripts much more reliably when you move by character (with left and right arrows). However it raises the pitch of the voice and makes them sound funny. To make them read reliably, one would use the speech and sounds manager. I'm sure someone, somewhere, has made a scheme for JAWS which is optimised for math and science reading, but I don't know about it. I couldn't get JAWS to say "kilograms" instead of kg, for example, but I'm using a slightly earlier version than the ones y'all are testing. That shouldn't make much of a difference, though.

You can get a rough idea of what JAWS is seeing by using insert-control-w to virtualise the window, which will display the text that JAWS thinks is in a window in the virtual viewer. JAWS gets easily confused if you move the mouse pointer; NonVisual Desktop Access, a free Windows screen reader, is better at doing that. You can find out what JAWS output through a refreshable Braille display looks like by using the Braille viewer which is in the start menu with all the other JAWS options.

As for the script, I don't know much about chemistry but to me the most useful part would be knowing if an equation is balanced. I did chemistry up to year 10 (or the 10th grade as they call it in the U.S.), and it was just equations, equations, and more equations. Graham87 07:08, 9 November 2008 (UTC)

Thanks, Graham, I thought you would appreciate the script, although I was worried that it was too long-winded. It'd be good to figure out how to break up its functions and arrange them on a page so that people can get what they want as quickly as possible. The present script checks whether the reaction is balanced, and reports on the counts of different types of atoms in the reactants and products. I was hoping that might help students and other readers understand what "balanced equation" meant.

It occurred to me that 10th-graders might be able to do some of their chemistry homework by typing their equations into a sandbox and running the script. I've been toying with the idea of writing a routine to balance the equation for them; it wouldn't be hard to write. It would be like a gift to my younger self, since I used to have difficulties balancing reactions in school. Proteins (talk) 07:26, 9 November 2008 (UTC)

[ec] Haha, well it doesn't change much past that point... still mostly equations, with a few random laws thrown in for good measure. I haven't taken a university level chem class yet but looking at the textbook I'm expecting more of the same. I liked Chem back when it was the periodic table, "how many protons does this atom have" etc. At a more advanced level, it's mostly math and as I mentioned earlier, math goes in one ear and out the other for me.

Proteins: I would love to take a look at your script, however, my coding experience begins and ends at <font color=blue>. How I ever got hired to do web design remains a complete mystery. I will, however, install it and see how it works. L'Aquatique[talk] 07:30, 9 November 2008 (UTC)

Thank you, L'Aquatique! In case you or Graham are interested, I just dashed off a script for assessing !voting at FAC discussions, based on the script I made earlier for Graham to summarize AfD !voting. I can already tell that the script isn't working perfectly, partly because FAC !votes are more free-wheeling in their structure. But it works OK and if you all had any suggestions for improving the script, I'd appreciate it. Going to bed, read my last chapter for today, Proteins (talk) 07:39, 9 November 2008 (UTC)

To be honest I think the AFD analysing script would be more useful. FAC is much more likely to have long threaded discussions, which don't make as much sense when simplified to a !vote count. The idea of FAC discussions is that they should be closed if there are no actionable objections, so if someone writes "Object, this article contains pictures" in a FAC for the article photography, the objection would rightly be ignored. Graham87 11:59, 9 November 2008 (UTC)

Superscript and Subscript

Ripping off Proteins' good idea, I set up my own sandbox to play with superscript and subscript (here: User:L'Aquatique/AccessSandbox). I found, at least with my current settings, that JAWS doesn't seem to care much about superscript and subscript, it read abc (a times b times c) and a^bc (a to the b, plus c) the same, attempting to pronounce it as a word: "ahbk" or something like that. The idea that I'm toying with, is taking a leaf from the html tags that code for subscript and superscript and writing "sup" "end sup", sort of like how JAWS reads parentetical statements "left paren" "right paren". For example, a^bc (a to the b, plus c) would be written a sup b endsup c. The question is, would other people recognize that for what it is, or would they just think, "dude, what the heck does sup end sup mean?". L'Aquatique[talk] 01:19, 10 November 2008 (UTC)

Yes, it'd be a good idea, but it'd take some getting used to. But as I said above, JAWS indicates subscripts and superscripts when you arrow through the document with left and right arrow, and the speech and sounds manager can be used to indicate them. Graham87 02:15, 10 November 2008 (UTC)

Right, and that works when you're writing in plain text, but when writing with the <math> attribute, which is [apparently] necessary to write more complex stuff, like sigma notation and differentials, throwing in a <sub> in the alt text actually breaks the code. For example, the math string ${A}^{1}+{B}^{2}\,\!$ is A¹ + B², but when you try to put that as the alt text, this happens: Failed to parse (syntax error): {\displaystyle 1</sup> + B²">{A}^1 + {B}^2 \,\!} . Using "sup" instead of <sup> seems like the closest we're going to get, unless there's something we haven't thought of yet...? L'Aquatique[talk] 02:39, 10 November 2008 (UTC)

It's great that people care so much about solving this problem. The "sup"-"endsup" seems a good approach, especially since expressions will always be unambiguous, no matter how many levels of super- and sub-scripts, which likely isn't true for other schemes. However, the sup-endsup approach might have a few drawbacks worth considering. (1) There's already a mathematical operation known as "sup", which stands for "supremum". It's relatively rare in most settings, but still there might be some ambiguity. (2) My first impulse is to keep simple superscripts such as a^b brief, and save the sup-endsup approach for more complicated situations. On the other hand, "sup" and "endsup" together are only 3 syllables. (3) I think people would wise up quickly to the meanings of "sup" and "endsup", although they're less familiar. You might be able to do the same thing, though, using parentheses, although they're slower.

Some other thoughts. The meaning of a sub- or superscript varies with context; in different equations, a superscript might be an index, a power, a contravariant component of a relativistic tensor, an atomic position within its molecule like the C^α atom in proteins, the electronic or spin state, etc. These different subscripts might be read differently. The "to the" phrasing works great for powers, but perhaps not so well for the other contexts; we wouldn't say "C to the alpha", would we? Ideally, the ALT text would clarify the context. You could do that by asking authors to ALT-label their equations themselves. Or you could make some templates to handle mathematical functions; I might try my hand at that tomorrow, if I can steal a little time. Proteins (talk) 06:31, 10 November 2008 (UTC)

Well, it doesn't have to be all or nothing. For example, if the equation was x = a^b (X equals A to the B), it wouldn't be necessary to write the whole thing: we could write A sup B, or maybe even just A^B, that's recognizable, right? I think whether or not the situation called for sup endsup would be something we would have to decide individually on a case by case basis, but it seems to me that we're not intending to use it as a fill in for "to the"- I guess I'm saying I don't see why we couldn't write C sup alpha, even though we don't say C to the alpha.

I also think we would see a significant amount less of sub than sup, because subscripts are often not read separately, i.e. in chemical equations there would be no real need to write H sub 2 endsub O, H 2 O would suffice just fine because of the nature of the layout of chemical equations. L'Aquatique[talk] 07:50, 10 November 2008 (UTC)

That seems reasonable, at least to me. If I understood your suggestion, we should omit the "endsub"/"endsup" for simple sub- and superscripts, and in some cases, even the initial "sub" and "sup" for the most common cases. For example, I'd be tempted to write "C-alpha" instead of "C sup alpha" for C^α since it's so common and since that's how it's spoken in practice by protein scientists. I think we can overcome concern (1) by reading "sup" out as "supremum", and I think the parentheses idea could backfire in some complicated cases.

I'm unclear on what happens next, though? After we agree on an accessibility approach, do we then try to convince others to adopt it? Proteins (talk) 16:56, 10 November 2008 (UTC)

Yes, you read my suggestion correctly. I think the function / endfunction markup could be used in other circumstances as well, for example say we wanted to write 2x + 2x³ + (4x + 2/3), we might say 2x plus 2x sup 3 + fraction numerator 4x plus 2 denominator 3 endfraction. It could also be helpful for things like rads, logs, etc.

As far as, if we get consensus here, then we add it to this page: WP:ACCESS, which is part of the manual of style. Then, maybe leave a note at the major affected WikiProjects so they know what's going on. Then, we roll up our sleeves and start putting it into place. This isn't the sort of thing that's going to get people riled up, since it isn't changing the look or substance of the page at all- the vast majority of the people will never know the difference, and those that do will be happy for the change [hopefully]. L'Aquatique[talk] 18:21, 10 November 2008 (UTC)

FAC-summarizing script

This might not be the right venue for discussing this script, since it's not closely related to accessibility, but we're here and it's still early days, so I'll just continue the conversation, if you don't mind. Maybe we'll hit upon something good.

I agree that the AfD-summarizing script is better crafted for its purpose than the prototype of its FAC counterpart. For one thing, I had your guidance in the former. But after studying FAC with my students for two years now, I sense the need for something to help the people involved and those who might like to become involved in FAC discussions. If the FAC is short and clear in its reasoning, such as Wikipedia:Featured_article_candidates/Rings_of_Neptune, it's easy to keep track of everything. But what about this one and others of that type? It might be useful to have a tool for organizing FAC's with 634 kB of data, and for organizing/summarizing the reasoning on both sides of such a debate.

My impression is that you'd like to keep track of the actionable, reasoned Oppose !votes; supports, neutrals, comments, me-too votes, etc. seem secondary. Perhaps a future FAC script could identify such opposes and reprint their threaded discussions in full, whereas it could just count the Neutrals, Comments and Supports and credit the corresponding users, as the script does at present. With such a script, even a newcomer could see quickly what remains to be worked on in the candidate article. To help such a script do its work, perhaps the FAC reviewers could agree to state the FA criterion on which they're objecting, and how their Oppose is actionable. To an outsider, the FAC process sometimes seems to generate more heat than light, and it'd be good to help with that. Proteins (talk) 15:11, 9 November 2008 (UTC)

Yes, that sounds good. Maybe the script could figure out when an objection is resolved, and state something like "objection about copyediting resolved". If the objection was resolved, there's not much need to read the discussion about it. I haven't dipped my toes into FAC for years, so you probably know more about the conventions on things like that than I do. Graham87 01:09, 10 November 2008 (UTC)

Rather strange to see this here and not at FAC; I really don't know how a script can decide what is actionable or not, when an oppose is resolved, and whether a Support is a drive-by, new editor or fan support. Unlike AfD, FAC is not a vote: I'm not in favor of making it appear to be one. SandyGeorgia (Talk) 18:51, 19 November 2008 (UTC)

We're discussing it here only because Graham has been helping me with writing scripts for accessibility, while I've been thinking of FA issues. His request for an AfD-summarizing script led me to think of an FAC counterpart. I haven't brought it to WT:FAC because it's not ready; I don't want to trouble Wikipedians with half-baked ideas. if you're interested, I wrote a script yesterday that checks for MOS:HEAD violations, as I mentioned at WT:MOS. Given your experiences at FAC, you might be amused by this nightmarish test file.

I've no wish to make FAC seem to be a vote, I completely agree with you. But I would like to give nominators and reviewers tools to help manage larger FACs. I understand that FAC and AfD are not votes, and that the wiki-way is to reach consensus through rational discussion. Would you agree that the number and quality of valid objections are more important than the number of objectors? A vote considers only the latter, but the focus at FAC seems to be on resolving the former.

I'm interested in your thoughts on the following question. Does every valid objection at FAC correspond to an FA criterion? Conversely, if an objection to an article is not related to a FA criterion, is it valid to raise it at FAC? I'm willing to pose these questions at WT:FAC, but I'd like to know your opinion before I solicit others'. Proteins (talk) 23:18, 19 November 2008 (UTC)

The discussion so far shows such an oversimplification of all that is involved in the decision making and review process, that it's hard to know where to start, but the endpoint will probably be the same: I don't think it's useful to attempt to summarize FAC with a script, and I believe any attempts to write a summary script will yield deceptive, vote-like impressions that won't be productive for the process. Concerns from some of the discussion above:

1. "If the FAC is short and clear in its reasoning, such as Wikipedia:Featured article candidates/Rings of Neptune, it's easy to keep track of everything." That statement is a deceptive oversimplification of the decision process—one that heads toward the notion that FAC is a vote and neglects all the factors that I look at in evaluating both Supports and Opposes. As just a few examples, on technical topics, I look for topic area input as well as layperson input. I look at WikiProject vs. non-WikiProject supports. I don't want to pass a technical topic that hasn't had both expert and layperson review. That's one example. Then, each WikiProject has different tendencies: some pile on fan support, some respectfully hold back, considering it a COI to support their own nominations. How many votes there are is truly irrelevant: I know what I'm looking for in each case, every article and Project is different, and it's Really Not A Vote. I don't think it will help the nominators, the reviewers, or me to give the impression that FAC decisions can be summarized to a vote or a table of Supports and Opposes.
2. "But what about this one and others of that type? It might be useful to have a tool for organizing FAC's with 634 kB of data, and for organizing/summarizing the reasoning on both sides of such a debate." Having read several thousand FACs, the four Roman Catholic Church FACs are highly atypical for reasons tangential to FAC; designing a script around them wouldn't likely be a productive venture and wouldn't capture essential aspects of the FAC decision. Most of the commentary had to be moved to talk as it frequently strayed off-topic or away from WP:WIAFA; some commentary ended up on article talk and in userspace; some editors didn't return to strike; some opposes weren't actionable: in short, the issues are too varied to capture with an automated tool. And, will a tool parse WikiProject memberlists to weigh fan support versus independent support? Will it detect newly registered users? Will it happen to notice canvassing that no one mentions? Any notion of vote tallying at FAC is deceptive, unlike any other area of Wiki.
3. "My impression is that you'd like to keep track of the actionable, reasoned Oppose !votes; ... " How is a script going to determine this? For example, pls read the last Hillary Clinton FAC; are all of those 1e opposes actionable and valid?
4. " ... supports, neutrals, comments, me-too votes, etc. seem secondary. " An oversimplification, depending on the reviewer, the topic area, the involved WikiProjects, etc. Often a comment is more valuable than a Support or Oppose declaration. A topic expert Support or Oppose has to be weighed differently than a newly registered drive-by comment.
5. "... identify such opposes and reprint their threaded discussions in full, ... " The train wrecks occur when nominators and reviewers don't thread their discussions and start multiple sections to say the same thing: a script won't be able to sort the rare messes, and the typical FACs don't need a script.
6. "With such a script, even a newcomer could see quickly what remains to be worked on in the candidate article." This is a dramatic oversimplification of FAC: Wiki is a volunteer venture on a website anyone can edit: there is often no such thing as "check, done". So, if one driveby reviewer says prose is fine, no one else should check? If one reviewer says images are fine, that's it? Although few image reviewers agree, and there is often lengthy discussion with unclear conclusions? If one reviewer says sources are fine, no one else is going to look at sourcing in greater detail? This is not only an inaccurate view of how FAC works, it will lead to dangerously superficial, check-the-box type reviews. Wiki depends on volunteers, and all reviews/reviewers are not equal. Six (fan support) reviewers can say prose is fine, until an excellent copyeditor comes in and shows the prose deficiencies. One reviewer can say the sources look fine, until someone who knows the area better comes in and digs deeper and finds that the sources don't support the text. How will a script determine something is "done"? And will a summary script, giving the impression the boxes are checked and the reviews are done, discourage deeper review?
7. "If the objection was resolved, there's not much need to read the discussion about it." Ouch. And what about the (surprisingly frequent) Supports that actually mention important deficiencies? How does a script decide when something is resolved when often engaged brains can't decide that ? What about the Opposer who begrudgingly strikes and leaves in anger because they got tired of trying to get a point across: is that resolved?
8. "Would you agree that the number and quality of valid objections are more important than the number of objectors? " I can probably provide a counterexample for any scenario you can imagine: see the last Hillary Clinton FAC.
9. "Does every valid objection at FAC correspond to an FA criterion? Conversely, if an objection to an article is not related to a FA criterion, is it valid to raise it at FAC?" See Wikipedia:Featured article candidates/Space Science Fiction Magazine/archive1.

FAC is simply too different from AfD for automation; with so many areas of Wiki that can benefit from automation, I just don't see this as a productive use of resources, and I see it as heading in a dangerous, vote-like direction, not acknowledging that article review on a volunteer Project is not a tidy, linear process. With many reviewers already complaining about the current trend of specialization at FAC, I don't see this heading FAC in the directions it needs to go, which is more comprehensive, less specialized reviews. We don't just tick off the 1a, 1b, 1c boxes and say "done", "passed". I think that anyone looking at the process from the outside, without having written FAs and reviewed FACs, is likely to see it as generating heat rather than light, because the process is not a checklist like it might be in a coroporate environment or in a journal review; critical review by volunteers, often laypersons, on a website anyone can edit, isn't tidy or linear and shouldn't be a "check-the-box" endeavour. I do not see this script proposal as a step in the right direction. SandyGeorgia (Talk) 05:13, 20 November 2008 (UTC)

I never meant to suggest that FAC was a vote, or that it was a check-the-box affair, or that automation would play any role other than helping the participants and encouraging newbies to participate more. I'm honestly bewildered how you could have formed such a mistaken impression; probably you're continually besieged by people with FAC suggestions, and just lumped us all together. I'm sorry that you got roped into this very preliminary discussion between Graham and myself.

I will look at the examples you cite and think more about FAC. My current understanding of the FAC process is that whether fifty people support or object is not relevant in itself; as everyone says, it's not a vote, but a consensus arrived at by rational discussion of the outstanding issues. Although votes don't count, the (actionable, valid) criticisms do; it is they that must be reasoned through and resolved. For example, if a reviewer supports but offer 12 criticisms, those criticisms should be resolved if they suggest a failing of a WP:WIAFA criterion, don't you agree? Conversely, if someone objects because WP:IDONTLIKEIT, that seems to offer no criticisms that need to be resolved.

Please don't worry yet about the technical details, as in "how would a script do X?" You might be surprised at what's possible. However, at the moment and for the foreseeable future, I'm just gathering information on what might be useful. Proteins (talk) 01:59, 21 November 2008 (UTC)

Infobox templates with data entered by column, rather than by row

See the template documentation for example Template:Football player infobox#Examples.

It instructs users to enter data like this:

| years               = 1992–1998<br />1999–2000<br />2001<br />2002–
| clubs               = Template United F.C.<br />Template Rangers<br />→ Example F.C. (loan)<br />Template City F.C.
| caps(goals)         = {{0}}75 (26)<br />{{0}}32 (11)<br />{{0}}15 {{0}}(7)<br />113 (46)
| nationalyears       = 1991<br />1996–1998<br />1999–
| nationalteam        = Examplia U16<br />Templatia U21<br />Templatia
| nationalcaps(goals) = {{0}}{{0}}1 {{0}}(0)<br />{{0}}{{0}}8 {{0}}(2)<br />{{0}}46 (13)

To get a table which looks like this:

Senior clubs1
Years	Club	App (Gls)*
1992–1998 1999–2000 2001 2002–	Template United F.C. Template Rangers → Example F.C. (loan) Template City F.C.	075 (26) 032 (11) 015 0(7) 113 (46)
National team2
1991 1996–1998 1999–	Examplia U16 Templatia U21 Templatia	001 0(0) 008 0(2) 046 (13)

Surely this is an accessibility problem for the blind children… accessing… Wikipedia's football (soccer) articles? I'm convinced it's bad news enough for editors with slightly uncorrected vision because the slightest deviation in byte count can push this pseudo-table out of alignment. Also it uses various white-space hacks instead of proper CSS alignment. Something should be done about this. — CharlotteWebb 17:41, 16 November 2008 (UTC)

By whitespace hacks I'm referring to Template:0 which is

<span style="visibility:hidden;color:transparent;">0</span>

— CharlotteWebb 16:34, 17 November 2008 (UTC)

Yes, it's a mess and linearises badly (which affects adults, not just children). Infoboxes can have multiple rows, using parameters such as year1, year2 etc.; which is what should be used in this case. I'm not sure how to code that, so you may need to ask for help elsewhere. Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 17:58, 16 November 2008 (UTC)

Well I know how to fix it. I'm just looking for a rough consensus that it's actually a problem. — CharlotteWebb 18:56, 16 November 2008 (UTC)

Graham will know best, but the HTML code you provide suggests that this template could be improved significantly. The two tables appear to be 4x3 and 3x3 tables, respectively, which I believe would be fine for people using screen readers to navigate. However, according to your HTML code, they're both actually one row of three cells; the appearance of four and three rows is created using linebreaks <br />, which is counterintuitive and not straightforward to navigate, I think. Using whitespace to align the table entries also seems impractical. I'm sure you've thought of this, but conversion to a normal table, with normal style conventions for alignment, seems like a good solution? Proteins (talk) 01:22, 17 November 2008 (UTC)

Yes, the bad HTML is a problem for screen readers. See Wikipedia:WikiProject Usability/Infobox accessibility. Graham87 02:22, 17 November 2008 (UTC)

Basically you'd want to do something like this:

| senior_clubs = 
{{senior_club_row|1992–1998|Template United F.C.|75|26}}
{{senior_club_row|1999–2000|Template Rangers|32|11}}
<!-- etc. -->
| national_team =
<!-- etc. -->

With something like this inside the row template (this also gets rid of the completely inane Template:0, the use of which usually indicates a serious design flaw, but for that matter so does the use of <br />):

<tr>
<td>{{{1|{{{years|{{{year|}}}}}}}}}</td>
<td>{{{2|{{{team|}}}}}}</td>
<td style="text-align:right;">{{{3|{{{caps|}}}}}}</td>
<td style="text-align:right;">{{{4|{{{goals|}}}}}}</td>
</tr>

Only it will be a pain in the ass to convert because you cannot select one row to copy and paste, only the columns. Have to use some kind of javascript I guess. — CharlotteWebb 12:29, 17 November 2008 (UTC)

Frankly I'm stunned that anyone would set it up the current way whether they're blind or not. — CharlotteWebb 12:38, 17 November 2008 (UTC)

Hi Charlotte, that seems like an excellent solution. Are you planning to write a bot to fix the instances of this template?

Perhaps you might consider leaving a note on the Talk page for this template? The people there might have other helpful suggestions, or long-standing improvements they've wanted to introduce. If they give you flak, you can always refer them to the consensus here, but I'm guessing that they only did things that way because they didn't know how to do it better. Good luck, Proteins (talk) 14:33, 17 November 2008 (UTC)

There are nearly 40,000 references to this template. I strongly suggest asking at Wikipedia:Bot requests for help in doing the conversion. -- Rick Block (talk) 14:40, 17 November 2008 (UTC)

Frankly I don't doubt they'll tell me to go jump in the lake or something less kind. — CharlotteWebb 16:29, 17 November 2008 (UTC)

I guess a bot could probably do it right 49 times out of 50, but I'm worried about "coincidental" alignment issues. For example, if there is a really long team name in the middle column, it might wrap onto two lines, forcing the other columns to use a double <br /> to keep the following "row" in proper alignment (which is unreliable to begin with because people have different font sizes ). Other ambiguous situations are also possible.

Anyway if the "new way" uses entirely new parameter names, we can use parser functions to let the "old way" continue to work in the meantime so the rush would not be urgent enough to risk further corrupting poorly presented data. — CharlotteWebb 16:26, 17 November 2008 (UTC)

New template created

{{Infobox Football biography 2}} should resolve this. I've tested it on Lutz Pfannenstiel (who probably has the biggest club list around). Can someone check that it works properly and then suggest a roll-out plan? Chris Cunningham (not at work) - talk 09:51, 18 November 2008 (UTC)

Like I said I'm not surprised by the hostile responses. — CharlotteWebb 13:46, 20 November 2008 (UTC)

On the other hand it is only a partial solution as it still uses the ugly goose-egg template {{0}}. Also having an explicit maximum number of rows isn't really a good idea (no matter how high it is). That's why I suggested using one parameter for the whole table, with a row template (containing the repetitive formating code, which needs to stay 100% consistent) that can be used 1–∞ times, rather than having an over-large template with a bunch of rarely used parameters. — CharlotteWebb 13:46, 20 November 2008 (UTC)

I picked a random soccer player to illustrate what I mean. Check out my edit to Argelico Fucks before it gets reverted. — CharlotteWebb 16:19, 20 November 2008 (UTC)

You know, I'm really thinking about splitting caps and goals into separate columns (3rd and 4th) rather than using the parentheses. — CharlotteWebb 19:13, 20 November 2008 (UTC)

The reason for using an explicit number of rows is that using the {{infobox}} subclass for the base infobox is a win on so many levels when it comes to standardisation and maintainability, but this negates the possibility of using a sub-template due to thw way template inheritance works. I picked a pathological limit for all parameters, beyond that reached by any footballer in history thus far AFAIK, so even if it needs to be increased in future it's not going to be too difficult to do so. Chris Cunningham (not at work) - talk 23:57, 20 November 2008 (UTC)

My initial suspicion is that one template expected to meet all possible needs of every type of infobox will create a greater maintainability nightmare in the long term, but I'm willing to see whether this can actually work. Do you mind if I edit your Template:Infobox3cols to allow the input of a parameter occupying all three columns? It's not impossible, in fact there are several ways to do it. — CharlotteWebb 14:24, 21 November 2008 (UTC)

Incidentally why do the headers become progressively wider after row 40? I'm pretty entering data there or below would make rows 1-39 appear incomplete (having only 3 columns) so I hope this is some kind of mistake? — CharlotteWebb 14:24, 21 November 2008 (UTC)

Check out the documentation at template:infobox/doc: entering a data line without the label will occupy all three columns. Can you provide a test case for the header issue? probably just a syntax hiccup in my code. Chris Cunningham (not at work) - talk 16:12, 21 November 2008 (UTC)

I didn't test it, I just looked at the code, which starts out doing this:

Row 1
-->{{#if:{{{header1|}}}|<tr><th colspan="3" style="text-align:center; {{{headerstyle|}}}">

but for reasons I don't understand it changes to this

Row 40
-->{{#if:{{{header40|}}}|<tr><th colspan="4" style="text-align:center; {{{headerstyle|}}}">

and this

Row 50
-->{{#if:{{{header50|}}}|<tr><th colspan="5" style="text-align:center; {{{headerstyle|}}}">

etc. becoming wider further down and reaching 9 columns at rows 90–99. — CharlotteWebb 18:26, 21 November 2008 (UTC)

D'oh! copy-paste error. Should be fixed now, thanks! Chris Cunningham (not at work) - talk 11:20, 23 November 2008 (UTC)