Luke posted a response to Paul Graham’s article and Don Dodge’s response. This got me thinking about my interview with Microsoft last semester. While I was there, I was able to talk with a manager with the WinForms API team. We got onto the subject of where the web is going, and what future computing will look like.
One of the ideas that had caught my attention at that time was that desktop machines are going to migrate towards thin clients. All you need is a monitor, keyboard, mouse and internet connection, and you were ready to go! The internet will be so fast, you won’t need a CPU to do processing, a video card to do your graphics, you will purchase CPU cycles and hard drive space from your local ISP; just as you do bandwidth. The price of computers would be drastically lower, and you would pay on a per-subscription basis instead of a big upfront cost. Upgrading your computer would be as simple as upgrading your subscription.
My interviewer asked quite a few good questions about this, such as how much cheaper would the PC’s really be? One of the things that was pointed out was the cost of the screen’s. Almost 1/2 of your computer cost on a budget PC will be put into that LCD screen. And want a bigger screen? You may end up paying more for the screen than for the computer itself. Therefore the idea of a “budget PC”, wouldn’t become so budget after all.
Also, this architecture wouldn’t be the greatest for people that want to do things that are video-intensive, such as playing video games. Currently, PCI-E 32x are at 8000MB/s. Will the internet be able to allow such performance over Ethernet? I highly doubt it with the current infrastructure, it would have to have a major overhauled.
I am not saying that the thin client idea isn’t possible, I am just saying that the internet would have to be awfully fast to support such architectures. I don’t see it being able to support such an architecture for more than a decade, and with recent news posts about how the internet may not be able to handle future loads, it makes you wonder what will come of a future Internet.
2 Responses to “The future of the web”
Another reason, to support your article, that computers wont go to thin clients is because, no matter how fast the internet gets, local hardware will always be faster. I think in the late ’80s people thought thin clients were the wave of the future as well. That’s why Purdue has all those old sunblade thin clients. Also notice we don’t use them anymore. They are completely unusable. Also, look in the new CS computer labs. The new labs have the most expensive workstations you can get.
The reason for this: the internet is scaled in proportion to hardware. And like I said before, hardware will always be faster locally and software is going to continue to require more resources. Some apps will be migrated to the internet (like office applications are right now) but, they are the oldest apps we have! The office apps were some of the first apps made for Windows so it only makes sense that they would be the first to go. It’s hard to see what the big new apps are now, but, the apps moving to the internet are being replaced.
I realize you said some of this in your post, I just wanted to put a different spin on it.
Plus who wants to hack on a thin client? It’s just another way to take the fun out of computing… LAME