Good enough for government work

If effective communication within and among governments and citizenries required these big guns [Word, OpenOffice Writer] then we'd have to find ways to accommodate them. But it just isn't so. A mere fraction of the power of these multihundred-megabyte behemoths suffices for basic communication; the rest is overhead. Software delivered as a service through the Web -- simple, lightweight, and universally available -- is clearly the better way forward.
Once we have an open document format -- and given XML's protean transformability I don't much care which one -- perhaps we can move on to the real challenge. I expect that Web-based software can meet all of the key requirements that are driving this debate, and should do so in ways that are native and ubiquitous. Governments advocating on our behalf should expect no less. [Full story at]

Responding to this week's column, Tim Bray points out that authoring software is a hard problem.

I've never seen software, designed for use by human authors, that has good usability and isn't a great big honking monster. [ongoing: Lightweight authoring?]
Tim, of course, is a diehard emacs user, as am I. So everything we say about mainstream wordprocessors should be taken with a grain of salt, because we don't rely on them as many folks do. On the other hand, folks who do reach for Word to dash off every memo and shopping list have provided us with a valuable source of data that, to my knowledge, has never been properly analyzed.

It's a truism to say that word processors were feature-complete by 1985 and have added nothing essential since then. But we've never connected the dots. Suppose we boiled down a large sample of office documents to their constituent elements and ranked them by frequency. We'd find paragraphs, lists, tables, images, links, basic styling, and then a long diminishing tail of rarely used features.

Here's the proof that tail doesn't matter much. If it did, we would not be successfully pouring the majority of our web writing through TEXTAREA widgets in blog and webmail composers.

For a decade I've been pointing out the vast gulf between TEXTAREA and Word. Analysis of a representative corpus of business and web documents1 should enable us to define a target set of features, and scope the difficulty of the problem. In this case, the right thing to do with the long tail is chop it off. Most of us don't need that stuff most of the time. We do quite desperately need a widget that does the five or six or eight things we all do all the time. And we need it to do those things in a way that's standard across browsers and operating systems, produces valid XHTML, and is cleanly extensible. The W3C isn't the right venue for this work, but something like the WHAT-WG might be.

Analyzing the right document corpus might also dispel some of the MSXML-vs.-OpenDocument fog. Goverments and citizens need technology that's lightweight, ubiquitous, and good enough for everyday use. Defining what's good enough for everyday use would be a great contribution to the debate.

1 Is there even a relevant distinction between the two terms?

Former URL: