We are a couple of weeks on from the Google I/O conference now, and many have had a chance to play with Wave. It is an impressive product, but still with some months/years to go before its ready for the mass market. Missing is the essential email integration, but when it comes the combination of Google Apps, AppEngine and Wave will be awesome (most likely). Apps is Gmail, Docs, Spreadsheets, Presentations, Chat and private youtube. AppEngine of course presents cheap hosting of apps, but most likely in the future will allow for custom underpinning of more Google enterprise services than 'just' Wave. Lotus Notes (it just refuses to die) and other legacy client/server collaboration 'solutions' are most definitely wrong footed. IBM/Lotus should rightfully be scared.

ThoughtWorks was at the I/O conference in force of course. Rebecca Parsons and Martin Fowler gave an enterprise leaning roundup of AppEngine and the cloud per se. Ola Bini chronicled his experiences with JRuby an Ioke for the Java AppEngine launch. We otherwise manned an AppEngine booth regaling passers by with our experiments, and lapped up the sessions. The conference closed on Thursday, but Ola and I went down to the Google Wave hackathon on the Friday to write appengine robots for the Wave sandbox. During the introduction to the technologies, Ola wrote an Ioke interpreter robot on AppEngine using the Wave APIs such that the robot could be invited to join a Wave-conversation. We sat next to groovy Codehaus buddy Guillarme Laforge for the day, and bumped into Ben Galbraith (never met before) Dion Almaer and ever reliable Dave Astels on campus. The Wave team ordered in an assortment of wraps and cookies; no more legendary lunches for blow-ins it seems.

AAWsome

Google is missing lots around naming though. They are too close to the woods to see the trees with many of their service/product names. Is Wave a service, a protocol or a doc-let thingy in a service? Google-Apps also has a dodgy name; it really should be 'Google Enterprise Apps' given what it comprises. Similarly 'Google AppEngine for Java' should be GAE/J or something snappy. It seems to me that AAW is a decent interim roll up of Apps, AppEngine and Wave, while we wait and see how they all work together. Unless it is messed up, it will herald a change for all enterprises and their collaboration habits. Some sooner that others, but all of them in time.

My mostly off-target Notes prediction.

I'd blogged before about Lotus Notes being core to Google plans for AppEngine. Well the blog entry was about the AppEngine launch, but GWT and Lotus Notes and AppEngine were a sidebar topic. The product I proposed to our studios team to development some three and a half years ago was Notes redone based on open standards (IMAP, NNTP) with what is now Cozmos for collaboration and what is now Swiby for easily constructed end-user apps. Obviously Wave is completely different, and much more advanced in its vision for the bits that are complete. I can't help but think that its weaker though in that its solely based on web technologies, and could well never have an off-line capability as Notes and our product would have done. Maybe offline comes next. But my prediction that GWT would have a truly fat mode for a Notes-like capability is way off target; Google wedded HTML5 in a private ceremony some months ago. So enthusiastic was the announcement of the wedding on the 28th May, that you would think Google were launching the Pre platform instead of Palm. Both GWT and Android feel odd amongst the Google range of technologies, given the love of HTML5 as a technology you would write for.

Wave's pervasive inbox

Back to Wave. When it is finished, Wave will present a pervasive Inbox for many items including email. Whilst its well known that multitasking sucks for actual productivity, its clear that it is here to stay for a good chunk of knowledge-workers days. Items in their in boxes will not be replicated though as many recipients as were crammed in the TO/CC lines, instead for multiple recipients at a single corporate destination, there will be as little as one server-side copy of the wave incarnation of the incoming email. Moreover for spammy items that are time sensitive, they are as likely to disappear from the inbox when they are suitably superceded. Imagine the build-daemon Cruise (one of the products we did build instead of many other proposals) spamming team member with build failed / build passes emails. For normal folks that is represented as a long list of similar subject line emails. For Gmail users (who are online versus IMAP) they are at least rolled up into the same subject-line groupings. For Wave if you go on vacation and come back after a week of repeated build breakage and fixing, you'll note that the summary is one line for Cruise:

From: Cruise Agent; Dates: Mon-Fri; Synopsis: Mingle 3.0 build broke 17 times, and was fixed 18 times, mfowler appeared to top perpetrator, recommend for pair-programming

In total 35 incoming 'emails' hit the pervasive inbox, but a Robot got busy while the recipient was soaking up the rays on a beach, extracting information from the incoming emails and rolling a summary for ultimate presentation. The robot did not necessarily know that the recipient was on vacation, but it did know that they were not reading email for say more than a day, so got bust digesting. Of course, Cruise make end up speaking the Wave protocol to a Wave doc-let-thing on the Wave service natively and not do any SMTP at all. SMTP has to die by the way, I'm not sure why we can't just use a send folder within IMAP.

Lastly, a word from my sponsor: ThoughtWorks looks forward to helping clients out of the Notes lock-in misery and into the AAW platform if the client wishes for help with that goal :-)

June 19, 2009