For the Love of Spreadsheets

[The opinions expressed represent those of the author, and cannot be attributed to ISO-TC184/SC4/T25 or any of the mentioned organizations participating in ISO15926.]

Many a CIO and Information Management professional has bemoaned the spread of spreadsheets containing business significant information in applications beyond the grasp of their management processes.

In well regulated organizations business critical information is managed in appropriately robust systems, but ever since commercial spreadsheet applications became part of personal desktops, accountants, engineers and managers (and a lot more besides) have developed useful mini-apps and data-bases – some not so mini.

The reason they are a mixed-blessing is obvious: However personal and useful they become, they lack any referential integrity, have at the very least doubtful normalization and redundancy, and what’s more the very flexibility that makes them attractive to create, means that quality control and validation (if any) of content is easily compromised by ongoing modification of both the structure and the content. All of that control and validation is configurable of course, but few using commercial spreadsheet tools start out with that as their plan.

But, before throwing baby out with the bathwater, it’s worth pondering the attraction of spreadsheets and distinguishing distinct aspects in these days of not only SQL-based relational databases but non-SQL, semantic-web and big-data technologies.

Firstly, it is necessary to distinguish between organization of data as a spreadsheet vs the ubiquitous MS-Excel tool most people use in the nineties and noughties to create and manipulate them. Secondly, it’s worth noting that spreadsheets are as consistent with the semantic web as any other mark-up language – yes, spreadsheets are a perfect mark-up format.

Drawn in the sand, scratched on slates, or cut into soft clay plates (choose your tabula rasa), spreadsheets are probably older than the Mesopotamian accountants crossing the Red Sea to Egypt to lend a hand with managing The Pyramids projects. With a 2D projection in the 3D world before your eyes, the spreadsheet is a natural way to capture and organize the information in your head, and share it with the next person. And even if you don’t plan ahead, there is always a column to the right or insert to “mark-up” what that next person thinks about what you propose. Better still, the [column, row, cell] structure is as pure a “triple” as anything the semantic web technologists could dream-up. And that’s because the semantic web as a graph of triples, long predates the technology, long predates Tim Berners-Lee’s vision of the WWW in fact. The predicate logic of triples is as old as Aristotle and the semantic web was coined by Foucault in “Les Mots et Les Chose” in 1966 if not earlier.




(Figures above, acknowledgement to Dr Thore Langeland, EPIM)

What this tells us that a spreadsheet is an excellent human user interface, even if MS-Excel gives CIO’s heart-attacks.

Of course they do say every picture paints a thousand words and every picture tells a story, so graphical user interfaces, particular dynamic graphics can be even more powerful. But what is often overlooked about these adages, is the words / story / semantics have to be implicitly understood before the picture / graphic / icon exists to bring the story to mind. Check the original “Printers’ Ink” source of the thousand words adage. You can always create better GUI’s but the human users need to understand their information content – the GUI doesn’t necessarily “contain” the information.

There are only so many ways information can be organized efficiently in a 2D projection. Formatted text, list reports, data-sheets – before pictorial representations become the obvious choice (but see above). Graphs are attractive in themselves, particularly for navigation, but have two problems – with the efficiency / information density as labels or icons on nodes and arcs, or with multi-dimensions that criss-cross in 2D, if not reducable to nested hierarchies.

Spreadsheets (multi-record-tabular-list and single-record-data-sheet forms) represent a good UI format for reporting from any well-structured application, and a mark-up format for feeding updates and additions back to, or directly into, such applications without any compromise of semantic needs.

Let’s be careful not to throw baby out with the bathwater.

[Update 21 March 2013.  Post Note – magazine story on the power of spreadsheets.]

[Post Note 26 Sept 2017. Even J K Rowling.]

[Post Note 4 Nov 2019. And Whaddya know, Spreadsheets to be the next $bn thing outta Silicon Valley.]

  1. #1 by Ian Glendinning on November 4, 2019 - 07:50

    Latest news – spreadsheets next big thing:

(will not be published)