The Zombie Stack Exchanges That Just Won't Die
PDF is almost a de facto standard when it comes to exchanging documents. One of the best things is that always, on each machine, the page numbers stay the same, so it can be easily cited in academic publications etc.
But de facto standard is also opening PDFs with Acrobat Reader. So the single company is making it all functioning fluently.
However, thinking in longer perspective, say 50 years, is it a good idea to store documents as PDFs? Is the PDF format documented good enough to ensure that after 50 years it will be relatively easy to write software that will read such documents, taking into account that PDF may be then completely deprecated and no longer supported?
Łukasz Lech
The PDF/A ("Archival" PDF) formats have been adopted by the International Standards Organization as an open standard for long-term document preservation.
PDF/A omits some features (such as proprietary compression algorithms) which prevent the PDF standard from being truly open and specifies additional features (such as embedded fonts) to ensure that a document's appearance stays consistent across platforms.
No file format is perfect, and we can't predict future market and technological changes with certainty, but PDF/A is a robust and well-supported standard for the long-term storage of electronic documents. It's what my institution uses in its digital repository.
(By the way, even for non-preservation PDFs, the format is well-documented enough that non-Adobe vendors and open source developers have written software to view and create PDF files.)
Note that everyday PDFs are not PDF/A compliant by default. If you have PDFs that you plan to store for long-term preservation, you'll have to convert them to PDF/A using Adobe Acrobat or some other software package.
You should also be aware that the recently released PDF/A-3 standard allows the embedding of arbitrary file formats, which could lead to a loss of interoperability if a proprietary or poorly documented item is embedded.
I've done a lot of work on Electronic Laboratory Notebooks, where you're looking to preserve the information in human readable format for 30+ years - often much longer.
The industry has more or less settled on PDF/A. It isn't perfect but it is documented, and there are multiple implementations of readers.
Importantly from my perspective... you aren't going to be the only person who wants to read PDFs in 2060.
So there would be a lot of other people wanting to read PDFs, and people would be able to write a reader for the file format... even if Adobe etc. are no more.
Given those two facts, I think you can be reasonably sure that the market will take care of ensuring PDFs are readable well after we've all stopped caring.
While PDF/A is specifically designed for long-term preservation, the PDF file specification itself is, as noted above, very well documented and a reasonably good bet for remaining accessible into the future, most especially if it includes searchable text, according to the Library of Congress sustainability of digital formats resource.
The same LC resource notes that PDF is a page-oriented format that "is not a preferred archival or master format for images." This means that it's best for textual-oriented documents.
The latest version of PDF (Version 1.7) is recognized as a international standard, ISO 32000-1:2008.
PDF/A (ISO 19005-1:2005 for PDF/A-1, ISO 19005-2:2011 for PDF/A-2 and ISO 19005-3:2012 for PDF/A-3) is especially suitable when the content already exists in PDF format, or can easily be saved as PDF, or even right away as PDF/A (for example, Microsoft Office / OpenOffice offer this feature since 2007/2008).
The focus is on 2-dimensional static final form documents - anything from letters, invoices, memos, brochures, publications like magazines, books to engineering drawings, technical documentation or manuals. Anything that 20 years ago was usually printed and distributed on paper. In addition, the 'preservation focus' by default is on visual reproduction, with an optional conformance level to also preserve text content as properly extractable text (think "Unicode") and also content order and structural semantics.
Preferably as much of the quality, structure and semantics of the original content should be preserved when moving towards PDF/A. This implies for example that it is preferable to save documents as PDF with a 'tagged PDF' option enabled whenever possible. This ensures that the reading order/flow of content is mostly or completely maintained, even on pages with complex layout.
Those who think of office formats, HTML, SGML, ... as an option to be considered: these formats lack the robust static-ness that PDF (which you could well think of as 'digital paper') and repeatable visual reproduction that PDF successfully offers. If you preservation focus is more on capturing "authoring content while being authored" (including some non-static stuff like formulas in spreadsheets or macros in a text processor) then PDF is not the way to go, or when being used (for visual reproduction, easy distribution format or fallback), must be supplemented by the original documents in their original formats etc. (causing it's own set of challenges - how do you know / ensure you can run today's version of OpenOffice 50 years from now?)
PDF/A tools and technologies have become very wide spread since the original release of PDF/A (as PDF/A-1 in 2005). In many cases PDF/A support has been integrated into (document management and other) systems and programs (like office applications), there are lots of options to do checking and conversion on all platforms, desktop systems and servers. An overview can be found on the website of the PDF Association at www.pdfa.org.
All in all PDF (ISO 32000-1:2008), or preferably PDF/A (ISO 19005 series), is a very reasonable option when it comes to archiving final form, static, 2-dimensional documents. Personally I do not believe any other format comes even close for this type of content/documents for long term preservation, but your mileage may vary. For other types of content/documents, other formats or approaches may be more suitable.