Slashdot Mirror
Aussie Government Gives PDF the Thumbs Down
littlekorea writes "The central IT office of the Australian Government has advised its agencies to offer alternatives to Adobe's Portable Document Format to ensure folks with impaired vision are able to consume information on the Web. A Government-funded study found that PDFs can present themselves as image-only files to screen readers, rendering the information contained within them unreadable for the vision impaired."
A thumbs down in the southern hemisphere is the same as a thumbs up in the northern hemisphere, as long as you name the file bruce.pdf. It saves confusion.
Given the number of times government officials around the world have failed to understand the difference between removing text in a PDF and replacing it with black and just covering the text over with black, they'd probably get it wrong about half the time even with best intentions.
Learn how to operate another program? Spend from the budget for another set of licenses? (the horror)... start to use Open Office or the like?
Questions raise, answers kill. Raise questions to stay alive.
ISO already has created the standardized PDF/X subsets used widely in the publishing industry. They lack support for extra features like scripting and other extensions.
The main problem with PDF for document archives is that it is a presentation format and doesn't adequately preserve text structure since everything is broken down into lines of text or individually placed glyphs. Analysis of a page layout can only bring back so much. There are better ways to store data that offer more versatility.
I am becoming gerund, destroyer of verbs.
Look at this page. It's for a local police department in a city that has lots of blind people because of the presence of the California School for the Blind. This is the first page that Google lists for the site. I can't imagine that a screen reader can make anything of the front page and there are no navigation buttons.
The real "Libtards" are the Libertarians!
No it doesn't sound like a bozo official since that style of pdf was specifically excluded from the user study they ran.
You could of course skim the report and know that, but I guess that would mean you couldn't launch into meaningless rants.
Of ocurse if you did that you'd know the report is available in PDF format which I guess would just launch you on a different meaningless rant.
And the rest of us say "Get rid of it". We do not access government documents to be blown away by their totally rad page style. We access them for information, and extracting the information from the glumph that encases it is sometimes hard for the best of us.
html all the way. Any formatting you cannot fit in a simple stylsheet can get left out.
Prediction for end of Universe #42: Fencepost error in Quantum_bogosort.cpp
Not necessarily. PDF does not preserve text flow. It breaks up paragraphs into lines (or less if kerning has been altered), and places them accurately on the page. If you have a multi-column layout, then a pdf-to-text algorithm (first step in screen reading) is likely to put column-2-line-1 between column-1-lines-{1 and 2}. Best of luck sorting that out.
Prediction for end of Universe #42: Fencepost error in Quantum_bogosort.cpp
Also consider pdfs with complex page layouts. Deciphering the text flow from them is often hard for eyeballs, let alone computers.
2 columns is enough to throw out many screen readers.
Prediction for end of Universe #42: Fencepost error in Quantum_bogosort.cpp
You do know that in Australia it is law that a company make their website accessible for vision impaired if at all possible.
...
What does it matter that they can't read the text? PDFs aren't about content, they are about preserving the layout. At least that is what it seems like to me when I am foolish enough to try and read PDFs on a device with a different number of pixels than the person who made the PDF file.
If the content matters at all, someone should invent a technology that allows text to be tagged somehow with indicators of the MEANING of that portion of text, like 'this is a title', and let the display device render the text according to how the reader can best view it. It sounds crazy, and it may take a few decades to do, but think of the benefits.
They whose government reduces their essential liberties for temporary security, receive neither liberty nor security.
I would expect a Microsoft format from our illustrious leaders.
Bingo. Anyone who doesn't see Microsoft's hand in this is hopelessly naive.
The correlation between ignorance of statistics and using "correlation is not causation" as an argument is close to 1.
I expect they could require that all they wanted, and it still wouldn't happen.
If my usability manuals are to be believed, people have neglected the safeties of nuclear reactors because those things are a chore and do nothing anyway. If you don't want your users to do something, then you design your system so that they never get the option.
Not necessarily. PDF does not preserve text flow. It breaks up paragraphs into lines (or less if kerning has been altered), and places them accurately on the page.
This is not true. PDF is capable of preserving text flow if the document contains such information. See this as an example: if you open it in acrobat reader and move the text cursor using the down arrow, you'll see it travel correctly among columns and paragraphs.
No page description format will help if the page has been generated in a broken way: for instance, try extracting text from the tables of an html page generated by javascript.
If you have a multi-column layout, then a pdf-to-text algorithm (first step in screen reading) is likely to put column-2-line-1 between column-1-lines-{1 and 2}. Best of luck sorting that out.
In this case it is the pdf-to-text algorithm to be broken, and should be fixed.
So basically they are saying that *because* it is possible to produce a shoddy PDF file which is basically an image dump, that this is reason enough not to use the format?
By this same reckoning, you could produce a really shoddy HTML page which also consists of images and no text... Virtually any format could be misused in this way.
So what's the alternative? That we all revert back to ASCII text since its incapable of holding graphics?
Personally i hate seeing poorly designed websites or pdf files as i described here, where the text is actually an embedded image (or worse - a flash file) and there is no clickable index etc.
We should probably start naming and shaming pdf creation software, and those who use (or misuse) such tools.
http://spamdecoy.net - free throwaway anonymous email - avoid spam!
This is not true. PDF is capable of preserving text flow if the document contains such information.
Yes, this can be done, but it is almost universally not done. Of all the pdfs out there, almost all of them that have anything but single column text flow incorrectly. The answer is of course to include this information every time, but I don't see how you can mandate that if the standard doesn't include it and most or all current software creates pdfs that don't have it.
If you have a multi-column layout, then a pdf-to-text algorithm (first step in screen reading) is likely to put column-2-line-1 between column-1-lines-{1 and 2}. Best of luck sorting that out.
In this case it is the pdf-to-text algorithm to be broken, and should be fixed.
I'm not sure that you can always figure out the text flow correctly a posteriori. Once the correct text flow information hasn't been encoded in the document, it's a bit of a crap shoot in some cases to figure out what was intended. Where should that floating box go? Many pdfs have text flow broken up so badly that they appear to read randomly. A few bits from one sentence, then a few words or parts from the middle of another paragraph. Literally the best option for some pdfs is to export them as images and import those to an ocr program.
To make the documents accessible, they will need to create them in such a way that the screen reader can read the text for the blind person. Believe it or not, extracting the text contents from a pdf file is actually a very non-trivial problem. Mostly the problems are caused by pdf authoring tools that render each glyph separately. The text extractor then has no idea about which characters belong to each line and has to guess based on the baseline of the character. Another problem is non-ascii characters and how the authoring tool decides to render them. The venerable free software tool pdflatex uses composite characters (basically it renders multiple glyps on top of each other) which makes it impossible to accurately extract the text.
So no, it is not about stupidity or bad Microsoft softare. PDF just is unsuitable for accessable documents.
Football Odds