google wave

[iMav]

Check it out here.

I've actually been thinking a LOT about Google Wave since I watched the demo video.  Google has a serious chance to succeed in making this the next generation, open email/IM/collaboration/etc platform.

Assuming they keep their word on wanting it to be open (and releasing all their work as open source), I could easily see companies moving to this type of solution.  Migration would be fairly easy.  Assuming they would rely on DNS service records and standard email address format (like the open standard jabber IM does), when you add a participant that is not a member of one of your domains, the service record is looked up for their domain.  If their domain provides a Wave service, great, add them to the conversation...if not, look up the MX record and either forward a copy of the conversation to them (non-optimal) or simply email them a link to an external portal page with an embedded Wave client...then they can participate fully in the ongoing conversation.

This also makes encryption dead simple to implement seamlessly.  Keypairs could be deployed to handle the encryption to multiple recipients.  Every time the conversation is updated, you re-encyrpt with the participants' public keys.  Every time a new participant is added, you add their public key to the mix and re-encyrpt.  GPG/PGP would be perfect for this.  Add in the afore-mentioned external portal and SSL...and you just put a ton of email encyrption vendors out of business.

And how this changes the spam game simply makes my head hurt.  If this does take off...one thing is for sure.  Google/Postini will be out in front of it and the rest of us will be playing serious catch up  

Personally, I think it make a great infrastructure for the next generation of web communities.  Networked sub-forums between sites would be simple to implement (think 21st century FIDOnet).  Lots of cool possible uses.

[o2dazone]

I'm the same way you are, since the day they released the demo I've been thinking about how it's hilarious how it hasn't been done earlier. How something like instant messaging is so similar to email, and how the two are treated mutually exclusive (even though they're amazingly popular). I also like the keystroke sending thing, making chatting with someone a little more engaging (opposed to send a message, then minimize and do something else for a while, and then come back to it)

I'm hoping people don't complain thinking it's just another 'social networking website'. Wave has tons of potential, looking forward to it. It should also give a violent shove into making browsers support HTML 5 (which now has open video...oh god - goodbye Flash)

[Mr.6502]

I'm excited to see it in action and can't wait to set up my own Wave server to try stuff out with.  I think it has some potential to be very fun to use alongside or in place of a web forum.

I've applied for sandbox access, hopefully I'll get it.

[majestouch]

Looks like fun, though this is more of an evolution than it is a *new* product, it basically merges threaded discussion forums and IMs; it doesn't obviate standard e-mail, but shows a bit of potential in the ease with which you can create an isolated communication structure that is easily managed by it's members.

From the video, the ground breaking part seems to be the control and visibility the google API provides; they're good at polishing ideas other people have had, but just couldn't implement with finesse. So, it pushes the envelope, but there are some technical hurdles with some of these ideas; things like drag and drop from the desktop and the "live typing" don't seem quite practical yet. HTML 5 is still a ways off and to actually see each character as another person types it requires either slow typists, or enough data padding to push a TCP packet out, or selectively setting TCP_NODELAY in your socket opts if you just happen to code your own browser:) - but for other browsers they'd have to degrade back to data padding to get the browser to push packets fast enough. Either way, you push either a crap load of packets or very inefficient data, ultimately, this kind of collaboration on a large scale could bring current networks to their knees - even with HTML 5 and the likes of "reverse" AJAX you still don't have socket level provisions. The demo shows all of 3 people working together on a closed network, not much of a stress test. Someday this will work, but methinks it will be a while.

[Mr.6502]

Looks like fun, though this is more of an evolution than it is a *new* product, it basically merges threaded discussion forums and IMs; it doesn't obviate standard e-mail, but shows a bit of potential in the ease with which you can create an isolated communication structure that is easily managed by it's members.

From the video, the ground breaking part seems to be the control and visibility the google API provides; they're good at polishing ideas other people have had, but just couldn't implement with finesse. So, it pushes the envelope, but there are some technical hurdles with some of these ideas; things like drag and drop from the desktop and the "live typing" don't seem quite practical yet. HTML 5 is still a ways off and to actually see each character as another person types it requires either slow typists, or enough data padding to push a TCP packet out, or selectively setting TCP_NODELAY in your socket opts if you just happen to code your own browser:) - but for other browsers they'd have to degrade back to data padding to get the browser to push packets fast enough. Either way, you push either a crap load of packets or very inefficient data, ultimately, this kind of collaboration on a large scale could bring current networks to their knees - even with HTML 5 and the likes of "reverse" AJAX you still don't have socket level provisions. The demo shows all of 3 people working together on a closed network, not much of a stress test. Someday this will work, but methinks it will be a while.

I think this is doubly true because of their basing it on XMPP, which has so much overhead.  If its really showing messages a character at a time I can't imagine the amount of resources a server would need to support even a small-ish online community.  

I've made AJAX chats before and typically I set a very slow rate of transmission of messages but have the client play them back.  So every 2 seconds it updates and spends up to 2 seconds playing back the messages its received.  Maybe Wave falls back onto a similar system to try and cut back at least a bit on the number of packets its sending around.

[majestouch]

I think this is doubly true because of their basing it on XMPP, which has so much overhead.  If its really showing messages a character at a time I can't imagine the amount of resources a server would need to support even a small-ish online community.  

I've made AJAX chats before and typically I set a very slow rate of transmission of messages but have the client play them back.  So every 2 seconds it updates and spends up to 2 seconds playing back the messages its received.  Maybe Wave falls back onto a similar system to try and cut back at least a bit on the number of packets its sending around.

Precisely. But...first generation software that degrades well? Bah!

The "open source" talk google is throwing around is very suspicious. The OSS + service model is the only reason a for-profit company would open their code. Google is just not suited for this model. You don't hire best-of-breed programmers and then ask other people to write software for you. It don't make no sense. What makes more sense is that it is only a guise to lure in site developers, a notoriously OSS-friendly crowd, in order to push this technology everywhere before they drop the bomb. The bomb being the rollout of the final frontier of web-based advertising: real-time contextual ad placement right were we pay the most attention - right along side our IMs and SMSes.

Something like this:

"Big weekend wave" (majestouch adds iMav and Mr.6502 to the wave)

me: Hey fellas, there's a vintage computer show this weekend, wanna join?
googlewavebot: Great deals on vintage clothing and accessories - yay!
iMav: Where's the show at?
me: Chicago, at the convention center, I can drive.
googlewavebot: Need car insurance in the Chicago area? insurance4u.com!
Mr.6502: what day?
me: Saturday, 10am.
iMav: I'd like to go, but the wife's got me on lock down this weekend...
googlewavebot: Bail Bonds R US! Get out of jail now and enjoy your weekend.[/I]
Mr.6502: Count me in! I'm looking for an AT cable.
iMav: Sorry can't make it, but call me if you find any stellar deals.
googlewavebot: Tired of cable TV? Stellar summer deals on Digital TV from DISH Network![/I]

webwit - why you wouldn't enjoy this kind of technology is beyond me...it's so...helpful. BTW, of course I meant control and visibility of the user created content; I'm sure the backend of the API will be just as tied up as every other monetized psuedo-standard platform.

[Mr.6502]

Yeah, the more I read, the more I realize I've encountered this sort of technology before.  But in the past it used much more efficient protocols for actually transmitting the data.  Their documentation points to the Jupter network multimedia architecture which Wave's operational transformation model is based on.  

Still, it sounds cool.  If there is an open source version that doesn't include any ads I'll look into it.  If the open source side of it is just the specifications for how the platform functions, I'll probably wait for someone else to do all the work, lol.

From what I've read, this will look very different once you introduce high latency.  When a change is made to a wavelet in the client it transmits that change to the server.  The client then waits for the server to process the change, update its state, and send a confirmation back to the client.  While the client is waiting its caching any changes that are happening in that client.

So if you have low latency, it can be sending updates for every letter you type.  If you have high latency, it'll be sending the data to the server in much larger chunks.  

The client then receives operations to apply to its own stored copy of the Wavelet.  It is up to the client to put those operations in order and since the operations will break down into rather small chunks, it seems trivial to make it play back those operations spaced apart based on the latency to give the appearance of a constant stream of information.  

However, doing that goes against the reasoning behind seeing the person type a character at a time in the first place, which was to eliminate the waiting while the client says "The other person is typing."

[talis]

The big problem I see with the whole concept is that, even if you have one friend/contact that doesn't buy in, you will still need to keep an IM client + traditional e-mail client around.  And that will fragment communications even further.  You'll need to transcribe "waves" for those that don't buy in to the concept, or choose not to use the Google implementation.

For business, where I can see it being most useful, I can see it being 5-10 years before enough people buy in to the concept for it to be really useful.  Even with a perfect implementation (no latency and no downtime)  it will take a long time for a lot of, especially big companies, to buy in.

[iMav]

Fortunately, the "client" is a browser...so "keeping another client around" is a non-issue.  It's no more a burden than getting sucked into another web forum and having PM's there.  

My feeling is that it should be dead easy to create an email notification to those that don't have a wave account...and let them log into a web portal with an embedded wave client.  (the embedded wave client has been demo'ed on the preview video).  

Lars said that there would be an open source server available.  They would release the "lion's share" of their technology and that you could compile your own server with what they make available.  I'm sure the open source community will jump on this and there will be many wave server's available.

Personally, I don't think they will implement ad's at all.  I see a model to allow developers to sell their extensions.  Think iPhone apps.  I could easily see a Google store that allows you to purchase extensions.  The reason they released the development version right now is so that developers could start creating extensions now.  They want lots of stuff available for when they release to the public.

[iMav]

Are you saying it is viral? So when Sandra wants to send a message to John, she doesn't email him as before, but John receives a message that Sandra emailed him from her Wave account and he can see the message by registering to the Wave?

That is how I envision messaging to those that don't have wave accounts (but, obviously, have email).  An embedded wave client portal coupled with email notifications to those without wave accounts.

[Mr.6502]

I believe in the talks about it the developers mentioned the system falling back to emailing people if no method to contact them via the Wave is possible.  It makes sense to suck people in that way.

I just hope when the open source version appears it isn't like Chromium.

[majestouch]

Fortunately, the "client" is a browser...so "keeping another client around" is a non-issue.  It's no more a burden than getting sucked into another web forum and having PM's there.  

My feeling is that it should be dead easy to create an email notification to those that don't have a wave account...and let them log into a web portal with an embedded wave client.  (the embedded wave client has been demo'ed on the preview video).  

Lars said that there would be an open source server available.  They would release the "lion's share" of their technology and that you could compile your own server with what they make available.  I'm sure the open source community will jump on this and there will be many wave server's available.

Personally, I don't think they will implement ad's at all.  I see a model to allow developers to sell their extensions.  Think iPhone apps.  I could easily see a Google store that allows you to purchase extensions.  The reason they released the development version right now is so that developers could start creating extensions now.  They want lots of stuff available for when they release to the public.

I was being a bit facetious about the ad-bot, the developer-app model popularized by facebook and the iphone is a much more likely outcome, but as I described, the bandwidth and packet utilization really is a serious hurdle unless they've done some serious "control-theory" type client degradation because Reverse AJAX & Comet are not the role reversal they claim to be at this point in time, they're still "faked" with periodic client polling, which is very inefficient form of data distribution due constant knocking. You need a very smart algorithm, maybe google is the one to do it...I see the paradigm in communications that something like this presents, but I'd like it a lot better if it were in the form of an RFC.

[iMav]

Anybody get their Google Wave sandbox account yet?

If so, you can catch me at "lherzog@wavesandbox.com".

[chongyixiong]

Sorry to bring up this old post.. but anyone got their Google Wave invites yet?

Let's have a public wave of Geekhack!

And sorry iMav, I don't think developer's @wavesandbox accounts and @googlewave.com accounts are able to talk to each other.. just yet.

Oh yeah, mine's chongyixiong@googlewave.com

[o2dazone]

mine is o2dazone@googlewave.com
I have a couple 100+ blip waves and it can get a little slow in Chrome v4, but in all, it's awesome once everyone is able to be apart of it

[ch_123]

I just hope when the open source version appears it isn't like Chromium.

What's wrong  with Chromium?

[Mr.6502]

What's wrong  with Chromium?

When it was released the source code for it was a long way from usable.  Since then the chromium team has done a lot of work to get it all going as well as they have.  My hope was that when Wave ends up released its source code isn't so far from complete.

[urlwolf]

mine is quesada@googlewave.com

[iMav]

larry.herzog@googlewave.com