Structured Online Laboratory Notebooks
January 18, 2007 – 4:11 pmI spent some time last week with a group of medical imaging researchers who manage their work with Excel. They explained that it’s almost ideal for their needs, since they can use cells to organize their experimental results, diagrams, code snippets, and whatever else they like in whatever free-form manner best suits the problem at hand. Unlike a web browser, Excel can do simple calculations and re-draw their charts; unlike a database, they can merge cells to create irregular structures if that’s what the problem calls for.
The two things they can’t do are:
- Merge independent edits. Excel uses a closed-source binary format, so all version control tools can do is say, “These two things are different.” (I’ve complained about this before.)
- Interact with their data over the web (more particularly, share live documents with colleagues).
A few years back, scientists in the US national labs were working on an uber-tool that would do all of this and more. That effort is still going on, but (a) it seems to have slowed down, and (b) it’s a big-hammer approach. I’m wondering if there’s a role here for an in-browser spreadsheet whose data format is diffable and mergeable. It’d be useful for much more than laboratory science: course grades spring to mind, as do address lists and a host of other applications… Anyone want to be rich, famous, and popular?
3 Responses to “Structured Online Laboratory Notebooks”
The problem with excel is not that the format is binary, but that it is closed-source. Isn’t it the same with Google’s spreadsheets? How is being in-browser better for me in this context? the format is still closed.
If you save stuff in XML, it’s not closed. You can do a line-by-line diff, but it’s not a nice approach. You can do more meaningful things, like node-by-node comparison, showing a list of modified nodes using xpath language, and so on.
With XML being an open standard, I expect that generic XML diff tools will improve in the future. However, for specific domains, like spreadsheets, you want more than that. You want to show the changes at the same level that they were made, right? I am not talking just about textual changes. How would you show, for example, that the color and font of a cell changed?
It is all about the transformation between presentation layer and storage layer. In order to show the diff in a domain specific way, you need to reverse that transformation.
I guess the reason nobody got rich & famous for this yet, is that every application/domain uses a different transformations.
By the way, even with text-based diff, not all meta-data changes are trivially visible. For example, when using Eclipse, why is it that both Subversion and Subversive can’t show you the changes to svn properties such as svn:ignore?
By Yoni on Jan 18, 2007
Sounds like a job for Dabble DB…
http://dabbledb.com/
(Specifically, the “You can export all of your data in several different formats” likely includes a diffable format.
Later,
Blake.
By Blake Winton on Jan 18, 2007