A bit more on Wave as a Scholarly HTML editor

This is a quick follow-up to my post last week about using Google Wave as a collaborative editor for scholarly works. I will start with some impressions of using the thing then go on to look at the Wave model.


Cameron Neylon has invited me to a Wave looking at a scientific article; he’s pasted one in to Wave as an experiment and there’s a discussion going on about how to structure things. I have to say that for a document of this length the performance is abysmal, looking at this on the weekend at home on a two year old low-end desktop machine I was unable to use the system at all effectively even scrolling is jerky and annoying and the replay is unusable. Also, it seems that for Wave to take off people are going to have to hang out there like they do in email or Twitter. So far this doesn’t seem to be the case. I have been invited to a couple of things but there’s not much happening so I forget to open Wave, which means that, well, there’s not much happening. This will change when we have our whole team in there, I guess. So this week’s impression is it’s:
  • Going to take a while to generate critical mass in groups of users. (OK, so that’s obvious.)
  • Too ‘messagey’ to be useful as a document editor. The idea that robots and gadgets will be useful, as discussed by Cameron Neylon seems right, but I think we might need to use those in a different client I’m not sure if that will actually be a Wave client, though. Even though Google docs also has awful HTML it’s much more workable as an editor. Can gadgets run there?

The Wave (non)Document Model

I said last week that I thought Wave had some potential as an editor that could output HTML even with its dodgy HTML, but I was wrong, I think. As Richard M. Davis rightly pointed out in the comments, we really do need editing tools that can understand basic structures like lists. I agree that if we are going to have useful Wave robots then they will need to be able to work with proper structure, like being able to identify the quotes in an article. The thing is that I had Wave completely wrong. It’s a messaging system with a very tenuous relationship to HTML. It’s a bit hard to find documentation on exactly what is going on behind the scenes, but the Draft Protocol Spec for the Google Wave Conversation Model makes it clear just how far from being an HTML editor this thing is. Here’s the example of a ‘blip’ document:
<contributor name="dadams@acmewave.com">
        <line></line>There is a theory which states that if ever anybody
        <line></line>discovers exactly what the Universe is for and why it
        <line></line>is here, it will instantly disappear and be replaced by
        <line></line>something even more bizarre and inexplicable. There is another
        <line></line>theory which states that this has already happened.
This is very discouraging from a document point of view. It has explicit line breaks in what looks to me like a paragraph. This probably explains why there is the weird (<p>text<br></p>) markup in the HTML source of a Wave. Waves in real life don’t seem to be broken up like this, it’s almost like the body element in this example is equivalent to a p element in HTML. This is more like a model for a chat client than a word processor, and we need to bear this in mind. I have yet to figure out the relationship between this line-based model and the formatting buttons in the wave editor. Regarding structures like quotes or lists I had an idea that maybe it would be possible to use Wave’s inherent nesting/threading to make structured documents, so you’d have a ‘blip’ when you want a list and embed another in it when you want a nested list but as you can’t copy or move blips, and you can’t put them between paragraphs (or lines) then that’s unworkable in the current version. Unfortunately, I think that Wave is too much a messaging system to be a useful editor in the short term. That’s not to say it won’t be a useful collaboration tool, but for drafting long structured documents I’m very skeptical. Quite a while ago Stijn Dekeyser and Richard Watson here at USQ looked at using Google Docs to author papers and concluded that the collaborative tools were great, and it would make a nice text editor in which to co-author LaTeX. I think maybe Wave would work for LaTeX or Markdown or another wiki-like text-based format better than as a WYSIWYG HTML editor at this stage; with collaborative editing of the marked-up document, and rendering as a separate process. At the time of posting I have a handful of invites to Wave, so drop me a line if you’re interested. © Peter Sefton 2009. Licensed under Creative Commons Attribution-Share Alike 2.5 Australia. <http://creativecommons.org/licenses/by-sa/2.5/au/>