Wikipedia talk:Manual of Style/Accessibility/Alternative text for images

Longdesc attributes

As explained in Creating Accessible Images, long descriptions, images containing too many important informations to be described in one or two sentences - such as graphs - need a link to a long description. MediaWiki does not currently support the longdesc attribute. What should we do, and how? Do you have any idea? Dodoïste (talk) 04:04, 23 August 2009 (UTC)

The simplest suggestion I can think of is to put the long description in the file page itself. For example, File:Oktadrachmon Ptolemaios II Arsinoe II.jpg #Summary contains a "Description" entry, which could contain the long description. Eubulides (talk) 06:08, 23 August 2009 (UTC)

From what I understand, there should be a sign to the screen reader that there is a longdesc somewhere. Plus, the longdesc may be hard to locate in the description entry of the file page.

Mediawiki's way to handle images and thumbnails is a pain, we should work on that, and prepare a few suggestions to the developers. Dodoïste (talk) 22:08, 31 August 2009 (UTC)

Alt text duplication with captions

A couple of recent edits had some problems.

One edit changed some examples and added commentary like "The caption should be duplicated in the alt text, since the interesting content is not in the image itself but in the caption." This edit appears to be based on a misunderstanding of how screen readers work: they read both the caption and the alt text in cases like these; they do not skip the caption. This topic is further covered in #Why not duplicate captions in alt text? above.

The other edit went to the other extreme, replacing alt text with extremely short phrases like "Child with autism" and again saying that alt text should duplicate the caption, albeit much more briefly. Again, this seems to be based on a misunderstanding of how screen readers work. Alt text is not there simply to prevent a screen reader from saying a file name: it's there to give visually impaired readers a brief sense of the useful info conveyed only in the image. Certainly "Child with autism" is not good alt text for an image whose caption already says that the image is that of a child with autism, as this alt text conveys no additional useful info. The change log for this edit cited this reference but the examples in that reference are not appropriate here, as they're famous images of George Washington, where an ordinary reader can be expected to know what Washington looks like and where "George Washington" in alt text is appropriate.

These errors have already apparently caused confusion in a template developer; see Template talk:Jct/shield #Shields and alt text. For now I've reverted them. Please comment here if there's any further interest. Eubulides (talk) 07:09, 23 August 2009 (UTC)

"The alt text should be concise and should emphasize the image's most important visual aspects: it should summarize the gist of the image rather than describing every detail." Well, that's at least what the first paragraph of the page says. But I don't consider any of the given examples to be concise and in terms of the second half of the quoted sentence I think they do more of the latter than the former. – TMF 07:16, 23 August 2009 (UTC)

"Concise" is a relative term, of course, but you make a fair point. I reworded the examples to make them more concise. If you have a specific suggestion for further improvement, please let us know. Eubulides (talk) 07:48, 23 August 2009 (UTC)

This was indeed a misunderstanding, but on how thumbnails work in Mediawiki, not on how a screen reader works.

I totally agree that "Child with autism" is not a great alt text. My point was: in this case, all the relevant information is already in the caption. So the alt text should normally be empty. But we can't have an empty alt text, because the image is the only content in a link. So, we should provide an alt text as short as possible. And "Child with autism" was the first idea that popped into my head. Dodoïste (talk) 13:54, 23 August 2009 (UTC)

You seem to be confused on something. The alt text should only contain information useful to understand the article. Decorative images, such as the picture of Pavel Bure, does not need to be described. That's the reason why I pointed to this reference.

"The examples in that reference are not appropriate here, as they're famous images of George Washington, where an ordinary reader can be expected to know what Washington looks like and where "George Washington" in alt text is appropriate." You are totally wrong here. We should not describe what the person looks like, whether he is famous or not. See Example 4 for further explanations. Dodoïste (talk) 14:10, 23 August 2009 (UTC)

I'm with Dodoïste on this one. And, may I add, in case we would describe any person, which we shouldn't, I find the rationale for not describing Washington as inconsistent, Americanocentic and inane. Dahn (talk) 14:26, 23 August 2009 (UTC)

• "all the relevant information is already in the caption" No, it's not. Not at all. The image conveys important extra information that's not in the caption, information about how the therapist and child interact via shared attention and reinforcement. This is an important point in the context of the Autism article, and it's obvious to someone who can see the image, but not obvious to someone who can only read the caption. (This, by the way, underscores the need to revamp WP:ALT to specify the contexts for the examples.)
• "Decorative images, such as the picture of Pavel Bure" The image of Pavel Bure might be decorative in some contexts, but it is not decorative in the context of being the lead image in the Pavel Bure article. In that context, the image provides valuable information on what Bure looks like. That's the point of the image: to depict Bure's visual appearance. Lead-image portraits in biographies require and deserve alt text describing what the depicted person looks like, because that information is otherwise unavailable to the visually impaired reader. (Again, this underscores the need to improve WP:ALT's examples to specify contexts.)
• "We should not describe what the person looks like, whether he is famous or not. See Example 4 for further explanations." Example 4 in the WebAIM article is about some other site linking to a Wikipedia article. In a case like that, it is appropriate for the alt text to warn the reader that the image will link them to another web site, and for this to be the focus of the alt text, as it's a somewhat unusual thing. The situation within Wikipedia itself is much different. In Wikipedia, images normally link to a page that further describe the image. This is the normal and expected use of images in Wikipedia, and readers (including visually impaired readers) expect it. In this environment, it does not make sense to waste valuable alt text in relatively useless placeholders such as "Link to file description page for George Washington". Instead, by a Wikipedia tradition so long that it dates back nearly to Wikipedia's birth, the alt text conveys the image's useful info that is not already in the caption or nearby text. This tradition is a logical consequence of how screenreaders work and how visually impaired readers use Wikipedia's "thumb" images. Other web sites work differently and have more freedom in how they use their alt text; also, even within Wikipedia one can use extended image syntax to achieve special effects using non-"thumb" images, which allow much more freedom about what should go into alt text. But for an ordinary portait of a person, the important info that it conveys is what the person looks like, and so that info should be in the alt text (unless it's already in the caption or nearby text).
• "I find the rationale for not describing Washington as inconsistent, Americanocentic and inane". There's no inconsistency here. The same argument would apply for any globally recognized figure, such as Gandhi or Mao or Napoleon. Furthermore, the WebAIM page is written for an American audience and so the argument about Washington applies even more forcefully to it.

Eubulides (talk) 18:28, 23 August 2009 (UTC)

• No, that's what you repeatedly claim is relevant. If the caption is the "what, where, when", that's the "how", and wikipedia don't do "how", for people who can or cannot see.
• Yes it is decorative, even there. The rest is frankly like saying that all of us who can see will "get" the same info on Bure by watching his picture. Also, if it's supposed to help people who see in any way, it's because they can then recognize Bure if they see other images of him. Let's not kid ourselves: blind people can't possibly "benefit" from that sort of info by definition. Or else what's next? We describe sound samples for the deaf?
• Same as above.
• Oh yeah? Well who is the authority to tell us where this "circle of familiarity" is supposed to end? Do Indian people using wikipedia know what GW looked like? Do most Americans know what Ceauşescu looked like, as most Romanians do? It's one more inconsistency that would be addressed simply by not going into such details at all. After all, it matters not that Ceauşescu was curly to anything he did. While some say that it matters that he was short (which isn't apparent from most pictures), this is on par with the cliché about GW's wooden teeth - if it's important, you'll find it mentioned in the article if it's of FA quality. But I'm sure that these alternatives mean little to someone who is set on claiming that describing how Ceauşescu looks is vital to the article. Dahn (talk) 19:27, 23 August 2009 (UTC)

Again, I am having trouble following the connections between the previous comment and earlier ones. But to try to address the points that I can follow: Certainly Ceauşescu has nowhere near the recognition of Washington, Gandhi, Mao, or Napoleon in the English-speaking world, and I would expect a typical reader of English Wikipedia to not know what Ceauşescu looks like. And I strongly disagree with the claim that "blind people can't possibly 'benefit'" from alt text. Clearly they can benefit: we have evidence from both a blind Wikipedian who strongly supports this sort of alt text, and from reliable scholarly sources. The claim that blind people can't benefit from alt text is purely unsupported opinion. Eubulides (talk) 01:51, 24 August 2009 (UTC)

Now you're just twisting my words. Here is what I have said: "Also, if it's supposed to help people who see in any way, it's because they can then recognize Bure if they see other images of him. Let's not kid ourselves: blind people can't possibly 'benefit' from that sort of info by definition." Here is what you produced from that phrase: "blind people can't possibly 'benefit'" from alt text". I'm afraid your obstinate claim that describing the facial appearance of guy in a 2D image to someone, using subjective terms, is "useful" to anybody is by now a question of logic (nb, "useful", not "nice", not "sweet" etc.; I'm saying this because you seem to get the two confused, and because wikipedia doesn't work with "nice" and "sweet"). Maybe you'll find the connection between my comments when I tell you this: go ahead and alt texts diagrams, facsimiles, and anything where the picture conveys relevant information that the blind reader may actually miss, and that is typically not covered in the legend; but alt texting something that that adds nothing to the article (and making it a requirement to do so for others to do the same) is a ridiculous experiment in how not to go about doing things.

And you keep repeating that old slogan about endorsement from users and how sources agree with you, when, let's face it, they agree with alt text in general, not to what you claim it should be used for. Your unwillingness to acknowledge that would really not be worth this much time, were you not the person who basically drafted this thing, useful and nonsensical parts alike. Without that claim that absurd things such as a "serious face" on a hockey player really help anyone, and without the repeated argument that one at all need to describe how a person looks in a portrait, this would be a shorter and much more applicable policy that I would endorse myself.

As for the "personalities": again, who is the authority? Let me rephrase that: who can possibly be the authority? Dahn (talk) 03:52, 24 August 2009 (UTC)

The best authorities we have are the Web Content Accessibility Guidelines, scholarly sources such as Buzzi & Leproni 2008 (doi:10.1145/1368044.1368049), and feedback from blind Wikipedia editors and readers. On the last front, after I made this edit adding alt text to Chelsea Bridge, a blind Wikipedia editor responded "Wow! Good work. I checked out the Chelsea Bridge article; it was interesting to hear descriptions of things that sighted people would take for granted". And yet the alt text that I added was not just about diagrams or text: most of it was the sort of text you're calling "absurd": text like "Suspension bridge crossing a wide river. A few buildings can be seen behind the bridge." Feedback from real users like this weighs a lot more heavily than abstract theorizing over why alt text is a "ridiculous experiment". Eubulides (talk) 06:14, 24 August 2009 (UTC)

That's being "nice", not being "useful", and the implications are still ridiculous (as per the descriptions offered on this page). Now let's have the same with descriptions of sound samples. But again: when I'll have an FAC, I expect you to fill in those for me, since you seem to know what is and isn't useful way better than me; I won't do it myself unless there's a diagram around. Dahn (talk) 09:15, 24 August 2009 (UTC)

Oh and, incidentally: a proper article on a structure, any structure, should include somewhere in the article text a description of the thing. In Chelsea Bridge, both the legends and various quotes seem to fulfill and even overdo that task already, even if granted it's better to have most such details in one paragraph. To push an unsourced and unsourcable additional description in the alt text for an image is almost the equivalent of a content fork. Dahn (talk) 09:44, 24 August 2009 (UTC)

Please stop right there. You're not being constructive anymore. Eubulides and I are going to improve the guidelines for alt text, as soon as we "get some time free from these debates (:-)". So please, let us work for a while, and we'll continue the debate afterwards. ;-) Yours, Dodoïste (talk) 11:58, 24 August 2009 (UTC)

I can only encourage that. If it was a problem of my posting, I would have refrained from doing so earlier, but I noticed that Eubulides is both adamant that the guidelines could only use cosmetic changes, and would rather discuss the flaws than remove them. sorry I took so much space here and so much of your time, and I'm glad there's at least one more editor who sees the problems this project has and has not given up on persuading Eubulides that they exist. Regards, Dahn (talk) 12:11, 24 August 2009 (UTC)

• "The image conveys important information about how the therapist and child interact via shared attention and reinforcement." In that case you are right, of course. But then we should clarify two issues:
  1. "A woman smiles broadly" has nothing in common with "A therapist and child interact via shared attention and positive reinforcement". The first is unuseful alt text and the second is an important information.
  2. When we provide guidelines about what kind of alt text to provide in a specific context, we should not forget to mention the context. In this case, we should mention that this alt text and picture are in the context of the Autism article.
• "The image of Pavel Bure might be decorative in some contexts, but it is not decorative in the context of being the lead image in the Pavel Bure article." Exactly the same as the previous issue.
• "Example 4 in the WebAIM article is about some other site linking to a Wikipedia article." You have mistaken example 3 with example 4. I am talking about the painting of George Washington crossing the Delaware River.
• In conclusion, we already agree on many things. I'm looking forward to continue the discussion. :-) Dodoïste (talk) 19:44, 23 August 2009 (UTC)

• "we should not forget to mention the context" Absolutely. When I get some time free from these debates (:-) that's first on my list for improving WP:ALT.
• '"A woman smiles broadly" has nothing in common with "A therapist and child interact via shared attention and positive reinforcement". The first is unuseful alt text and the second is an important information.' This comment is based on the assumption that "A woman smiles" is not useful. But that assumption is incorrect. The fact that the woman is smiling conveys important information about autism therapies. Many non-experts think of behavioral therapy for autism as being an unpleasant business involving electric shocks or whatever. This image helps to show that modern behavioral therapy does not have to be that way. Here's another way to put it: like the image itself, the alt text conveys information that is useful in more than one way. It's not just about shared attention: it's about the overall gestalt of what the therapy looks like. And the alt text here is the only thing that is conveying that gestalt to visually impaired readers.
• "You have mistaken example 3 with example 4." Ah, sorry, I looked at the 4th image in the WebAIM article rather than at the image labeled "Example 4". But the Example 4 image is not relevant to this discussion, since it has no link. From the WP:ALT point of view, it's OK if the Example 4 image has no alt text. (It should have alt text anyway, but discussing this would be a digression as Wikipedia images typically do not work like the Example 4 image.) Anyway, the point remains that the longstanding Wikipedia tradition for thumb images is that the alt text describes visual aspects of an image that are not mentioned in caption or adjacent text, and we should not be attempting to overturn this longstanding tradition, which is used by the visually impaired, without some very good reasons.

Eubulides (talk) 01:51, 24 August 2009 (UTC)

Major rewrite

In response to #Practical matters and #Alt text duplication with captions above, as well as the long discussions of alt text in Wikipedia talk:Featured article candidates/archive39 and WT:ACCESSIBILITY #Unicode characters and screen readers, I have rewritten much of WP:ALT. This is a bit of a work in progress but I figured it was time to get something out the door. Changes include:

• Alt text is no longer required for math mode formulas, and the suggestion in many cases is to not supply it. For these formulas, the default alt text generated by the Mediawiki software is often good enough.
• Examples have been scattered throughout the guidelines rather than being segregated into a single section.
• Each example supplies a context.
• Many more examples have been added.

Comments are welcome. Eubulides (talk) 09:09, 27 August 2009 (UTC)

It's great, really. Well done! :-) That's a huge improvement. Dodoïste (talk) 09:31, 27 August 2009 (UTC)

Galleries

I filled a bug about <gallery> some time ago, asking to allow to add an |alt= parameter for every single image in the gallery. You might want to look at it: bugzilla:18682. Yours, Dodoïste (talk) 10:07, 27 August 2009 (UTC)

Thanks for bringing up galleries; that was a gaping hole in the project page. I added a subsection that discusses them and mentions that bug report. Eubulides (talk) 19:02, 27 August 2009 (UTC)

Template:Infobox rail line

Could you please fix {{Infobox rail line}} so that it is alt compatible at Washington Metro and other places.--TonyTheTiger (t/c/bio/WP:CHICAGO/WP:LOTM) 17:38, 27 August 2009 (UTC)

I fixed the template so that it should work for that example, anyway. The template's documentation talks about WP:Route diagram template but I didn't see Washington Metro using that stuff so I left it alone. Eubulides (talk) 18:08, 27 August 2009 (UTC)

Also, please fix {{Infobox Skyscraper}} so that it accepts the double image alt text at Trump International Hotel and Tower (Chicago). It seems that the documentation may not clarify the use of two images properly.--TonyTheTiger (t/c/bio/WP:CHICAGO/WP:LOTM) 21:08, 27 August 2009 (UTC)

You can simply add the alt text to each parameter. Parameter values contain extended image syntax, so there's no need to modify the template. Please see WP:ALT #Writing the text's discussion of {{Infobox}}. Eubulides (talk) 22:54, 27 August 2009 (UTC)

I did not notice that the image were not really a part of the infobox code.--TonyTheTiger (t/c/bio/WP:CHICAGO/WP:LOTM) 02:43, 28 August 2009 (UTC)

Alt text for video clips

When doing alt text for video do we just describe the first still image shown or try to describe the entire video clip? -- œ^™ 04:44, 31 August 2009 (UTC)

The latter. Please see the video clip in International Space Station #Assembly and structure for an example. I'll add a section to WP:ALT about this; thanks for the query. Eubulides (talk) 05:23, 31 August 2009 (UTC)

Sigh. Must I? :) The video clip I'm having trouble with is Ribbentrop#Trial and execution. I don't know how to describe this without it taking up a lot of text, and since it's a newsreel there's already accompanying audio commentary, so I was thinking I could just do something like "See commentary" in this situation, as mentioned in the Wikipedia:Alternative text for images#Placeholders section.

One cannot assume that the reader can hear the video. The reader could be hearing-impaired. As it happens, I am currently using a browser with no audio capability whatsoever; my display device doesn't even have a speaker. I can see the video, though, and it does communicate useful info that's not in the adjacent text. For starters, it has a prominent text title "21 NAZI CHIEFS GUILTY" that should be in the alt text as per WP:ALT#Text. I suggest a brief summary of the video's contents (say, 40 to 60 words, including the title; 100 words would definitely be too much). Also, it would be helpful to put into the file description page a text transcript of the video; I realize this will be some work and goes well beyond what WP:ALT requires, but you did ask, and in my current state of not being able to hear the video I can attest that it would be quite useful. I added a paragraph to WP:ALT#Videos about this. Eubulides (talk) 20:03, 31 August 2009 (UTC)

So?

So how are you ladies and gentlemen making this "guideline" more applicable? The only thing I see so far is more self-assured absurdities about furniture being "elegant" and people gazing "commandingly", and now the utterly absurd command to textualize audio clips. I therefore have to assume that the technique applied here is "ignore all objections and keep at it as if nothing happened". Which means that, at this point in time, applying the "guideline" is not only unintelligent, it's also demeaning. Dahn (talk) 21:23, 31 August 2009 (UTC)

Specific suggestions for improving the guideline are welcome. The only thing I gleaned from the previous comment is a criticism of the alt text suggested for Image:Jacques-Louis David 017.jpg in WP:ALT#Verifiability; if this can be turned into a specific suggestion for improving the wording example there, that would be helpful. Eubulides (talk) 22:12, 31 August 2009 (UTC)

This answer is precisely an illustration of the "ignore all objections..." system I outlined above. I do believe I have repeatedly told you that even attempting this kind of description veers into the trivial and subjective. You want a specific suggestion beyond that? "A portrait of Napoleon" or, at most, "A portrait of Napoleon standing" should do. The rest is testing the limits of what wikipedia is all about, and creating a playground for users who want to say how images make them feel. Dahn (talk) 23:18, 31 August 2009 (UTC)

Thanks for the specific suggestion. I changed the example wording in the brief version to "Full length portrait of Napoleon standing", which is pretty close to what you requested. In the longer version that edit also removed the "commandingly" and "elegant" that you objected to earlier in this thread. Eubulides (talk) 23:34, 31 August 2009 (UTC)

Am I going to have to do this with and for every picture, or do we now both see the slippery slope that comes with letting users apply this indiscriminately? Btw, version 2 is still outside reasonable limits, however you define those. Dahn (talk) 00:15, 1 September 2009 (UTC)

There is no slippery slope here, unless by "slippery slope" one means the sort of freedom that always exists under editorial judgment. The new alt text is somewhat terse but it is not unreasonable. Eubulides (talk) 00:53, 1 September 2009 (UTC)

Nope. All other editorial judgments, which are not as loose as you would have us think, fold back on sources, not on the eye of the beholder. We've been through this. Dahn (talk) 02:19, 1 September 2009 (UTC)

Like I told you before, this kind of comment is destructive.

There are still issues to be fixed, but they can't for now because Mediawiki is bad at producing accessible images. So we still have the hard part of the work left, which is to improve Mediawiki itself. If we do it right, the caption could actually be enough for most of the thumbnail images, and the decorative thumbnail images could finally have an empty alt. But of course, in order to do that we need to remove the link on the images, so we have to think about another way to link to the description file. I've been thinking about that for a while now, but I haven't got the right idea yet. Dodoïste (talk) 22:29, 31 August 2009 (UTC)

"Destructive"? Anyway: since these are supposed to be the exception, and not the rule (unless this is something which just surfaces in conversation from time to time), surely you could replace this whole pointless and unfathomable "guideline" with a non-binding small project/taskforce, where you could simply work on each and all example of where alt text "should" be applied, and just stop thinking about it as a default option for regular images. This provided the "essential" info users are missing is not that Napoleon has a commanding gaze or whatnot (people, we've been through this). Dahn (talk) 23:18, 31 August 2009 (UTC)

Yes, we have been through this, and the task force suggestion is a reasonable one. I'm not much of a task-force person; I've never been a member of one and don't particularly want to start now. However, any editors who which to form one are certainly welcome to. Wikipedia could definitely use the help. Eubulides (talk) 23:34, 31 August 2009 (UTC)

No, you're missing the main part of the point. This, the page we're discussing, is a guideline. This means that, defiant of objections, it stands as a requirement for FAs (and is was already questioned there and here, over and over again). That in turn means that, if one has an FA coming, someone can block the FA nomination by simply invoking this admittedly shoddy guideline, until it fits a specification that can be stretched to mean anything. This happens while the very people who came up with this guideline have no clear idea of what and how they would want applied.

Now, instead of this constrictive and self-contradictory guideline, which is effectively imposed on all of us for the possible benefit of a few (at best, on principle it may help some in some cases), and which seemingly relies on some esoteric knowledge of what is necessary and where (namely, your esoteric knowledge), I have proposed to turn this vision into a project. A non-binding project, where you and like-minded individuals can start sorting out the minority cases where this is actually needed (as opposed to proclaiming that it's needed as a rule, which it patently isn't), and begin doing it yourselves instead of spending eternity defining inapplicable generic criteria that some 98% of wikipedia could actually do without. The other 2% is actually manageable by a task force. (And I'm gonna refrain from commenting on the irony of you disliking taskforce work but creating a new "guideline" midway through the project.) Dahn (talk) 00:15, 1 September 2009 (UTC)

Objections were raised when alt text was made a requirement for FAs, as they were raised for pretty much every FA requirement, but a consensus process was followed and they were eventually added by consensus. There is no need to replace the WP:ALT guideline with the task force; it would be helpful to have multiple ways of supporting the same goal of WP:ACCESSIBILITY. The comment "on principle it may help some in some cases" seems to ignore the evidence presented in Wikipedia talk:Featured article candidates/archive39 #Alt text helps the visually impaired. I suggest avoiding wording like "your esoteric knowledge"; talk-page comments should be about content, not the contributor. I am becoming inclined to agree with Dodoïste that this thread is not helping the encyclopedia. Eubulides (talk) 00:53, 1 September 2009 (UTC)

Consensus can change, and I do believe that what you call consensus only reflects part of the debate.

"There is no need to replace the WP:ALT guideline with the task force; it would be helpful to have multiple ways of supporting the same goal of WP:ACCESSIBILITY." Do you really not see the need? And, at long last, would you stop misconstruing my words to read like I'm suggesting ways to proliferate this "guideline"? As for the tedious link to the discussion page, it proves just that "on principle" - as stated, the one example you could find of a person citing this as helpful is for an article where the appearance of the bridge should have been mentioned, in plain terms, within the article itself.

I am not commenting on contributor, I am stating facts: this page looks like it does because you think it should look like this, things are claimed to be helpful when you claim they are helpful, and subtleties are decided upon by you. Don't worry: it could be just about any user, attempting the same futile task in whatever way, with similar results - because there is something innately flawed in how this guideline was conceived and is defended, and it will be flawed as long as humans will be humans. In plain terms, this is pushing a square figure through a round hole, and the only rationale yet for why we should even attempt that is your esoteric knowledge. Your esoteric knowledge about why it's important to say that a woman is smiling, your esoteric knowledge about the importance of hockey players having "serious expressions", your esoteric knowledge about the significance a "Brutus haircut" must have for a blind man, etc. I cannot but personalize this, as long as it's you who added the absurd descriptions, you who kept them, you who defend them, and you who will change some but not all. Yes, it could be just about anyone else performing the same experiment, but it so happens that it's you.

And I'd thank you for not ignoring this point this time around as well: transforming this into an FA guideline has effectively made it everyone's business to follow up on your arbitrary and non-transparent decisions of what is and isn't necessary, of what is good and what is better, but you won't even create a system in which the burden falls on those who want to apply this. No, it falls on everybody, even those who see its manifest problems (and, no, this is not the case with any other guideline).

"I am becoming inclined to agree with Dodoïste that this thread is not helping the encyclopedia." As opposed to shutting all windows and doors and going on with this experiment, I suppose. Dahn (talk) 02:19, 1 September 2009 (UTC)

Implications

well I haven't been in Wikipedia in recent times but this is really, really new to me. I always knew that there was a vauge guideline to add alt text to images but making it mandatory was off the charts for me. My syphomies go to the visually impaired but as mentioned before it seems like another chore in many ways. A few questions:

1. Is there going to be a push to try and get all pages alt-ized?
2. When exactly did this become?
3. Most importantly, at least for me, can a FA or FL be failed because it contains no alt text? Is this now a cross-wiki guidline for all pages?

In my opinion this usability initiative thing going on is pushing the envelope a little o_o. A Wikipedia fork and new guidelines is a bit much to take in at once. ResMar 18:22, 4 September 2009 (UTC)

Oh my. I don't like the direction this is going. You want to change Wikimedia so that it works better with this? And admittedly no one will, sorry to say, give a crap about this guideline, because unless you're doing an FA or FL, it doesn't matter as Wikipedia guidelines are mostly non-binding.

...But its a guideline...a binding guideline...ResMar 18:31, 4 September 2009 (UTC)

BTW. For this to work you're going to have to change 90% of the templates on Wikipedia. And can we make it so that the alt only displays when the image fails instead of when you put your cursor over it? Cause that's just plain silly. ResMar 18:42, 4 September 2009 (UTC)

• Support for alt text was added to the MediaWiki software last October, the WP:ALT guideline was updated a couple of months later, and it has been required for FLs since June 23 and for FAs since July 1.
• In theory an FA or FL can be failed purely because of missing alt text. As far as I know this has never happened for FAs. (I don't follow FLs much.) Generally speaking, editors have been willing to add alt text and the issue hasn't come up. For example, if you take a look at the five featured-article nominations just before yours, three had alt text before being nominated, and two editors added alt text soon after being asked (I presume they also didn't know about the newer requirement). It's not a big deal in practice.
• As far as I know there is no systematic effort to add alt text to all 6,866,709 articles in English Wikipedia. It'd be quite a large job to do it all at once. However, alt text is being added to many articles; the effort is certainly not limited to featured articles and lists. And please bear in mind that featured articles are supposed to represent the best of what Wikipedia can do, including the best accessibility for visually-impaired readers.
• Every template used by recently featured articles has already been updated to support alt text (assuming it didn't support alt text already). It's not that hard to update templates in this way; please see WP:ALT#Writing the text for more.
• Unless you're running IE7 or earlier (which has a bug that is fixed in IE8 or later), the tooltip displayed when you hover the mouse over the image is the title attribute, not the alt attribute. So your request to improve the behavior will be granted once you upgrade your browser. For more on this, please see WP:ALT#Appearance.

Eubulides (talk) 18:50, 4 September 2009 (UTC)

So the point here is to ignore the historical and anecdotal reference and focus on describing the image? (for example, "Napolean standing at his desk" no, but "An aged man in his thirdies with blah blah style haircut standing at a desk (I know that's a bad example but I don't feel like writing a good one). With this type of model i simply do not see how one can be pressed to elaborate as well. In addition this seems to be geared almost comepletely towards screen readers; how useful is "man-in-his-thirdies" when your image doesn't load? Or if your images are off? If this is to work it should focus on all three of its goals instead of the one that is of the greatest minority. I do not find this model to be insanely useful. While its obvious that my argument was lost I must question how useful "a three story building" or "a middle aged man" really is. ResMar 23:46, 4 September 2009 (UTC)

• "So the point here is to ignore the historical and anecdotal reference" No, useful historical or anecdotal info should not be ignored. The point is that useful info like that should be put into the caption or other adjacent text, so that sighted readers can see it too. Another way to put it is that alt text should not be the only place containing a useful piece of information.
• "focus on describing the image" Yes, alt text should focus on the visual impression given by the image; please see WP:ALT#Essence.
• "I know that's a bad example but I don't feel like writing a good one). With this type of model i simply do not see how one can be pressed to elaborate as well." The best examples are real images in real articles. In your 00:29, 5 September 2009 comment below, you give a real example (thanks); I'll follow up there, to help illustrate how this is done.
• "In addition this seems to be geared almost comepletely towards screen readers" Please see #Ah, but this isn't all about the visually impaired is it? below.

Eubulides (talk) 05:21, 5 September 2009 (UTC)

Ah, but this isn't all about the visually impaired is it?

The only question is whether editors should bother to aid the visually impaired. Eubulides (talk) 17:43, 5 August 2009 (UTC)

Graham87 was saying that alt text is useful in describing a logo, for example. But in many of the examples we've seen of alt text, if not most, the descriptions are often quite unhelpful. The example I gave above was describing Immanuel Kant as a man in a wig looking to his right, or Clare College Cambridge as a three-storey building. It's hard to see how that would help someone, and yet this is what you're arguing should be essential for FA. The issue is that it's yet another stylistic burden on FA writers, when people are arguably already overloaded, and in an article with a lot of images that are tricky to describe well, it adds quite a bit of extra work. -SlimVirgin

Ah, but this isn't all about the visually impaired is it? The greater majority of those who this is useful to are those whos images don't load or who turn off or have no images on their browsers. I have to agree that it's another stylistic load on FAers, and that it should be relagated to a less important role, like "you are required to add appropriate alt texts to your images" instead of "you will add alt texts to your images and we will oppose if it is not satisfactory enough."

It's quite vauge what you want to get out of this, exactly. And it's obvious that you're playing by Graham's assesment to make this is a blind-man thing when the guideline itself states in its goal that It should also make sense in a graphical browser with images turned off, and it should fit with the surrounding text when viewed with a text-only browser. I cannot fathom this focus on the visually impaired. (by the way this paragraph is a seperate argument, not against the guidelin-ability of this but against the stess put on the visually impaired). ResMar 00:07, 5 September 2009 (UTC)

3D rendering of the Seamount.

And since you stress examples so much and your defense argument will most likely rest on how I present no evidence, here's some evidence for you. Now on my FAC you tell me that several of my alts are copied from the captions. True enough. Although I was never very good at debating I see a fatal flaw in your arguments. You told me that I best change the caption to the following: A north-south ridge, trending slightly east of south. Pele's Pit, at its peak, is about 1000 meters below sea level; further south the ridge gradually descends about 3500 meters to the sea floor. My problem with that is that this advice goes against the guidelines you yourself wrote. For the flag of Mali the page recommends that you alt it Vertical tricolor flag (green, gold, red). So in your first guideline you write that I must elaborate on the image further with information that is unique, not copied from the article or the caption, but in the guideline itself you require one to describe the image itself not give any information about it? Bull. you follow a conflicted line of argument that changes with the image and there are no, and should be no, direct guidelines on what's what. I remain to understand wether you want us the describe the image or to give information about it. And skwirm your way out of the vise as you may try, it is an impossible and impractical line of thought to follow. Decide upon one or the other. You seem to adapt from one to the other when the situation seems fit. So tell me:which one is which? What do we follow? ResMar 00:29, 5 September 2009 (UTC)

Ah I never stated what I meant with the image. The caption is a simple word combination that gives a basic description of the image. There is no anecdotal or historical signifigance to discuss otherwise, and it does not draw any geographical conclusions. Thus one is hard-pressed to come up with an alt that is both unique from the article (I do not like witholding information from the text body, we do it enough with captions already, which are far more important then alts) yet states the exact same basic thing without being simalar to the caption, and I am hard pressed to push past making a "different enough" version. In addition I am to question wether you want to "link" the two, as you said on my FAC that it must "elaborate" upon the image, WHICH THE CATPION ALREADY DOES. ResMar 00:35, 5 September 2009 (UTC)

...However, its changes to the other alt text entries did not address the repetition problem. For example, for Image:LoihiBathemetric.jpg a screen reader will read the alt text and caption aloud, resulting in something like "Link Bathymetric mapping of Lōʻihi, with an arrow pointing to Pele's Pit in the center. Link Bathymetric mapping of Lōʻihi; the arrow points to Pele's Pit." (alt text italicized); the repetition obviously is not helping the visually impaired. (Please see WP:ALT#Repetition for more details about this.)...Eubulides (talk) 22:32, 4 September 2009 (UTC)

You're right. It's not useful at all. Here your solution was to give geological points of the image. Strike two for the anti-guideline. Now I get what you want. you propose two different systems for two different types of images. For those that already do the alt's "supposed job" you want us to add "unique" information, whilst fot those like a potrait caption, only mentioning the potrait's name and year of creation, you want us to describe in extensive detail the potrait. Please tell me how that is NOT contrary. In addition I am to think that perhaps just not leaving an alt with my first example is better since the caption already does the job. ResMar 00:44, 5 September 2009 (UTC)

• "Ah, but this isn't all about the visually impaired is it?" Correct. Although visually impaired readers benefit the most, other readers also benefit. For example, as you mention, people on low-bandwidth connections can look at the alt text to decide whether to click and retrieve the image. These other usages for alt text benefit from the same alt text that screen readers benefit from.
• "I cannot fathom this focus on the visually impaired." The impression I got from earlier discussions on this topic, as well as from reliable sources on alt text, is that support for visually impaired readers was the most important reason for alt text.
• "'you are required to add appropriate alt texts to your images' instead of 'you will add alt texts to your images and we will oppose if it is not satisfactory enough'" The featured article criteria merely require "brief and useful alt text when feasible", and this is all that is being asked for. Please don't assume the worst about the process. I don't know of anybody who has ever opposed a featured article nomination purely on alt text grounds. The requirement is not that much work in practice.
• "It's quite vauge what you want to get out of this, exactly." The stated objective for alt text is in WP:ALT#Goal. What is vague about that section, exactly? Can you suggest specific wording that would make it less vague?
• "this advice goes against the guidelines you yourself wrote... So in your first guideline you write that I must elaborate on the image further with information that is unique" No, WP:ALT never says that the information must be unique. On the contrary, WP:ALT#Brevity explicitly says "As it is not meant to be a unique identifier for an image, alt text typically does not supply enough information to characterize an image unambiguously." However, I see now that WP:ALT's first example, which used the flag of Mali, was a bit too simple, and gave the reader the misimpression that alt text should uniquely characterize the image. I have tried to address the problem by substituting a more real-world example, a photo of a Danish flag flying, taken from Flag's lead. I hope this helps to clear up the confusion that is evident in the "two different systems for two different types of images" comments above. There is just one system, not two.
• Much of the commentary above is about Image:LoihiBathemetric.jpg, and yet the image shown above is File:Loihi 3d.gif, a different image. It's not clear to me why the mismatch is present; perhaps you intended to show the former image instead of the latter?

Eubulides (talk) 05:21, 5 September 2009 (UTC)

• Flag: better I guess; see below.
• FA reviewers are forceful to death that's why. <(o.0>) I see now its actually CLEAR what you want. Null point.
• Yes but you're more likely to see a low-bandwidth or text browser user so I STILL don't like that it focuses on the visually impaired, with these two as a secondary thing. They should at the least be equal.
• Ah, well actually it doesn't matter because this image is follows the same point as that image, but I just read See caption so ok this is a null point.

Better. ResMar 12:34, 5 September 2009 (UTC)

Question

Should every image of an article have an alternative text? Felipe Menegaz 16:22, 9 September 2009 (UTC)

Purely decorative images should not have alt text (or a link); other images should all have alt text. Please see WP:ALT#Purely decorative images for details. Eubulides (talk) 19:17, 9 September 2009 (UTC)

Identifying articles with images without alt & getting them into cleanup listings

I suspect there are hundreds of thousands of articles with millions of images lacking ALT text. Would it be possible to get a bot (or similar mechanism) to label them in some way (not visible to readers - similar to the way User:Erik9bot tags article lacking references) and them puts them into a category, or using Template:Alt text missing to put them in Category:Unclassified articles missing image alternate text. If the article is also tagged by a wikiproject this should enable them to be included in project cleanup lists and may encourage editors to add alt text to images in areas in which they have an interest/expertise.— Rod ^talk 10:43, 12 September 2009 (UTC)

I'm sure a bot like that could be written. However, I'd guess the vast majority of articles would be in the new category: would such a category be useful? Eubulides (talk) 13:51, 12 September 2009 (UTC)

Since I suggested it, User:Erik9 has submitted a bot request for approval to do this at Wikipedia:Bots/Requests for approval/Erik9bot 12 for discussion. I think it would be useful as the wkiproject banners would enable the mega category to be parceled up to relevant wikiprojects - it would also give a better idea of the scope of the task.— Rod ^talk 18:46, 12 September 2009 (UTC)

Garbo descriptions in ALT text

I was not under the impression that when writing ALT text, we should be inserting POV wording such as "breathtakingly beautiful" or "beautiful". These descriptions should be as neutral and descriptive as possible without interjecting our POV interpretation of someone's beauty. Garbo may be breathtakingly beautiful, but it is not for an encyclopedia to determine. I also have a bit of a problem with one person rewriting the description to match his or her own POV wording in policies/guidelines. This tends to endorse the POV wording officially. That's simply a no-no. Wildhartlivie (talk) 00:36, 15 September 2009 (UTC)