The future of our hobby (was Re: ImageDisk project is canceled)

It's taken me several days to mull on this and the inevitable progression of computers. Anyway, the subject line only comes up rather deep into this reply to Tony about his "I don't replace boards" stance: It was thus said that the Great Tony Duell once stated: The first computers were designed and built out of discreet components, which gave away to integrated circuits. Not only did it become cheaper to build computers, but easier as well, but I'm sure there were guys out there who ridiculed those that couldn't hack a computer with discreet compoments. "What? You have to replace an *entire* chip because a single transitor inside blew?" they might say. "How can you detect which chip is bad? With components, I can place probes *anywhere*, trace the fault down to the precise transistor and relatively quickly replace it. See?" But over time, the nature of a "component" changed yet again, this time to a larger unit---the card (or motherboard). The functional units are now larger. Whereas there are multiple companies that make a quad two-input NAND 7400, there are also multiple companies that produce (or rather, produced) two port serial cards for the PC. Sure, it's may be just as easy to replace a chip as a card, but then again, it should be easy to replace a blown gate, right? The 7400 chip is documented to work under certain conditions, with these inputs and output. Just as a two port serial card for the PC is documented to work with the ports at such-n-such address with so-n-so IRQ and what not (or else, not many people will buy the thing). Discreet components -> IC -> boards -> computers "Here" being your house? Or England? Because it's certainly a reality here in the States, and I can name one company that does this as a standard operating proceedure---Google [1] (heck, even the company I work for does that to some degree). With 60,000 machines [2] it's probably more cost effective to let the machines die and work around them than to attempt to fix them (they probably do pull the dead machines, but only if there are enough to justify the expense of doing so). Even more amazing, Google is going further---not only does it consider a computer a mere component to be replaced, they're trying to make *data centers* a component! [3] A 40'x8'x8' data center, ready to be shipped anywhere in the world, the ultimate in plug-n-play. And thus we come to the subject line. In 10 years time (or heck, say 20) when Google starts to decomission some of these "data center containers" for pennies (maybe even mills [4]) on the dollar, what are we to make of such a system? It's about the size of the early computers, but with *way* more functionality and equipment. And its components are today's commodity PCs running a commodity operating system. Maybe we should start saving those Wintel boxes after all ... -spc (The technical specs on Google are very interesting ... ) [1] Ever since the early 90s I've felt that there were no more systems for a hacker (in the good, exploratory sense) to get into. TOPS-10, TOPS-20, ITS, heck, even VMS, were all dead or seriously on the decline, with the only multiuser systems left running Unix (how boring). It wasn't until last year I realized that there *is* a system that I would love to get into and play around with---Google. Sure, it's built on top of Linux, but only because it's easier that way. [2] http://www.tnl.net/blog/entry/How_many_Google_machines [3] http://www.pbs.org/cringely/pulpit/pulpit20051117.html [4] 1/10 of a cent.