Analysis After three years of monastic silence, Lotus Notes' creator Ray Ozzie unveiled his latest project Groove Networks in New York this week, and instantly became a kind of godhead for the peer-to-peer networking buzz.
P2P, doncha know, is going to be the defining computing model for the next ten years ... or so a very powerful and well-heeled portion of self-interested CEOs, analysts and VCs (Esshhther!) will tell you.
So after the dust had settled, and the tech media played its: "I saw this before you did!"/"Yeah, but I knew Ray before you did!" games, we had a poke around.
If you don't read any further (and it only gets interesting there), then take this away: Groove is indeed a very well-thought out reimplementation of the best Lotus Notes ideas - for creating ad hoc collaborative workgroups with some messaging prepackaged - only this time for the Web, rather than the corporate LAN.
Like Notes, it's being promoted as a stealth product: as a client that needs only some minimal overhead at the server (yes, Notes admins can stop laughing now, but that's how Notes was originally sold, believe it or not) which comes with a platform attached. But it fixes one of the features of Notes, in making spontaneous workgroups, or work sessions, much easier to arrange.
Get into the Groove
Groove taps into the same management zeitgeist that Notes tapped into, and really it's a very good one. You'll know it pretty well, but the terms decentralised organisation, dispersed knowledge, ad hoc workgroups, worker empowerment, hive mind, downsizing(er... scrub the last one ...) were all familiar ten years ago, and they're back again now.
Groove is a downloadable client (Windows only for now, but Linux to follow with a wink in the direction of Mac OS X), a platform and an API. It piggy-backs onto XML-RPC or the Microsoft-backed rival SOAP protocols and, as a remarkably refreshed-looking Bill Gates pointed out in a video testimony, it hooks into COM too. This must be the first public mention of COM by a senior Microsoft executive in two years, and a fairly good indication of how cocky the old guard is (Andy Groovem sorry Grove, also leant a hand) is about P2P swinging the fashion back to fat clients.
However, despite the vested interests and the buzzword glue, you can't fault Ozzie for lack of consistency, or for shying away from the tricky technical challenges.
Groove focuses on packaging the really hard parts of such a framework: synchronisation and security, just like he did in Notes. And if you doubt that these are anything less than critical to the success of such a model, then hark back to the winter of 1995, shortly after Netscape's launch, when received wisdom had it that proprietary client/server groupware would in short order be buried by whizzy, IETF-blessed patchworks of Internet software.
Today, Notes has 60 million users, and Collabra (or your fill-in-the-blank protocol-centric alternative) has considerably fewer. The fact is, although Internet standards have taken us a long way, for a particular kind of buyer, they don't go all the way unless they're wrapped up.
In fact at the launch one of the more ironic testimonies came from Viant's CTO, citing "my friend John Perry Barlow" (a warning alarm bell went off for us there) who proposed "the Net has eliminated containers... containers are a side effect of technology".
Absolute garbage, of course, as Groove is nothing if not a container. It is of course a very neat container which packages all of the tricky parts of business processes that commodity internet protocols haven't quite got licked.
There are a couple of interesting side effects which the P2P bandwagon hasn't yet addressed. And it will need to if it's going to roll that much further into the future.
Into the Groove
First, what parts of Groove will the company make free available as free software if only as a bait to accelerate the platform?
Groove could adopt a model similar to Napster, where it it tolerates software libre client clones while keeping other protocols under wraps. But that would be a hard sell to business customers.
Second, Groove does not address trust metrics. For now it's an infrastructure play that leaves aside how people collaborate.
These days it is a lot harder than in the early days of Notes to chooses likely collaborative partners when creating ad hoc groups as there is much more information media to choose from. Who's smart, and who's a clown? Working this out should be transparent, and the science is evolving pretty rapidly. Whether Groove intends to swallow such trust metric logic into the platform, or leave it for third parties, will be its next test.
And finally, and this is a question to which all P2P brainstormers should have a some kind of answer, is whether you'll really be able trust the data you're working with.
Sun's chief scientist John Gage told us back in June that he thought one of the biggest headaches in computing in fifty years would be disappearance of the canonical text. He didn't just mean bit-flipping, as any regular Napster user will confirm.
Faced with several dozen versions of Heartbreak Hotel, many of different lengths, recorded at different bitrates, how do you know which version is the definitive recording? Which one drops the start, and adds dead space at the end? At least the central server model ought to guarantee some redundancy. The redundancy in the Napster model is really a redundancy of incompleteness.
These questions alone should be enough to sober up the quite frantic P2P hype - at least a little. For now though, Groove appears to have made an excellent debut, particularly in making its developer material available and friendly. ®