W3C home > Mailing lists > Public > wai-xtech@w3.org > July 2007

Re: reCAPTCHA implementation problems

From: Gregory J. Rosmaita <oedipus@hicom.net>
Date: Fri, 13 Jul 2007 13:10:59 -0400
To: wai-xtech@w3.org
Cc: joshue.oconnor@cfit.ie, Gez Lemon <gez.lemon@gmail.com>, Ben Maurer <bmaurer@andrew.cmu.edu>
Message-Id: <20070713163840.M84781@hicom.net>

aloha, again!

first, sorry for confusing gez with josh -- the only excuse i can 
offer is that you both look alike to me <wink>

i agree that free-form navigation using the virtual cursor, can 
be much more confusing for many more users than is generally 
admitted, and hence falls under the classification of an "undue 
burden"

i have to disagree with david on his assertion that screen-reader
contextualization alone suffices -- the screen reader will only 
provide the dimensions of the table, which would be confusing to 
many in this instance, because they have not loaded the page to 
navigate a table, but to fill out a form in order to move on to 
whatever task they are attempting to accomplish, so knowing that 
one is entering a form with x controls, listed in tab order, IS 
the sort of essential contextualizing that only an author can 
truly provide, at least given the current state of things (one 
can imagine a script that searches for id slash headers 
associations (if present) and applies that information to its 
barebones outline of the table (X columns by Y rows), and a 
similar repair could be attained from scope, or by an 
algorithmic calculation, but that is one of the problems -- 
relying on one's assistive technology to provide repair 
mechanisms should not be necessary, if the proper and complete 
markup is utilized correctly by the page's author.  besides, 
the user, as i indicated before, doesn't absolutely need to 
know that the form is ensconced in a table, but DOES need to 
know that inside the table is a form consisting of...

josh, i didn't hear any vertical bars on the page, and i always 
have all graphic detection mechanisms enabled, even though it 
drives me crazy when i visit a heavily nested layout table with 
its various spacers and other kludges revealed to me by the 
aural rendering of the document, so it may well be that there 
is something (the video intercept?) affecting the aural 
rendering of the page when running JAWS on a virtual windows 
machine in the mac environment?  DavidP, i believe, is the one 
to ask about that peculiar behaviour, as he uses a mac far more 
often than i...

oh, and ChrisB, your posts have been particularly useful -- thank 
you -- that is precisely the type of feedback that ben needs 
from a developer's point-of-view; it could well may be that 
a WCAG2 technique or two can be refined or added due to this 
investigation, and some consistency in the realm of HTML4/XHTML1
forms perhaps achieved...  i greatly appreciate your concrete 
implementation advice, especially since i know that y'all at 
AOL have been working on an accessible captcha-type challange...

actually, everyone's posts so far have been extremely useful -- i 
just wish that those ben sent to the list on wednesday and thursday 
morning had percolated their way down to the list archive -- i was 
fortunately included in the recipient field when ben was first 
attempting to post to the list, but now that that has been taken 
care of, i believe that we're making real progress...  i was unable 
to make the DHTML SubTeam meeting today, at which i was going to 
ask BeckyG to review the thread and add any advice she might have 
as a dojo toolkit implementor, so if she doesn't pick up on this 
thread, i'll email her directly to ask her to do so...

gregory.
-------------------------------------------------------------
ZEAL, n.  A certain nervous disorder afflicting the young and
inexperienced. A passion that goeth before a sprawl.
		    -- Ambrose Beirce, The Devil's Dictionary
-------------------------------------------------------------
Gregory J. Rosmaita: oedipus@hicom.net
     Camera Obscura: http://www.hicom.net/~oedipus/
        Oedipus' Online Complex: http://my.opera.com/oedipus/
-------------------------------------------------------------


---------- Original Message -----------
From: Joshue O Connor <joshue.oconnor@cfit.ie>
To: "Gregory J. Rosmaita" <oedipus@hicom.net>
Cc: Gez Lemon <gez.lemon@gmail.com>, Ben Maurer <bmaurer@andrew.cmu.edu>, 
wai-xtech@w3.org
Sent: Fri, 13 Jul 2007 17:25:52 +0100
Subject: Re: reCAPTCHA implementation problems

> Gregory J. Rosmaita wrote:
> > 1. user doesn't know what to expect when encountering the form
> 
> Exactly, and also if this CAPTCHA is going to be different from 
> the way the usual CAPTCHA  will work (or not) then the user will 
> have to be informed of such. The skillful use of @summary in 
> this case would be really good.
> 
> > 2. user may never discern that the graphical tier of alternate 
> > slash assistance links even exist, eliminating (or limiting) 
> > the likelihood that a user will use JAWS' virtual cursor to 
> > inspect the entire page, [...]
> 
> Relying on the user to be able to navigate their way around 
> using the JAWS cursor is a bridge too far (in my book) as it is 
> usually something that only power users, or at least rather 
> experienced screen reader users will ever do so IMO it is not 
> ideal to rely on it for the user to find - what is essentially 
> to be core functionality - of the accessible CAPTCHA.
> 
> > quick and dirty fix: since the form is contained in a table, have 
> > you given any thought to adding a brief description of the form 
> > (how many fields, the tab order, that an audio alternate is 
> > available, etc.) -- using a summary would greatly assist a 
> > non-visual user in conceptualizing the form [...]
> 
> Agreed.
> 
> The switch to audio button could be a graphic with suitable alt text
> that describes to the user what it is for. This would enable it 
> to have focus when the user tabs to it. Is it because of the way 
> Javascript has been applied in this instance that the two 
> buttons 'reload' and 'switch type' are both somehow out of the 
> tab order? If so would it be possible to use an element that 
> would stay in the tab order and still   be able to call a JS 
> function to trigger the switch to audio mode? I guess so.
> 
> I also noticed a huge amount of 'vertical bars' in the interface,
>  which are really tedious to listen to, did anyone else get 
> that? Or is it indicative of my ongoing driver issues while 
> running JAWS with Parallels on a Mac Book Pro :-(
> 
> Josh
------- End of Original Message -------
Received on Friday, 13 July 2007 17:11:27 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 13:15:43 GMT