nullstream weblog - Google OS / thin clients / hosted apps

« That 3D Desktop Acceleration is so hot right now | 1600x1200 Desktop Wallpaper »

Google OS / thin clients / hosted apps


March 25, 2005 12:41 AM PST

Every week more stuff about Google building their own OS, yadda yadda yadda. Here is just a sample:

Article on OSNews.
Another one on Cnet
One on Microsoft watch about Ajax killing the smart client
And another blog post just being contrary to those above

I donít really want to speculate about if Google is or is not doing it. Paul canít comment anyway. Iím more interested in discussing if there is a good reason to do it. And if there is, how to do it. And by Ďití Iím talking about a thin client, hosted app model where all of your apps and data sit on some big server farm far away. Oh and by Ďyourí I mean some Ďcommoní user, not uber geeks that live and breath tech all day. Any thoughts? As always Iíve got my own opinions but Iíll hold them back for a bit.

Comments (7)
Paul, March 25, 2005 02:03 PM:

In my 20% time, I'm building an ark which will contain 2 of each kind of web page. When the revolution comes, the GoogleColonyShip.exe will launch (in beta) and the universe shall be ours.

Paul, March 25, 2005 03:40 PM:

Here is something related that Joe Beda (former Avalon lead at MS, now at Google) is saying:

George, March 25, 2005 08:30 PM:

Will there be a mac version?

Paul, March 25, 2005 11:29 PM:


J, March 27, 2005 10:43 PM:

Nice set of articles, John. I'll post my notes below for what they're worth. If anything, blogging gets me to think about things and present my conclusions clearly. So it remains a useful excersise.


The first was obviously written by a googly-eyed fanboy (pun intended): "All of it is free, all of it is easily accessible through a powerful, extensible web browser, and all of it simple and easy to use (it's still Google, remember)."

Google has done a decent job of a DHTML interface, but it's certainly not the best work I've seen out there. Picasa is the nicest UI they have, and they didn't write it. Did they write google maps, or buy the guy who did? Regardless - they've made a richer web UI mainstream (as the original post shows), and these interfaces will become more commonplace.

He's awfully overzealous about ads and free content: "Or imagine a database of freely-downloadable music, from top artists, television shows, news videos, etc., paid for by inconspicuous Google text ads."

I don't know who's providing all this content for free, or how much revenue is going to be generated by ads that people find inconspicuous. Why not just drop $100 on an OS every couple years?

In summary - the big picture this guy is missing out on is that no one wants one company to control all this stuff! The beauty of the interconnectedness of the web and web applications, is that through a browser you could interact with an email provider, word processor, IM, etc. You don't need one company to provide it!


I see the argument of storing your data on highly available centralized servers. But there's a bigger problem than reliability or privacy that the article doesn't mention - choice. If my data is stored on Google's servers - it must be accessible programmatically through 3rd party apps (via Web services presumably). This means that as new, better Web apps are written by 3rd parties - Google becomes an online data store, like .mac. The drawback for 3rd party software providers is they will not have access to modifying the back end features, and we'll start hearing about MS-type "hidden apis" used in Googles web apps. :-)


I agree that "ajax" is an interesting methodology. It's another step in Web UI development, but I don't see it as that revolutionary. Vs. Microsoft and .net, I like it because it is SIMPLE. In the current environment of rapid software development and change - SIMPLE rules because people understand it and push forward. No one successful is sitting around on 5 year development projects for some complex, proprietary technology (see MS and GA).

I like the comment from MS about how web apps don't handle disconnects. Javascript can access local cache. And if anything - Outlook and Microsoft Domains are the worst example of offline networking detection I've seen. I love how the entire screen will freeze for 30 seconds and Windows explorer is completely non-responsive if you click on an unavailable network drive - it's called background threads or non-blocking software Microsoft! Use one of those 30 outlook threads for something useful.


I like this article - it is even keel and while giving respect to Google, doesn't over-laud them.


I've been totally amazed lately at the lack of creative thinking in technology. Everyone seems to be sitting around waiting for Apple to release the new "superchinpokomon iPod phone", or for Google to release a new Beta web site. Outside of these companies - what are other hardware and software companies doing? I think people want the security of one or two big companies doing things. They just don't trust the millions of new cool "dog food over internet" type companies that were going to change the world. As usual - this is swung over too far the other way. Somewhere in the middle, where a large company creates a de-facto standard, and independent developers create the "next paradigm" is where it will go.

It's not going to be a situation where "one company harnesses the Internet". The Web is a platform that no one will control.

And for those keeping score, my earlier rant on MS seemed more vehement than my negativity towards Google or Apple. So there, Paul!

john, March 28, 2005 12:47 AM:

I agree with many of your points, but I'd like to take a this discussion in a different direction.

Ok lets start with the 'should' it be done question - refering to a network base OS or thin client model. A good way to start this is to first determine who the target market would be. I think that almost every existing computer user can come up with a big list of reasons why they would not be interested in this model. The arguments range from I don't want my data on someone else's server, to you would never have enough bandwidth to do full apps like video editing, or playing games. Fine, all valid, lets move on. In fact let's just come out and exclude ALL existing computer users from our demographic. Whoa you might be temped to say, who does that leave? Well only the majority of the people on this planet.

The truth is the majority of the population cannot afford a computer or the software to run on it. For this audience even a low end PC with open source software is out of reach. So how do you solve it? One approach is to offer the computer, software and
internet access as a subscription for a small monthly fee. If you use follow the standard PC / OS model the hope would be to make up the initial hardware cost over time. The initial cost would have to factor into the monthly cost in order to try to break even as soon as possible as well as offset cost of maintenance upgrades and service. One risk with this model is that someone could just take the computer and re-format the drive and have a fully operational machine and no more need to pay for the service. Approaches to counter this range from encrypted bios to filling the whole box with plastic. None of which is perfect. The other model to take however is to build a really low cost thin client computer that couldn't be used as a standard desktop PC. By driving the cost out you could even lower the monthly subscription making it even more affordable.

So what kind of experience could you offer, and would it be good enough to fill the needs of the target market? It really comes down to what do the majority of people do with their computers? My guess is that 80% of computer owners get by with doing less than 50% of what the computer is capable of. If that 50% use case lends itself to being either web based or thin client hosted (remote desktop, remote X, VNC) then bingo you have success. Don't forget you are bringing this to people who don't have computers today. They are gaining 50% not losing it.

So what would the model look like? Lets start with the use case again. What do the majority of people do with their computers? Again, not you and I, but the rest of the huddled masses. Here is a brief list:

Surf the web. We'll divide this in the basic web, and fancy pants web, and get back to this later.
Send and receive email.
Use IM.
Author basic documents / store and retrieve information. (info apps)
Manage their life / money / taxes / business (manage apps)
Download, store and playback music.
Download, manage and view photos and digital media.
Play games (simple)
Play games (fancy pants)
Play video. DVD, streamed or other.

Now how much of this would work in the thin client model. Again we should probably start by at least setting a minimum bandwidth bar. My guess is that for most of this market we won't even be able to leverage the an existing phone line infrastructure. I could be wrong, but lets start by solving this with wifi and wiMax. It is a low cost way to cover long distances with good bandwidth etc. But lets assume we need to share it amongst lots of users. Lets set the bar at 128kbit, with reasonable latency. What can you do with that? Well the basic web is no problem. Heck most people are dealing with the web just fine on phone lines under 56k. What about remote hosted apps? Well I know from experience that remote desktop at 128k is perfectly usable. In fact I was able to everything my job required across this type of connection no problem. The only limitations from a normal 'work' experience are music and video. For those you would need good dynamic streaming on demand solution. A solvable problem.
Full games on the other hand, oh well can't have every thing.

Ok this is ready longer than I wanted it to be, and I'm running out of steam. You can fill in the blanks. I guess the bottom line is that I think there are interesting opportunities out there that most of us don't see because of our own bias. I think that the first few companies to see the formula and get it right will be very successful, not to mention expand the lives of a lot of people.

J, March 31, 2005 12:55 PM:

My quick response is that "Ajax" provides a richer interaction, but that was already there with Java. The issue with both Java and Firefox/Ajax/XUL as a platform for developing countries, is that the hardware requirements are really steep, and bandwidth is extremely expensive and hard to find.

So it makes more sense to offer a $200 machine that you can plug into the TV, or recycled 15" monitors from more modern countries. This machine would run linux and free, open source software. An added benefit is you'd have a huge new market for developers of new products on this platform where the hardware was cheap and the software was free.

The problem with a hosted/network based solution is someone always tries to assert control (be it through censorship or financial). Open platforms are much more scalable and cost effective to reach the market you're talking. I agree that wifi/wimax/phone could all work for the raw pipe.

For high end games, kids can go to their local internet cafe and play them there. This is already very popular in countries where the average person can't afford a high end PC.

I guess the final question I have is: do people really need computers for this stuff?
+ Surfing the web could be slow/expensive depending on bandwidth. Also governments are often concerned with censorship.
+ Many non-tech people prefer to talk on cell phones than email/IM.
+ What is desktop publishing buying most people?
+ Do they have a digital camera or other digital content creation tools?
+ Cell phones have games
+ TV has video (that actually works)

The main thing I see this being commercially viable with is in what you called "Manage their life". There are so many small, family businesses in developing countries. A small point of sale type computer that would have a spreadsheet, money manager, credit card reader, and other basic business tools would be really interesting. It's actually an idea my uncle in Thailand has talked about for years. You wouldn't even need networking on board. The vendor could take the machine to his bank and plug it in to upload all the transactions and stuff. Or with a serial cable, connect to a cell phone and send out emails that had been batched up.

Maybe you're not talking about 3rd world, but computerless people in the states. I think Mac Mini (or would someone do a linux PC clone of this thing?) and a modem fits the bill. No thin client needed.

All links will be marked with the nofollow tag, making them useless for search rankings. Any posts containing spam URLs will then be deleted.