Apology [was: Re: Org HTML export accessibility (was: org exported pdf files))

[Tim Cross <theophilusx@gmail.com>]

Dear all,

I think I owe everyone an apology. I have allowed frustration from
another area of life colour my response here and as a result, my tone
and assessment was too negative.

While it is correct that we cannot use org mode to generate accessible
PDFs and that does mean in environments where policy mandates accessible
content (which is PDF), we cannot use org mode.

The ability to generate accessible HTML is on the other hand quite
possible. Unlike PDF, the burden for doing this rests primarily with the
author and not the data processing framework. There are probably some
things we can do to improve or encourage accessible authoring, such as
alerting authors to content which looks like it should have an alt
tag. I will give some thought to that and when I get a chance, will see
how well various html based back ends deal with accessibility checking
tools.

For those interested and because it might help with understanding in
this area, I thought I'd outline the actual cause of my frustration. The
following isn't directly related to org-mode, but may be informative for
some. However, it is a little long, so feel free to just delete and move
on if your so inclined.

There is a little irony here as well. I've been using org mode since it
was first released. I even recall email discussions with Carsten when he
was first looking at how to improve outlining. It is org mode which
allowed me to generate really good quality documents and track all the
data and tasks I had to manage in my various job roles. People often
commented that they found it interesting that some of the best looking
documents produced in our area were from the person who is legally
blind! The irony being I cannot easily access the PDF output I created
and I became part of the problem by generating inaccessible documentation!

One very long standing frustration I have had in my career has been to
do with access to training materials. Most training organisations are
extremely reluctant to provide electronic copies of their learning
materials. I have lost count of the number of non-disclosure forms I
have been forced to sign in order to get electronic documents from a
training organisation (even though they are legally required to provide
their materials in an accessible format). Even when I have managed to
sign the necessary paperwork and get the documents, they have often been
in the form of DRM protected PDFs with an expiration date. While those
without any disability can retain the learning materials for future
reference, it is not a luxury afforded to anyone dependent on assistive
technology. Worse yet, most DRM protected formats also require the use
of non-free platforms, such as Windows or MacOS (I did often get some
perverse satisfaction from cracking the DRM protection, which in most
cases, is fairly easy to do). However, there is an ironic component here
as well. Usually, the DRM protected PDFs are actually very accessible
once you jump through all the necessary hoops. They are typically well
tagged and easy to navigate. On the other hand, the non-DRM PDFs are
rarely accessible despite correctly formatted PDFs actually being one of
the most accessible formats available. Often, once forced to provide
electronic copies of their learning materials, training organisations
will provide image PDFs, generated from a scanned version of their
materials. Image PDFs are 100% inaccessible - they are just pictures, so
you cannot even extract the text using tools like pdftotext[1] Even when
not image PDFs, they often lack the necessary tagging etc (though, this
situation has improved in recent years as many tools now default to
accessible output rather than requiring it to be enabled).

Even once you jump through all the necessary hoops, your not out of the
woods yet. My current frustration has been with obtaining the important
bit of paper which says your trained and certified. After completing the
course I looked at what I needed to do to sit the certification
exam. The exam is one which has to be done at a large certification
examination centre and it is done electronically. It is actually run by
a very large US based training organisation, who I will not name.

It runs out that I cannot do the training at this time. I have to give
them  a minimum of 12 months notice to sit the certification exam
because due to my 'special' needs, the whole examination centre has to
be booked out just for me! To make it worse, the assistive technology I
have to use is a program called JAWS, which only runs on windows and
which I am totally unfamiliar with. My suggestion to just have a
sighted person assist me by reading the questions and entering the
answers has been rejected as well as all other suggestions and
appeals. It is highly likely I will just forgo certification. While it
would have been handy, it isn't essential.

I outline all of this not for sympathy but to try and promote
understanding of the challenges faced by many who need access to
accessible content. Accessibility is also an area which isn't well
serviced by the open source community. This is not a criticism, just an
observation. It is also easy to understand why. Most successful open
source projects are about scratching an itch. Org mode was born by one
person (Carsten) scratching an itch which turned out to be an itch many
others also had. With accessibility, the number of people wit the itch
is significantly smaller and there is a lot of itch variation. Those
with the technical skill to scratch it are even smaller. 

regards,

Tim

[1] I have had some people say to me that the situation isn't that bad
with PDFs as you can use tools like pdftotext to extract the text from
the files and then read it with a screen reader. Unfortunately, this
approach does not always provide good output. Often, the structure of
the document is lost and bits become tangled/jumbled (consider what is
the best way to dump text from a table in a linear manner when you have
little meta data about that table. The output also often contains
'artifacts' and odd characters as well as spurious spaces which make it
difficult to understand or process correctly by the text-to-speech
system.