TECH TALK: The Network Computer: The Idea Returns

Recently, there was speculation that Google was building a network computer along with its own browser. The New York Post wrote:

The broader concept Google is pursuing is similar to the "network computer" envisioned by Oracle chief Larry Ellison during a speech in 1995.

The idea is that companies or consumers could buy a machine that costs only about $200, or less, but that has very little hard drive space and almost no software. Instead, users would access a network through a browser and access all their programs and data there.

The concept floundered, but programmers note that Google could easily pick up the ball. Already, its Gmail free e-mail system gives users 1000 megabytes of storage space on a remote network — providing consumers a virtual hard drive.

"I think a similar thing [to the got network computer] is developing in a more organic way now," said Jason Kottke, a New York-based Web developer who follows Google's moves. "People are ready for it. Instead of most of your interaction happening with Windows or Mac, you're spending a lot of time with Google-built interfaces."

News.com wrote: “Google has also been rumored to be working on a thin-client operating system that would compete with Microsoft in areas beyond search. Techies have even discussed the idea of Google becoming a file storage system.”

A commentary on ZDNet added:

What Google must do is get itself on the desktop. The obvious Google-shaped hole is local searching, where Microsoft has a history of conspicuous failure. A browser plug-in that amalgamated general file management with knowledge of Outlook, multimedia data types and online searching would be tempting indeed. Add extra features such as integrated email, instant messaging, automated backup to a remote storage facility and so on, and it gets very interesting. That would need considerable browser smarts, but would extend the Google brand right into the heart of the unconquered desktop where it would stick like glue.

By effectively combining local computing and the Web in this way Google would open up multiple revenue models. As well as advertising-supported and subscription services, it could start to offer very effective antivirus and other security functions--your data, safe in their hands--as well as any number of cleverly targeted sales opportunities based on what it knows about your personal file mix.

It would also remove one of the big barriers that stops people moving from Windows to open source. If all your important data has been painlessly stored on Google's farm and there's a neat, powerful Java browser-based management tool to retrieve it, you can skip from OS to OS without breaking into a sweat.

Google may not be the only one thinking about networked computers. A recent story in Business Week mentioned that “AMD is planning to announce as early as October that it is teaming up with contract manufacturers to create an inexpensive, networked PC for sale in India or China. It's part of [CEO] Ruiz's ambitious plan to help connect 50% of the world's population to the Internet by 2015.”

So, is the network computer just a dream or will it become a reality? Given that we already have ever-cheaper computers, cellphones, TVs and gaming consoles, do we really need a fifth device? Will the network computer succeed in its second avatar? Is the network computer idea the harbinger of a deeper shift in computing?

As we seek to answer these questions, we need to first understand what a network computer is.

Tomorrow: What Is It?

Tech Talk | PermaLink

TECH TALK: The Network Computer: What Is It?

Wikipedia has this to say about the network computer:

A network computer is a lightweight computer system that operates exclusively via a network connection. As such, it does not have secondary storage such as a hard disk drive – it boots off the network, and it runs applications off the network, possibly acting as a client for an application server. During the mid to late 1990s, many commentators, and certain industry players such as Larry Ellison, predicted that the network computer would soon take over from desktop PCs, and everyone would use applications over the internet instead of having to own a local copy. So far, this has not happened, and it seems that the network computer "buzz" was either a fad or not ready to happen.

The idea actually goes back a long way however, back to the text-only dumb terminal, and later to the GUI of the X terminal. The former needed no software to be able to boot, everything was contained in ROM, and operation was simple. The latter requires some files to boot from the network, usually using TFTP to get them after obtaining an IP address via DHCP and bootp. Modern implementations include not only the X terminal, but also the Terminal Server in Microsoft Windows 2000 and XP, and others. The name has also evolved, from dumb terminal to network computer, and now to thin client.

Webopedia adds:

A computer with minimal memory, disk storage and processor power designed to connect to a network, especially the Internet. The idea behind network computers is that many users who are connected to a network don't need all the computer power they get from a typical personal computer. Instead, they can rely on the power of the network servers.

This is really a variation on an old idea -- diskless workstations -- which are computers that contain memory and a processor but no disk storage. Instead, they rely on a server to store data. Network computers take this idea one step further by also minimizing the amount of memory and processor power required by the workstation. Network computers designed to connect to the Internet are sometimes called Internet boxes, Net PCs, and Internet appliances.

One of the strongest arguments behind network computers is that they reduce the total cost of ownership (TCO) -- not only because the machines themselves are less expensive than PCs, but also because network computers can be administered and updated from a central network server.

Sun too has said since its inception that “the network is the computer.” There is something appealing about the idea about low-cost, simple computers connected to a centralised computing platform. The network computer has had many names – thin clients, diskless workstations, information appliances. It is one of these enduring ideas in computing that refuses to die and keeps floating back every few years.

The world of today is very different now as compared to the mid-1990s when Larry Ellison first proposed the idea of a network computer. To understand if the network computer can succeed in today’s world, we first need to travel back and see what went wrong when the network computer was first introduced.

Tomorrow: Ellison’s Ideas

Tech Talk | PermaLink

TECH TALK: The Network Computer: Ellison’s Ideas

Oracle CEO Larry Ellison first touted the idea of a network computer as early as 1995. Ellison introduced his vision of the network computer, a small, inexpensive device that makes it easy to run applications that access information via the Internet.
Wally Bock takes up the story:

The reasons were obvious, at least to [Ellison]. PCs had gotten too complicated, he said. And besides, every fifteen years or so there's a new revolutionary product in the computing business that replaces what went before.
Network Computers were supposed to be slimmed down versions of Personal Computers. They might have a screen, a microprocessor, some memory chips, a keyboard and a mouse. The most important component would be the network connection.

That magic part would connect the Network Computer to the Net. There would only be rudimentary software and memory on the Network Computer. Most software and serious memory would be out there on the Net where it could be easily maintained. The system would run on Java and use Oracle databases. Microsoft software would be nowhere in sight.

The idea caught on with the industry insiders, journalists, venture capitalists and other trumpeters of the great Internet Bubble. In mid-1996, Business Week devoted a Special Report to Network Computer.

By then there was a price target, $500 and all kinds of companies were lining up to make products that would capitalize on this powerful Network Computer trend. Bandai, a Japanese company, announced that it would soon be selling a $600 Network Computer in the US. Apple had designed one called the Pippin.

There were only two figures in the computer business who didn't share the enthusiasm for the Network Computer concept. Andy Grove of Intel hedged his bets by having teams work on projects that would set Intel up to be a player if Ellison was right. Bill Gates set Microsoft on a path of developing its own solutions.

Larry Ellison said that Network Computers would be widely available in 1996 but by mid-1997 that hadn't happened. Gateway had released a Network Computer in May but that was about it. Still, no lesser authority figure than the Economist was saying that "there is broad agreement that NCs are indeed the future."

Not much more happened in 1997 but in 1998 the Economist was still predicting that the PC would be "entering its twilight years by the beginning of the millennium." The idea of the Network Computer as an "information appliance" had caught on and there were predictions that the folks who would really shift the Network Computer revolution into high gear would be companies like Sony.

Well, as it turned out, network computers didn’t really happen. But the idea refuses to die. Ellison resurrected the idea. In November 1999, this is what News.com wrote:

Though hotly debated in computer industry circles in 1996 and 1997, the network computing concept failed to gain a market foothold, in part because PC prices suddenly fell to historic lows--lessening the need for new, low-cost systems. When Ellison and others first began touting the network computer, traditional "standalone" desktops typically cost well over $1,500. Today prices begin around $400.

But as prices dropped, people seemed to conclude that controlling sophisticated software applications is less important than using the Internet. Home consumers in particular often rely on so-called Web-based applications such as Hotmail, and Internet-based corporate networks are commonplace.

PC makers and electronics companies have accordingly turned their sights toward manufacturing easy-to-use "information appliances" that deliver email and Web access. Thus the network computer, one of the first devices to contemplate doing away with personal hard drives and relying on network storage instead, could be positioned for a comeback--even if questions about the server end of the equation remain.

Ellison, the flamboyant head of the world's leading database maker, said Wall Street's reaction to Liberate [which was earlier called Network Computer] endorses his vision of "thin client" devices such as network computers, telephones and palm-size devices that work with applications from central computers.

"The personal computer is a ridiculous device," Ellison said, arguing that while information appliances won't obviate the need for PCs, the latter have hidden costs, create more labor for corporate information technology departments and don't make sense for many users with scaled-down PC needs.

So, what exactly went wrong with the Network Computer?

Tomorrow: What Went Wrong

Tech Talk | PermaLink

TECH TALK: The Network Computer: What Went Wrong

One of the most detailed analyses of the reasons behind the failure of Ellison’s vision of the network computer (NC) comes from Bhaskar Chakravorti in his book “The Slow Pace of Fast Change.” This is what Bhaskar writes:

For an answer, we must recreate the qualifying conditions for an NC-favorable endgame. Consider four crucial parties:

Buyers: demand-side players who decide to purchase PCs or NCs for the department

Users: demand-side players who work in the department

The NC coalition: supply-side players from Oracle, Sun and IBM

OEMs: supply-side players who manufacture PCs

Choice Factors for Buyers
Being responsible for buying the equipment used by the employees in various departments, the buyers were motivated by the various applications that the computers would facilitate. The other major motivator, as well as constraint, was the budget outlay necessary to meet the department's needs for computing. Buyers would also be motivated by the need to keep systems service calls under control.

Expectations about the choices of buyers in other units or firms with which this department interacts would also play a role. It is important to maintain compatibility to smoothly communicate or exchange information. The expectations about what others are buying also drives expectations about the software and support that would be available from the providers.

In larger corporations, the computing architecture used a client-server model configured with the PC as the client. Any change in the client device would have required upgrading the capabilities of these other components of the system. This would be a distinct constraint to adopting an alternative to the entrenched status quo; the costs in the existing network and servers were already sunk. Change would require the activation of a new buying process for these other parts of the information technology infrastructure.

Choice Factors for Users
Users are, in general, motivated by the desire to do their job without having to relearn how to use a device or get used to new software or interfaces. They would usually prefer the attributes of the PC over those of the NC since they do not bear the direct costs of purchase. The PC gives them the control and flexibility to utilize a vast amount of computing power independently. With a PC, the user can run programs with minimal reliance on connection to a wider network.

Choice Factors for the NC Coalition
The coalition was motivated by the desire to supplant the PC with the NC. However, for each coalition member, the degree to which it would be willing to invest in selling NCs was constrained by several other factors. The NC applications and operating system had not sufficiently matured. There was insufficient market impetus for their development at optimal scale. With the Internet and e-business initiatives emerging as the single biggest attention-grabber for executives at Oracle, IBM and Sun, as well as their most demanding customers, the coalition's marketing and sales resources were feeling constrained.

Choice Factors for the PC OEMs
A critical constraint governing the PC OEMs' choices was the PC industry structure. When the NC was being launched, the PC had become more of a commodity, with relatively low entry barriers into the PC manufacturing and assembly business. Among the so-called tier-one OEMs, there was intense competition for the high-end PCs. A similar pattern existed among lower-end PCs as well, which were continuing to take potential customers away from tier one. This dynamic was reinforced by a highly competitive component-manufacturing industry serving the OEMs.

The combination of easy entry into PC assembly, increased competitiveness, and standardization resulted in a diminished potential for product differentiation across different brands of PCs. Much of the motivation among PC OEMs was becoming focused on taking costs out of the system. The OEMs were being pushed further in this direction by the competitiveness among component makers, by the continued streamlining of production and supply-chain processes, and by the simplification of the distribution model.

Bhaskar summarises: “The NC's primary point of value had been focused on the notion that it was a less-expensive alternative to the PC. The nature of the choice factors driving the highly competitive PC industry had effectively resulted in a closing of the price gap. The PC industry had de facto neutralized the NC's differential value proposition through its own internal competitiveness across PC brands. Buying behaviors were structurally incapable of changing over to the NC in the way it was positioned. The lower-cost-positioned NC was not on course toward its intended endgame.”

Tomorrow: Information Appliances

Tech Talk | PermaLink

TECH TALK: The Network Computer: Information Appliances

One of the reasons for the enduring appeal of network computers is the inherent importance and complexity in today’s computers. We can’t live with computers, and we can’t live without them. We long for something better, something more humane. And the dream endures.

In a book entitled “The Invisible Computer” (published in 1998), Don Norman wrote:

Do you really want to use a computer? Do you want to use a word processor? Of course not. The fact that you think you do is the triumph of marketing and advertising over common sense. Now, maybe if you are a confirmed technology addict, or a computer programmer, sure you love using computers, but not the rest of us. We want to get on with our lives.

I don't want to use a computer. I don't want to do word processing: I want to write a letter, or find out what the weather will be, or pay a bill, or play a game. I don't want to use a computer, I want to accomplish something. I want to do some activity, something meaningful to me. Not "applications," not some bizarre complex computer program that does more than I ever want to know about and yet doesn't really do exactly what I need. I want computing that fits my activities. I want the technology hidden away, out of sight. Like electric motors. Like the computers that control my car.

Computers ought to be like the embedded ones that tell you how far you can drive with the fuel remaining in the fuel tank. Invisible, automatic, and useful. It's invisible, so you don't have to do anything to it. It provides valuable information. Start driving more efficiently, and the remaining distance goes up. Start driving less efficiently and the distance goes down. It wouldn't be difficult to add a time estimate: "Twenty minutes to empty."

This is the way the fuel tank meter ought to be: get rid of the current gauge that tells what fraction of the fuel tank still has fuel in it and replace it with one that says how far or how long we can go. Notice too that this computer is very limited in its functionality: it tells the range of driving with the remaining fuel. Nothing more, nothing less.

This is the way computers ought to be, not just in the car, but in the home, at schools, and in the office. Useful for doing things, for getting answers, for having fun, presenting us with the information we need to know, information we can use directly without further thought. Under this model they will be far easier to use. They will be designed specifically to fit the task, to fit the needs of their users. This also means that they will be specialized, so we are apt to need many of them. No problem, because they will be like all our other appliances: we buy just the ones we want, just the versions that fit our lives. Their simplicity and utility make up for their specialization.

The major problem with today's PC is that of complexity. The complexity of the PC is pretty fundamental: it is built into its foundation. There are three major reasons for the complexity: the attempt to make a single device do too many things; the need to have a single machine suffice for every person in the world; and the business model of the computer industry.

Don Norman’s solution was specialised devices – what he termed as “information appliances”: “Devices that fit the person, that fit the task. Devices that are easy to use, not only because they will be inherently simpler, but because they fit the task so well that to learn the task is to learn the appliance.” While information appliances are not exactly the same as network computers, the thinking is identical – to simplify computing as we know it.

A couple weeks ago, Jim Smith suggested we look at the toaster for inspiration.

Monday: Application-Specific Computing

Tech Talk | PermaLink

TECH TALK: The Network Computer: Application-Specific Computing

The idea of simplifying computing was echoed a couple weeks ago by Jim Smith, a general partner with Mohr Davidow Ventures. In an article on Always-On Network, Jim wrote:

It may come as a surprise to some people that the cost of operating a computing infrastructure now dominates the cost of acquiring it. At the same time, while we've made huge leaps in all areas of computing we struggle to make use of those advancements.

For example, we've proven that we can build 10GHz microprocessors, but we leave a massive portion of those cycles unusable. We can build petabyte-scale storage systems, but absorb a huge chunk of that storage with often inaccessible data. We can connect to one another at 10Gb per second, but don't know what information to exchange.
The key to lowering operating costs, reducing complexity and making better use of technology advancements is application-specific computing.

The low cost of computer components today allows system vendors to target systems to specific business processes. By reducing the general-purpose nature of these systems, configuration demands can be minimized. By reducing the base of software that resides on a machine, there are fewer knobs with which malicious users can abuse infrastructure. By reducing the number of applications that share infrastructure, the likelihood that applications will corrupt one another's data stores is greatly reduced. A byproduct of application-specific systems is the opportunity to more efficiently use computing resources and vastly improve the performance of those systems.

Consider how we use toasters and personal computers. When you unbox that new toaster, you can plug it in, pop some toast in, and push the button. Contrast that with when you unload a new desktop. You set it up in a few hours and constantly update the software. Mind you, the overwhelming majority of that software you will never use. Malicious users floating through the network may use it, but you likely won't.

When the toaster doesn't perform up to expectations, you throw it out and buy a new one. When the computer (loaded with all types and flavors of applications) reaches a breaking point, the specter of transitioning to a new machine is almost always (and rightly so) cause for hesitation.

A crucial aspect of application-specific computing is raising the level of abstraction at which enterprises operate computing machinery. Were a toaster to operate like a general-purpose machine, a user would be forced to specify the temperature at which to toast the bread and the amount current to supply to the heating coils, and time-share that energy with the DVD player. Instead, the single-purpose nature of the machine allows it be a simple choice between light and dark.

While Don Norman comes at the issue of the complexity of computing from the consumer viewpoint, Jim Smith addresses it from the enterprise side. Their views may have come a few years apart, but they underline a common thread – that of making computing less complex, more manageable, and reducing the total cost of ownership. Add to this the growing demand for affordability from the next billion users in the world’s developing countries, the growing use of open-source software in the creation of sites like Google and Yahoo, the rapid proliferation of broadband networks. This is the backdrop in which the resurgence in interest in network computers needs to be evaluated. So, the question that arises is, are the conditions any different now to ensure the rebirth and success of the network computer?

Tomorrow: The World Today

Tech Talk | PermaLink

TECH TALK: The Network Computer: The World Today

As an idea, the network computer idea has been around with us for us for long. Its first manifestation was in the form of dumb terminals connecting to mainframes. It persisted with the use of the minicomputers. Then, came a paradigm shift with desktop computing – all that the users needed was available locally to them. The desktop computers have driven the various cycles of computing since then, resulting in about 700 million users worldwide.

The idea of a computer connected and dependent on a centralised platform came back into vogue with the emergence of the Internet and web browser. Since Sun talked about the network as the computer and Oracle’s Larry Ellison touted the network computer, not much has changed – for the most part, we still continue to buy and use desktop computers.

But there are now a few factors which could finally see the emergence of the network computer.

As a Business Week story put it recently, tech’s future is increasingly going to be dictated by the users in the developing countries. This is where the next billion users of technology are going to come from. For these users, affordability and simplicity are key requirements. The PC, with its dollar-denominated components, remains expensive for the significant majority of potential users in the developing countries. In addition, even after all these years, the desktop computer remains a complex device to master. For all practical purposes, the PC remains a developed market solution – with a limited reach in the top of the pyramid of the developing countries.

The availability of high-speed, always-on communications networks is the second differentiating factor in today’s world. Through a mix of broadband and wireless technologies, connectivity is increasingly becoming ubiquitous. Even in countries like India, data networks are pervasive – Reliance Infocomm has enveloped hundreds of cities and towns with its CDMA data networks offering speeds of up to 144 Kbps. The telcos and cable operators are responding with the promise of broadband via their networks. Mumbai has started seeing ads promising broadband for as little as Rs 240 ($5) per month. Even though what is really on offer is always-on narrowband (128-256 Kbps) with limits on data downloads, this is a good start.

The third factor which is favourable for network computing is the availability of open-source software. The likes of Google, Yahoo, eBay and Amazon are Internet-scale platforms which have demonstrated the use of open-source software to build a massive digital infrastructure for their services.

So, a mix of a shift in markets and consumer needs, combined with the availability of networks and open-source software, is creating the environment for the (re)-birth of the network computer. Technology’s next users in the world’s emerging markets, who so far only had a choice between non-consumption and piracy, will be at the forefront of driving this disruptive innovation, which will eventually make its way into the developed markets also. We are entering a world of service-based computing.

Tomorrow: Service-based Computing

Tech Talk | PermaLink

TECH TALK: The Network Computer: Service-based Computing

At DemoMobile 2004, Chris Shipley discussed her vision for the future of computing. Even though the context of her keynote was the developed markets, much of what she talked about is universally addressable because the underlying infrastructure (communications networks, mobile devices) is being deployed globally. This is what Chris Shipley had to say:

On the surface, [the] idea of device computing makes sense. The locus of our computing and communications activities has moved from desktop PCs to mobile devices. Over forty years, computational power has migrated from large mainframes to mini-computers, to workstations, to PCs, to laptops, to handheld devices.

And with each progression, some pundit stood up at an event like this and declared that the computing paradigm had made the shift to the new platform. Mainframe computing. Mini computing. Desktop computing. Mobile computing.

With that history and context, device computing seemed the logical next step.

But, it’s not.

Quite frankly, while all these devices are amazing in their capability and power, they are virtually nothing on their own. Just plastic, LCD, silicon and solder, buttons and jog dials that deliver very little value if they aren’t connected to some other computer or service. For the most part, mobile devices – whether a cell phone or a PDA, game gear or the custom slate the ups driver carries – these mobile devices are little more than beautifully designed, computationally rich input/output and storage devices.

It’s not until you add a connection – a radio, IR, cable, or docking station – that mobile devices become truly useful and very exciting.

So, to say that we’d entered the age of device computing was to miss the point.

Certainly, we are moving to a new paradigm in computing. Of course, mobile devices play a key role in the new definition. And, it’s easy to see how the devices themselves might get all the attention. Some of you wear cell phones and email devices on your belts like the gunman of the new digital frontier. We agonize over which cell phone to buy, which device is best for remote email. We talk like proud parents about our latest laptop purchase and will demo at the drop of a hat the features and functions of our favorite handheld.

The device is the visible, physical instantiation of what is truly going on here. The device is the end node of a connected system of computing. A system that is larger than mobile devices, or even wireless networks. A system that fundamentally changes the way applications and data are delivered to the point of interaction. A system that profoundly affects the architectures and opportunities of computing from the enterprise to the home.

A system that I call service-based computing.

Service-based computing delivers applications and data from a managed computing platform to a relatively simple end device – the point of interaction with the data. Service-based computing puts the onus of managing the computing environment on the service provider, and liberates the end-user to engage with the information. Service-based computing will drive elegance into application and device design. Service-based computing not only enables, but requires, simplicity and reliability in end-point devices, no matter if they are a cell phone or a desktop PC.

Indeed, service-based computing is bigger than today’s mobile and wireless market. It is broadly encompassing of most enterprise, small business, individual, and convergent consumer computing. Service-based computing is the future model for nearly-all computing and communications.

We are already seeing Chris Shipley’s vision of service-based computing around us – in the likes of the Internet service platforms like Google and Yahoo. As more and more of our data starts residing on servers, the client devices will just become the endpoints to interface with the data. This is the Internet OS that Tim O’Reilly has been talking about.

Tomorrow: The Internet OS

Tech Talk | PermaLink

TECH TALK: The Network Computer: The Internet OS

Tim O’Reilly discussed the idea of the Internet Operating System (built around open-source software):

The first generation of any new technology is typically seen as an extension to the previous generations. And so, through the 1990s, most people experienced the Internet as an extension or add-on to the personal computer. Email and web browsing were powerful add-ons, to be sure, and they gave added impetus to a personal computer industry that was running out of steam.

(Open source advocates can take ironic note of the fact that many of the most important features of Microsoft's new operating system releases since Windows 95 have been designed to emulate Internet functionality originally created by open source developers.)

But now, we're starting to see the shape of a very different future. Napster brought us peer-to-peer file sharing, Seti@home introduced millions of people to the idea of distributed computation, and now web services are starting to make even huge database-backed sites like Amazon or Google appear to act like components of an even larger system. Vendors such as IBM and HP bandy about terms like "computing on demand" and "pervasive computing".

The boundaries between cell phones, wirelessly connected laptops, and even consumer devices like the iPod or TiVO, are all blurring. Each now gets a large part of its value from software that resides elsewhere. Dave Stutz characterizes this as software above the level of a single device.

I like to say that we're entering the stage where we are going to treat the Internet as if it were a single virtual computer. To do that, we'll need to create an Internet operating system.

The large question before us is this: What kind of operating system is it going to be? The lesson of Microsoft is that if you leverage insight into a new paradigm, you will find the secret that will give you control over the industry, the "one ring to rule them all", so to speak. Contender after contender has set out to dethrone Microsoft and take that ring from them, only to fail. But the lesson of open source and the Internet is that we can build an operating system that is designed from the ground up as "small pieces loosely joined", with an architecture that makes it easy for anyone to participate in building the value of the system.

Tim added recently: “I'm talking about the emergence of what I've started to call Web 2.0, the internet as platform. We heard about that idea back in the late 90s, at the height of the browser wars, but that turned out to be a false alarm. But I believe we're now starting the third age of the internet -- the first being the telnet-era command line internet, the second the web -- and the third, well, that tale grows in the telling. It's about the way that open source and the open standards of the web are commoditizing many categories of infrastructure software, driving value instead to the data and business processes layered on top of (or within) that software; it's about the way that web sites like eBay, Amazon, and Google are becoming platforms with rich add-on developer communities; it's about the way that network effects and data, rather than software APIs, are the new tools of customer lock-in; it's about the way that to be successful, software today needs to work above the level of a single device; it's about the way that the Microsofts and Intels of tomorrow are once again going to blindside established players because all the rules of business are changing.”

If there is one candidate to build the Internet OS, it is Google.

Tomorrow: Google OS/PC/Browser

Tech Talk | PermaLink

TECH TALK: The Network Computer: Google OS/PC/Browser

Rich Skrenta wrote about Google’s platform shortly after the launch of its Gmail service in April:

Google is a company that has built a single very large, custom computer. It's running their own cluster operating system. They make their big computer even bigger and faster each month, while lowering the cost of CPU cycles. It's looking more like a general purpose platform than a cluster optimized for a single application.

While competitors are targeting the individual applications Google has deployed, Google is building a massive, general purpose computing platform for web-scale programming.

This computer is running the world's top search engine, a social networking service, a shopping price comparison engine, a new email service, and a local search/yellow pages engine. What will they do next with the world's biggest computer and most advanced operating system?

Jason Kottke added: “[Google’s] real target is Windows. Who needs Windows when anyone can have free unlimited access to the world's fastest computer running the smartest operating system? Mobile devices don't need big, bloated OSes...they'll be perfect platforms for accessing the GooOS. Using Gnome and Linux as a starting point, Google should design an OS for desktop computers that's modified to use the GooOS and sell it right alongside Windows ($200) at CompUSA for $10/apiece (available free online of course). Google Office (Goffice?) will be built in, with all your data stored locally, backed up remotely, and available to whomever it needs to be (SubEthaEdit-style collaboration on Word/Excel/PowerPoint-esque documents is only the beginning). Email, shopping, games, music, news, personal publishing, etc.; all the stuff that people use their computers for, it's all there.”

Headshift’s Lee wrote: “The concept of a Google OS is a more fundamental danger to the Windows cash cow that Microsoft is based upon. A distributed system powered by Google's computing and search power, but which is run through a browser and Web services, could simply render Windows obsolete.”

Of late, there has been a lot of speculation about Google and the possibility that it may be creating its own browser. Google’s goal would be to create its own alternative to counter the integration that Microsoft’s Longhorn would offer between desktop computing and the Internet. Google is in an excellent position given its dominance in search to create alternative computing platforms. Business Week also argued that “Google should be in the browser biz…The move would be key in helping the outfit expand beyond search, tie its many offerings together, and hold off Microsoft.”

Morgan Stanley’s Mary Meeker wrote in a recent report on Google: “Particularly, with the launch of Gmail, we became intrigued at the possibility that Google could create a distributed computing model layered over user-generated content. Right now users can have 1GB of webmail storage—but with potentially tens of thousands of servers, and commensurately cheap storage space, we wonder about the possibility of Google providing a thin application “desktop” that resides on the browser, where users could jot brief notes (GWord?), do basic calculations (GExcel?), and of course, search. The April 2004 registration of gbrowser.com by Google could lend some credibility to this line of thinking. Ultimately, we believe the company could have a significant opportunity ahead of it in Search / Find / Obtain well beyond the Google.com domain.”

Monday: Browser as Network Computer?

Tech Talk | PermaLink

TECH TALK: The Network Computer: Browser as Network Computer?

A comment by Haig in response to one of the earlier columns in this series built on the idea of the browser as the network computer (NC): “I think the NC is alive and will continue to grow in software, not hardware. The NC is basically the browser, a self-contained, thin-client network computer within our computers, giving us what the NC promised without the need to limit our hardware. And like another comment said, Google, Amazon, et al. are our services and they already push their offerings to our NCs, to our browsers.”

So, is the browser the next network computer?

My thinking is that the “browser as network computer” approach may work fine in the developed markets where computers exist everywhere, but it is not good enough for the emerging markets, where the cost of the desktop computer continues to be an inhibitor. Of course, one could argue that the commoditisation in computing will ensure cheaper computers. This is just one part of the solution – what is needed is a $50 computer (excluding display). It is highly unlikely that any existing computer vendor will make such a device because of the fear that such a device could cannibalise from their mainstream business.

There is a second reason as to why the browser as network computer may not be good enough. This is because of the fact that there is a huge library of applications (like Office suites) that already exists – and these are not necessarily browser-based. Over time, they will be rewritten to make them Web-friendly. But for now, they need a rich client interface.

One could argue that the next users do not necessarily need the Office suite – a web-based word processor or spreadsheet would work just fine. This is a plausible argument. But my counterpoint is that it will still take time for the web-based applications to offer the functionality of these applications, and a reasonable time and money investment needs to be made to make them web-based.

There is an alternative – to run these applications on a “virtual desktop” on a network computer. This goes beyond the browser-only approach. The virtual desktop is what applications like Citrix and Microsoft’s Windows Terminal Services make possible. So far, they have been used to reduce the complexity of the desktop client and deliver computing from servers primarily in enterprise environments. What is needed is the same concept to large-scale public computing.

This can be achieved by three key components: a centralised computing platform (think of this as the “grid”), network computers, and software that enables the creation and delivery of “virtual desktops” from the server to the client. The assumption here is that there is always-on connectivity of the order of 128-512 Kbps between the network computer and the grid – in the networks that are now starting to emerge, this is not an unreasonable assumption.

The virtual desktop could encompass the full desktop that we see on today’s computers – both Windows and Linux computers have this as the starting point for user interaction. The browser would be subsumed within the desktop. In addition, multimedia support would need to be made possible by using client-side processing wherever needed; this approach can then support video playback on the client side as also voice-over-IP.

A point to note in this approach is that this does not need the rewriting of any existing applications. The entire library of applications that already exist can be supported on the grid. Over time, as browsers support richer modes of interaction, it is entirely likely that the browser could become the only interface on the network computer. As such, the network computer could even make inroads in developed markets as an adjunct to existing PCs.

Tomorrow: The Four Devices

Tech Talk | PermaLink

TECH TALK: The Network Computer: The Four Devices

So, why will the network computer be the fulcrum around which this new computing platform will be built? We already have four devices – the PC, TV, cellphone and gaming console. Why will it not be one of them? Let us consider each of them.

The two primary issues with the personal computer are affordability and manageability. In the developed markets, complexity and the cost of operation for computers is becoming an issue. In the developing markets, the cost is of utmost importance. Even today, a barebones PC in India costs about Rs 15,000 ($330). While there are ongoing efforts by companies like AMD and VIA to bring this price point down, that will only happen by impacting performance and user experience.

More than affordability, the more serious issue over time is going to be the complexity involved in managing the computer. Most desktop users run pirated copies of Windows, leaving them vulnerable to viruses and spyware. That could be solved with using Linux. But the issue of adding and upgrading applications and managing user data still remains. This is where a “thin client” network computer scores over the “thick desktop.” It brings both affordability and simplicity – by centralising processing and storage, and making software delivered as a service from the grid. The network computer needs no management – just like a telephone. If it stops working, it needs to be replaced, without fear of any loss of data – since the user data is stored centrally.

So, can cellphones or PDAs be the answer? After all, countries like India have more than three times as many cellphone users as computer owners. I think not – for a simple reason. The small input/output footprint of the cellphone and PDA makes it usable for niche applications, not for sustained work. Yes, the screens could become better and there could be keyboards attached – but then we are no longer talking portability.

The cellphone is a personal device. It can complement the network computer, but it cannot replace it. In fact, the cellphone with a capability to attach to an external keyboard, mouse and monitor could double as the network computer – and this is the directions I see network computers evolving in over time, as the incremental cost of adding communications protocols like WiFi or GSM/CDMA becomes close to zero. Until then, the cellphone will not be able to replace the network computer.

The third option is to use the TV connected to a set-top box. The problem in this TV-STB option is two-fold. First, the display resolution on TV does not compare favourably with that of a computer monitor. The TV screen also inherently limits the display to less than that of a 640x480 resolution (which could result in horizontal scroll bars for most web pages). While it is possible that these limitations may be overcome, the second limitation would then kick in. In most Indian homes, the TV is a shared entertainment device – it is still not a “personal” device. As a result, using it for computing would be limited to non-TV watching times only.

It could be argued that given that people have no real options, they may prefer to use the TV-STB combo itself as a network computer. That could be especially true in light of the planned offering by Reliance Infocomm for Indian homes which would bundle a set-top box with a broadband connection. My feeling is that in these scenarios the TV would at best be used for entertainment-related functions. So, while this could be a possibility in homes, it definitely rules out the enterprise market. Given a choice between a TV-STB combo and the network computer, I believe people will increasingly opt for the latter.

The fourth device that is widespread in the developed markets in the context of entertainment is the gaming console. So, could the gaming console be the network computer? I think not. What the gaming console does is provides an alternative to the set-top box – the display still remains as the TV. So, it suffers from all those limitations. In addition, the business model of the gaming console industry is to subsidise the console and make money off the software (games). That model is unlikely to work well in developing countries where software piracy is rampant. As a result, the game console cannot be subsidised and so its inherent price advantage disappears.

That leaves us with having to create a new device, the network computer.

Tomorrow: The Fifth Option

Tech Talk | PermaLink

TECH TALK: The Network Computer: The Fifth Option

The network computer that I am envisioning is a $60-$65 (Rs 3,000) device, excluding the display. In India, a refurbished colour monitor (about 3-4 years old) would cost about Rs 2,000, while a new monitor would cost about Rs 4,000. Thus, the network computer would cost about Rs 5,000-7,000 ($110-150). This is 50-65% lower than the equivalent cost of a personal computer today, and a little more than the cost of a mobile phone.

Let us delve into the network computer a little more and discuss the hardware composition, the software on it, and the connectivity options.

From a hardware standpoint, the network computer needs to use a platform that is commoditised. That provides us with two options – an x86 base or using chips that are used in cellphones. The x86-base would probably create a much more costly solution. What we really need is a processor that costs $5-10, and thus can keep the overall system cost to no more than $50, including packaging. The two important characteristics of the design are the need to support an OS like Linux and be able to drive a standard VGA display. A bonus benefit would be the ability to manage multimedia encoding and decoding in hardware on the client-side – this would allow efficient use of the client-server bandwidth while handling audio and video applications.

The software on the device needs to do two things: provide an OS which can drive the various peripherals (keyboard, mouse, display, network, USB ports, audio in and out), and support a remote display protocol like VNC (virtual network computer). The OS can be Linux.

On the connectivity front, it will be necessary for the network computer to support a wide range of options, though not necessarily on the same device. After all, without connectivity to the network, the device would be useless. The various networking options would be Ethernet (for LANs), Wi-Fi (so as to eliminate the need for cabling) and perhaps, GSM and CDMA. The wireless options could be supported via an onboard software radio, which could dynamically use the most appropriate connectivity option.

In addition, the network computer will need power. This can be provided for via the mains. Some versions of the network computers could also come with battery support – these versions are more likely to resemble cellphones and come with integrated keyboard-display modules so as to create an integrated unit.

Technologically, the network computer is not a very radical device. It doesn't need to do, and should not be. It should essentially provide all that a desktop computer provides, except that storage and processing are not done on the device. It should be possible to build such a device for about $50-60.

The natural question: how does one make money selling the device? The short answer: one doesn't.

Tomorrow: Business Model

Tech Talk | PermaLink

TECH TALK: The Network Computer: Business Model

The network computer by itself will not make money for its makers. It needs to be part of a wider ecosystem – where computing is proffered as a service. From the network computer manufacturer view point, there are two options: to work with a small 10-15% margin, or think of the device itself as a service and make small amounts of money on a monthly basis, just like a utility company does.

I believe the ideal solution lies in offering the network computer for a refundable deposit which covers the cost of the unit, and then charge a small monthly fee. From a customer viewpoint, the network computer is not likely to be purchased as a standalone unit – rather it will be used with service (be it at home or at work). Thus, the network computer company of tomorrow is more likely to resemble a utility company than a computer maker. By making computing a utility, users are also being provided with the flexibility of cancelling service anytime – something that is not an option when computers are bought on installments. This is possible because the network computer is a device that really does not age and become obsolete – other than hardware failure, one has no worries. In other words, the lifetime of a network computer could be as much as double that of a regular computer (which typically has a 3-4 year lifetime).

The magical monthly fee for the computing service should be no more than Rs 700 ($15). Consider the case of the home user. In this situation, Rs 200 would go for the network computer, Rs 350 would go for bandwidth, Rs 100 would go to the grid for the computing and storage facilities along with a wide variety of software and content, and Rs 50 would be available for providing support. In the case of the enterprise user, the bandwidth costs would be lower since they would be amortised over a larger number of users. But an additional cost of a local server may be needed in the event that a two-tier grid (think of this as a “grid cache” and “grid core”) is deployed. In this case too, the per user cost would be in ballpark of Rs 700.

Will this model work? I believe it will. A no-commitment Rs 700 per month offer is about 50% higher than what users pay for cellphones in most cities in India. What is on offer is a computer that looks and feels like a regular desktop computer but without some of the hassles associated with it. As a critical mass of these network computers gets deployed, software and content providers will be attracted to this platform since now they have a much more cost-effective way to deliver their offerings to users. This will in turn create a positive feedback loop for adoption.

All previous efforts at network computing have ended in failure. But I firmly believe that the techno-business developments will create the necessary atmosphere for the network computer succeeding this time around. The question that arises is: which of today's companies will make it happen? Or, will it be a new set of companies?

Tomorrow: Making It Happen

Tech Talk | PermaLink

TECH TALK: The Network Computer: Making It Happen

My belief is that the likes of Intel, Microsoft, Sun, Oracle and Google are not the ones who will build and capitalise on this new platform. The company or the set of companies to build the network computer and the grid will emerge from the developing countries.

For the likes of Intel and Microsoft, the network computer is a disruptive innovation. Their existing customers in the developed markets are not asking for it. These companies, even though they have the greatest resources at their disposal, are unlikely to want to upset their gravy train of billions of dollars in profits from today's users of computers – each of whom is well ensconced in an upgrade cycle that delivers new hardware and software every few years.

Intel doesn't really have an emerging market strategy. It offers its dollar-denominated chips to these markets with the belief that as the economies grow, adoption will increase. And so it has been happening. Growth is 30-40% in countries like India and China. The irony is that it could be many times that if the network computer were integrated with the rest of the ecosystem (grid, broadband, software and content).

Microsoft's emerging market strategy seems to be low-cost, limited functionality, local language versions of Windows XP (called Starter Editions). This is a flawed approach, as I wrote recently on my weblog: “Microsoft is caught between piracy, non-consumption and Linux in the developing markets. Rather than low-cost, reduced functionality Windows, it should look at reducing cost of the desktop computers (think thin clients) and running Windows off centralised platforms, with a pricing of $1 per month. Not just the limited versions, but the full-featured versions. But this requires Microsoft to think not like a monopoly but like a utility company.” [Also see some of the comments from readers as part of this post.]

Google is the other option. It has the reach and the cash to bring such a strategy to market. But it will not focus on devices. It will continue to focus on software, and will probably opt for the “browser as network computer” approach. It will build out services on its computing platform and work on delivering them to desktops, cellphones and other devices via a Google-branded browser. Google is after all a media company, and is focused on increasing the space that it has to serve its ads. [One could argue that controlling the virtual desktop on the network computer would be the ultimate prize of them all.] What Google lacks, like Intel and Microsoft, is an understanding of emerging markets.

The opportunity then is for a new network of companies, each focused on key elements of the emerging computing ecosystem. This company has to be born in the emerging markets because these will be the first markets. Local context is important, and it is hard to get this when management is sitting a few timezones away. This is the opportunity for entrepreneurs in India. So far, we have been offering services on computers to global companies. The time has now come to offer computing as a service to our own brethren – and their ilk across the developing countries, the middle and bottom of the global pyramid. Can we do it?

Tech Talk | PermaLink