W3C WBS Home

Results of Questionnaire ISSUE-66 - Image Analysis Heuristics - Straw Poll for Objections

The results of this questionnaire are available to anybody.

This questionnaire was open from 2010-05-12 to 2010-05-19.

10 answers have been received.

Jump to results for question:

  1. Objections to the Change Proposal to Strike Paragraph
  2. Objections to the Change Proposal to Keep The Paragraph As Is
  3. Objections to the Change Proposal to Be More Explicit

1. Objections to the Change Proposal to Strike Paragraph

We have a Change Proposal to remove the image heuristics paragraph from the img element section. If you have strong objections to adopting this Change Proposal, please state your objections below.

Keep in mind, you must actually state an objection, not merely cite someone else. If you feel that your objection has already been adequately addressed by someone else, then it is not necessary to repeat it.

Details

Responder Objections to the Change Proposal to Strike Paragraph
Ian Hickson The first paragraph of the rationale suggests that it is impossible to "recover from images with missing" alternative text. This is a straw-man argument, since the text in the specification (or proposed in one of the other change proposals) does not claim to enable user agents to "recover", merely allows user agents to make best-effort attempts to help users try to interpret such images.

The second paragraph continues arguing against this straw-man, citing software that has shown a remarkable ability to identify images, and claiming that it cannot solve all problems. Yet no claim is made that any software can solve _all_ problems. Any improvement in accessibility is better than no improvement, even if the improvement is only helpful in limited case. For example, OCR is a mature technology that can be of huge help to users unable to see images, as there is a large class of images that merely consist of styled text. Certainly, relying on OCR software is no substitute for real alternative text, but no claim is made that it is — quite the contrary in fact — and _any_ improvement is an improvement, even if it is not a complete substitute.

The change proposal continues in this vein throughout.

Removing these paragraphs removes helpful text that might inspire user agent implementers to improve their accessibility. This would be a mistake, and I object strongly to doing so. It is always better to be explicit in stating what user agents are allowed to do than to rely on user agent implementers making educated guesses about what is likely to be the most helpful solution.
Matthew May
Larry Masinter
Laura Carlson
Julian Reschke
Cynthia Shelly
David Singer
Gregory Rosmaita
Martin Kliehm
Krzysztof Maczyński

2. Objections to the Change Proposal to Keep The Paragraph As Is

We have a Change Proposal to keep the HTML5 specification as is. If you have strong objections to adopting this Change Proposal, please state your objections below.

Keep in mind, you must actually state an objection, not merely cite someone else. If you feel that your objection has already been adequately addressed by someone else, then it is not necessary to repeat it.

Details

Responder Objections to the Change Proposal to Keep The Paragraph As Is
Ian Hickson
Matthew May The rationale states that the current text should remain because "OCR has existed for decades". To be blunt, so what? OCR is not a specification in itself; it's an entire field of research, consisting of an unknown number of algorithms and processes, many if not most of them proprietary and patented. If you specify it further, you run the risk of patent challenges. If you leave it as vague as it is now, you will end up with numerous implementations of questionable value to the user. The editor cites OCR as "one simple and unambiguously possible technique" when it is neither.

Anyway, the original reason the issue was raised was the even more nebulous "image analysis heuristics" passage, which remains in part in the current draft. UA developers don't need to be told in the specification that they're allowed to do more for the user. One would hope they've got ideas of their own.

I'm in favor of either fleshing this out in a way that everyone can agree is actually beneficial for accessibility, or striking it. This is a half-step which is worse than either of the other two options.
Larry Masinter see 4 below
Laura Carlson 1. The spec mixes directives to authors and directives to user agents, which will cause confusion.

2. It is not clearly noted that the cited paragraph [1] is not an escape clause for authors to get out of writing text alternatives. Some authors may take the paragraph as saying,

If:

User agents can help users with visual disabilities make sense of an image

then:

I don't have to worry about it.

3. Any User Agent image repair verbiage should be left to or vetted and coordinated with User Agent Accessibility Guidelines Working Group (UAWG). They are the domain experts. Any such verbiage in HTML5 would need to be carefully worded so the two specs are harmonized. Currently the two specs are out of sync.