Slashdot Mirror
Sun President Says PCs Are Relics
christchurch map writes "Jonathan Schwartz, president of server and software maker Sun Microsystems, said that the personal computer is increasingly becoming a relic. Instead, what has become important are Web services on the Internet and the majority of the world will first experience the Internet through their mobile phones." From the article: "Schwartz points to the increasing wealth and power of companies, like eBay, Google, Yahoo and Amazon.com, that profit from free services available over the network. Among his audience, many more people said they'd rather have access to Internet services than their desktop computing applications. And Microsoft--the company with the biggest financial stake in the PC software business--has struggled to cope with the arrival of Web services."
Yeah, right. Like a 128x92 screen is as useable as a 1600x1200 one.
"National Security is the chief cause of national insecurity." - Celine's First Law
Its not exactly surprising that these words are coming from Sun, seeing as their motto has been "The network is the computer" for at least the last 10 years now. The web as an application platform has been making notable steps forward, but there are always going to be large enough differences in browser platforms so as to cause problems in non-homogenous environments. The web browser is increasingly becoming a new 'operating system', and as with our existing operating systems, it has all the differing configurations and incompatibilities between versions that we've come to expect from any such platform. Moving from one of these environments to the other has made sense for simple data-based applications for a long time now and we're increasingly seeing interactive applications move forward with AJAX/Flash based approaches, but its not a total replacement for the native desktop applications at the moment. Thats not even to mention the vast variations in bandwidth availability across the world, and the limitations that can place on development.
Business Voyeur
Honestly, this is the simply wishful thinking (bordering on delusion) under the guise of expert analysis.
This is even less true now than it was ten years ago.
A better question will be who will buy Sun.. I'm guessing Dell.
Personally, I think there's a middle ground here. Basically, I'd like sort of a "home mainframe", and a bunch of terminals around the rest of the house. I've got maybe 5 computers in my home, and like you said, they're all 99% idle most of the time. If I could condense all of that down into one box, it'd be great. I'd hopefully be able to access the same desktop from any room(terminal) in the house, when I decide to replace/upgrade hardware, I only have to do it once, and I only have one computer to administer. But most importantly, all my personal data and files are still somewhere that I physically control. Such a system would need to be a little different than today's PC's, but it wouldn't require the complexity or performance of corporate mainframes or anything like that.
I guess you could run into the problem of more than one terminal doing really intensive stuff at the same time, but maybe since I'm only buying one box, I can spend a little extra and put some nice hardware inside to mitigate that problem. As it is, only one of the five machines that I have now is anywhere near state-of-the-art, so it wouldn't be that much of a difference anyways.
One time I threw a brick at a duck.
If you simply use your PC to "do e-mail and the internet" then yes, I agree that the PC is rather ill-suited to the task. There's a vast amount of wasted capacity if you're only running an internet browser on your PC.
However, the PC is also a platform for a variety of other things:
For the sake of redundancy, I'll mention that the PC-less world relies much more heavily on bandwidth than the market currently provides at reasonable cost. PCs are primarily a storage device, and until you get another system with adequate cache to store all of the things that you want to keep after you download, you'll probably be stuck using a PC.
If you're an avid gamer, then you're definately putting a much larger portion of your PC to work than the "average" user described in the article. It does seem that consoles are becoming much more powerful in terms of delivering games than PCs are, but they are much less flexible at this point and don't support user-modded games, maps, addons, etc.
If you're a media fan, then the PC offers you speed, reliability, and flexibility that the internet world does not. Granted, you can get your music online, but I'm sure we all sleep much better at night when we know our favorite music is on our PC and not going anywhere, rather than being subjected to the whim of our ISP or whatever site we stream from.
The internet is a growing market for just about everything. Unfortunately, it also means that greedy people are starting to catch on, and there will be more and more pricetags for online services in the years to come. It doesn't cost me anything (aside from the electric bill of course) to play a song that's on my hard disk, but the internet is not so friendly (and I expect that it will become less-so as time goes by).
Streaming videos just don't rival the quality of a DVD at this stage. If you were able to compress a stream and still maintain quality at a reasonable rate, you'd still need a processor on the end-user side to decode the stream. There's also the issue of bandwidth and transportability of media. I can take a DVD with me to the room downstairs or even out of state on a plane and it never loses quality because the signal gets bad or my connection changes.
While the news, e-mail, forums, information, etc. may becoming increasingly internet-specific in terms of its execution, there's still a great deal of use for a PC. I'm certainly not going to give up my hard drives any time soon (xbox 360 can go to hell).
So what's the motivation for all of the internet stuffs, from an industry perspective? What you do online, they can see. What you do on your PC, they can't. Unless installing spyware becomes the new fad soon, that's not going to change. It makes much more sense from a business perspective to have all of your applications in the same place you have your data-collection--online.
Until the internet gets a Ctrl-S, I don't think I'll be giving up my PC. I can't count the times I've lost a lengthy post to the evil internet. And I like being able to keep my media out of the clutches of some greedy CEO as well.
A lot has changed in that time. A couple of years ago, we built an ASP-based service. At the time, we were really worried about high-speed bandwidth adoption. To our surprise, this has not been a problem at all. We seldom ever get calls from people with dial-up service.
The biggest problem that we face is one of perception. People believe that if they buy the software and install it on their PC that somehow they'll have a better experience. They forget that as soon as their PC is full of viruses that the program stops working. They forget that they have to backup their data on a regular basis. They forget that their data is locked in a single place rather than being accessible from any web browser in the world. I think that people are reluctant to build dependencies on others, but that will change over time. After all, when was the last time somebody stapled their own wires to a phone pole because they didn't trust the telephone network?
If you don't want crime to pay, let the government run it.
Just like retail is facing a death march, so is the PC, the TV, the phone, the iPod, the DVD player, the cable box, the newspaper, and so much more.
Convergence is not coming, its here. Its only going to get "worse."
Wireless broadband everywhere is just around the corner. Why store data on a PC or a LAN at all? Constant repair/upgrade/update/crash concerns. When 2Mbps wireless is truly a commodity, change will be imminent.
What data do YOU store? How about the average household? MP3? Movies on DVD? Thesis? Magazines in a bin for the past 3 years? Family photo albums? No, they won't disappear, not immediately.
Once that 2Mbps wireless is that commodity, data warehouses will be, too. No more backup concerns, no hardware-go-booms, no constant PC replacements. Just rent the space as you need it. Need more power? Its there.
Software rental (client-server thin networks) will be the next step. It will happen. No patching, no $250/year license for Ofiice 2006, no virus concerns, just pay-as-you-go. IT consultants beware.
The new TVs are just 1024x768 plasmas or LCDs. A $50 set-top box transcieves to Internet2. Your PDA will have the same access to your data as your home dumb terminal and office dumb terminal. All your contacts, movies, songs, personal and business data.
Why even buy music or movies? Pay-per-play!
Privacy? Few care. DRM? They're working on it for this future, not for piracy today.
Dont get me wrong, I'm a massive fan of "the network is the computer" and all that jib-jab. But if web services is the great extent of it, count me out. Web services is fine for checking your email, but theres a world of real work which needs to be done at a near-OS level to create a distributed computing environment. Plan9, IBM's SoulPad, Synergy, these are the few and the brave willing to go out and fsck around with the traditional concept of a computer, to unweave the ideas of one computer, one monitor, one mouse, one system. To reduce network is the computer to WS-* is just a wretchingly awful idea.
The human-computer-I/O needs to be made network capable. I'll get back to you on it.
Myren
You're right, but I should clarify: Right now network equipment doesn't yield much bang for the buck. You have to invest in some pretty expensive network hardware to do things like clustering over long distances.
On the other hand, the bang you get for every $1 of CPU or disk today will take you pretty far. $1000 of CPU and disk will get you a pretty powerful box, but $1000 of fiber optics will just get you an 8-port switch.
However I see that changing as we reach the limit of silicon-based chips. We're already at the point that we can't just pack more transisters on a die. Now we have to do things like multi-core chips and hyperthreading, which unlike pipelining or branch prediction requires pretty major changes to the way we write applications. We're running out of CPU tricks to give applications a free speed boost.
In the meantime network prices are plummeting. Gigabit is standard on home machines; fiber to the curb is a reality in some places. Eventually $1000 will get you a fiber network fabric that can move data at local bus speeds; can you imagine writing applications that could harness that kind of raw decentralized power?
Clearly, you are right. Bandwidth will always be the limiting factor. Clearly, it is impossible to get HDTV level graphics through the same sort of connection that I use for my cable internet. A true revolution would be required in Internet connectivity.
Now, that sarcastic comment aside, I agree that latency will always be annoying for the vision of the network computer. The speed of light will always kick you in the ass. (And it will kick you in the ass as fast as possible.)
...is that if you want to make money, it is useless to target the PC. The PC is dead as a target when it comes to commercial application development.
He isn't trying to replace your PC, he's trying to explain why companies just aren't developing PC software anymore.
All the revenue-generating applications these days are on the Internet. (Games are one of the big exceptions, but even PC games these days have to use the Internet in some way to be commercially viable.)
Paul Graham has been saying the same thing for some time. And I think they’re right!
Sunlit World Scheme. Weird and different.
Try this. Just needs a hard drive and an OS, and woila, a $300 PC that comes fairly close to those specs. 'Course, it needs some assembly, but for $300, what do you expect?
/. zen: Imagine a Beowulf cluster of Beowulf clusters...
This is one reason why .Mac sucks: Why would I want to store my personal stuff with them? And if I were to store it there, I wouldn't want a measly gigabyte for that price.
Very insightful post. I absolutely agree. I recently setup my own server for music and, eventually, video and e-mail. Now when I want to run an application that I can't install on my work server, or I'm stuck somewhere with Net access, but nothing to do... I contact my server, install the software, or stream my digital media, and I have access to all of the data I want on my terms. When I'm traveling, I can take digital photos and send them home.
Stepping back in time, in college I used Linux almost exclusively one year. I setup the system to do Appletalk networking and Samba networking and was able to access my compter's drives from anywhere on campus. I walked around without any floppy disks (which could break and loose data) and could log on to any free computer in the cluster and do my work.
Once you get used to having your own networked server, nothing else will do. One barrier I see right now is that server administration is probably too complicated for the average computer user. Securing the machine needs to be easier, back ups need to be easier, and setting up secure remote access needs to be easier. The other barrier is ISP's treating broadband as 'smart TV'. Capping uploads at a slower rate than downloads and banning the use of server software severely limits the utility of a broadband connection. Our salvation so far has been MMORPG's and X-Box live. Games need fast connections in both directions and run server processes to support multi-player gaming.
There is a balance that can be struck here. Right now, big companies think they have to control the hardware and the content. But there are a lot more computer users than there are hosting and content companies. What if Sun sold an easy to administrate home server? The entertainment companies could sell you a license to serve content from your home server to any of the devices that you own. Any display device would just be a dumb terminal to the content. Consumers would be happy; they can access their content on any device they wanted, when they wanted.
This scheme could run up against the DRM debate... but what if you owned a portable networked media player? It has a key to access your home server and the content you purchased. If you want to go to a friend's house and listen to music or watch a movie you own, just take your access device and hook it to your friend's TV. Now you can do what we've always done; loan content to friends on a limited basis or share the experience of content with a group of people, but when you go home, you take your movies and music back with you. Heck, the server could have a loan-out feature. Issue a temporary key to your friend to access your content; there will be no more scratched CD's or DVD's that you will never see again. And content companies would really be dealing with a market they are used to. They still made money when I could bring CD's to a friend's house and listen.
Whoever figures out how to make 'information furnaces' easy to use, standardized, and cheap, is going to eat everyone's lunch. Tivo showed that people are ready for it; we just have to convince the content producers that they can make money without restricting their users.
======
In X-Windows the client serves YOU!
This sounds mysteriously like predictions from 10 years ago, and 10 years before that, and 10 years before that. I don't think Sun was the one to make that prediction the first time, but they sure were making it 10 years ago. So far it just seems to keep getting less and less true. The network is not the computer. The network is the input. The computer is the computer.
There are quite a few problems with the remote-PC option. For one, latency is a killer which we can only overcome by client-side predictions, so most UI will be intollerably unresponsive without enough power to run things locally.
For another, just because the computer is physically remote doesn't mean the user doesn't have to administer it. It's still their 'GoToMy' PC. They can still screw it up, unless you're not going to let them install applications, at which point it becomes a bit useless as a computer. If users want autoupdating, why not just write software that autoupdates?
Third, we all know that network black boxes in this country come as tied to specific services. And we know that technology dongles like this fail.
Fourth, while some network apps have taken off, like webmail, others have failed miserably. Browser-based text editors come to mind. Some things you just want local.
And Fifth, with computers so cheap, why network? Where is the huge performance or convienience increase that would convince everyone to switch?
Latency basically kills the possibility of playing games over a black box even with high-speed broadband. You would need to do the kind of expensive client-side predictions currently in use to keep the game playable, at which point you would by definition have a client capable of playing the game.
But ultimately I think the basic problem is that people want to own their things. They don't usually want to lease their telephones, or rent their software by the year. When I buy a computer, I want that feeling of "well, i've got that computer problem solved." I want my private data on a local disk. I want to be able to kick something. I just don't see the compelling argument that would alter computing from the current independent model to a client-server model.
The ______ Agenda
You get the best of both worlds; ability to install your own apps and no need to physically maintain a machine.
The system administration could be drastically simplified for the common case, and security issues could be patched by an automated updater, similar to Debian apt-get.
The problem is that ISP's don't want this model; they want to lock people into keeping their data in proprietary systems.