The Notebook Not Taken
I had the pleasure of seeing Alison Hill talk about computational notebooks last week. Jupyter, Quarto, Wolfram Notebooks, and the like are now many scientists’ preferred way to think in code. Having used several, I can’t help but wonder if there’s a universe out there in which we took a different path. Instead of starting with Markdown and slowly edging toward a full-featured editor, did someone on Earth-978 write a plugin for LibreOffice to run code and insert its output into the document? It’s technically feasible: there’s no reason something like the Jupyer protocol couldn’t have been invented with a WYSIWYG editor as the first front end instead of a browser.
Earth-977’s history took yet another path. There, Microsoft included what they called a “computational bridge” in Office 2007. It was designed to help people create automated reports, but scientists almost immediately adopted it as well: most of them were already using Word and Excel, and found that learning a bit of VB.NET to push around their dataframes was a lot easier than shifting to an entirely new suite of tools. By the time Google Docs appeared, active code blocks were as normal (and as expected) as drawings and tables. A few holdouts continued to use Python and R in text-only editors and IDEs, but once a consortium sponsored by Microsoft, Google, and Apple implemented diff and merge for office documents, the battle was effectively over.
I suspect our universe unfolded differently for two reasons:
Most computational scientists don’t know Java or .NET.
Most programmers look down on WYSIWYG editors. As a result, auxiliary tools from
grepto Git can’t handle things that aren’t backward-compatible with punchcards.
I believe the LibreOffice path is still viable in our universe. The unseen 99% of scientists (data or otherwise) don’t yet use computational notebooks of any kind. With the distinction between desktop and cloud growing ever blurrier, and with so many of the pieces needed for this approach already available, I think a startup could make a compelling case that accountants, marketing executives, and others would prefer something evolutionary over a browser-based dashboarding tool or something as alien as today’s notebooks.
Note: this post was inspired in part by my move from Twitter to Mastodon. The differences between the two have got me thinking about how chancy and evitable our technologies are, and about how many alternatives we have yet to explore.