The General Approach to Computing - A Ramble

Warren Wolfe wrote: I tend to agree with this sentiment, I do love having old machines around, but I use the most modern machine I own/can access for daily work. It doesn't mean I don't appreciate the Lisas, Commodores, or old Macs in my collection, nor does it mean I don't use them often. That said, some machines I rarely turn on because I don't want to risk harming them, and in those cases, I find using emulators for them a better choice. i.e. the 5M profile hard drives on my Lisa, or the Xerox Star's floppy and hard drives - after all, the most common components to fail are mechanical ones. But if you're thinking in terms of common things to do, perhaps those don't need a lot of CPU power (i.e. word processing email, and surfing the web - though recently between heavy javascript and flash sites, you do need the CPU power.) If on the other hand, you do something a bit more interesting, like development, and find yourself compiling the same code dozens of times a day, it pays to have a faster machine - that doesn't mean you have to run the evil OS from Redmond, nor does it mean it has to have a CPU from intel, but you can consider in terms of efficiency. MIPS (or perhaps GIPS these days) per watt. While both my Xerox Star and my Macbook can be used to write something in a word processor, and both have a very nice desktop, my Macbook doesn't require its own 20AMP circuit. And if a hard drive dies, a 2.5" SATA drive is fairly cheap and easy to replace. How about the drive in the Star? And should I need to print that document, where do I find a printer for the Star? Doesn't mean I love the Star any less. This is very true, and is one of the costs of modern electronics. The other part is while the high end items are expensive at the start (say before the warranty starts to run out), they can be had for less than half the price in working order a year or two out, once they're used. And perhaps less if they're sold as non working. So it's not too bad in all cases. Where you're able to this, by all means, that's wonderful. And as you note, not everything is easily serviceable this way. It's just the reality of things. When each transistor is microscopic, how would you fix just that one? And when it's part of an SMT chip with hundreds of pins, how would you replace it when it was surface-mount soldered onto a board? Is any electronics factory going to care that we'd want them socketed if it'll cost them $1-$10 more per chip, or are they going to care that the consumer and maker wants them as cheap as possible and the maker wants them to fail the day after the warranty expires? (And the maker wants them to be as proprietary as possible so they can control their market?) Ya, well, you know what they say about common sense. :-) A lot of this boils down to how practical something is. Collecting and repairing historically important computers isn't the same as reviving a somewhat modern but obsolete, by a couple of years, computer. I'm going to care far less about throwing out a 20G IDE drive instead of attempting to repair it (even if I could), versus an old 5MB 5.25" full height MFM drive (which I might be able to.) Perhaps in 10 years the 20G IDE drives will be scarce and have some sort of collectible value to them, but that doesn't mean I'm going to stockpile the dead ones in hopes for that to happen. Sure, someone out there today will see an opportunity, and might stockpile them now when they can be had for cheap. You can be certain that in 10 years we'll be complaining about why they charge $100 for them on ebay and how ebay sucks so much and how these drives shouldn't cost so much when for $100 you could have a 200TB drive... All while secretly kicking ourselves for not keeping more of those around. :-D But such is the life of a collector. Not at all. You be certain a lot of this stuff is grossly inefficient. The current "in" trend is to use things like garbage collection instead of doing your own memory management, and it's also to use dynamic languages which are really easy and fun to work in, but are extremely difficult to compile (i.e. ruby - and I'm not knocking them by saying this). So if you think of a popular application that's likely to be used by hundreds of thousands of people, the trade off is that it costs the programmer less time to spit out a piece of code, but what they don't look at is how hundreds of thousands of people now have an application that runs several orders of magnitude slower than it could, and requires several dozens of times more memory. Why? Because the industry loves having replaceable programmers, kids who have passed some introductory course in Java and churn out whatever the spec requires. This isn't to say that there aren't really good programmers out there anymore than can code low to the machine, (you'll find those writing device drivers for example), and perhaps those will get paid maybe 2x-3x more than the replaceable Java kids that corporations love so much, but IMHO, what they produce is worth several thousand times more, yet, they'll never get paid several thousand times more. That said, Java, and most other languages have a plethora of libraries available to them, so they don't have to reinvent the wheel (quicksort, linked lists, dynamic arrays, etc.) so for the EXPERIENCED programmer, it can be a highly efficient and powerful tool. But that's the trouble, most places are set up with very few experienced programmers and hordes of code monkeys. So that's how you wind up with say things like Office Suites that take up a whole DVD and have billions of unused features full of bugs that few people need. As long as the rest of the corporate world continues to use the bloated inefficient tools, they remain the standard and everyone wants a document in that format. It's a self-sustaining ecosystem. So far no one's complained that I've used OpenOffice to edit spreadsheets (even at work) or text documents as long as I export them in that non-native format they seem to think is the best thing since sliced bread. Infact, they never even notice it unless I happen to trip on some bug, and even then do they rarely care. IMHO: if you remove all the ugly inconsistencies in the Java language, fix up the syntax a bit more, get rid of the garbage collector and have a really good optimizing recompiler that spits out native code, you might be able to write very efficient code that would run very nicely on all platforms. Just my $0.02. "The future is already here. It's just not evenly distributed" -- William Gibson. Somehow he managed to say all that in 11 words instead of two paragraphs. :-) Damn, I wish there were more writers like him out there. A lot of this is that the tools are kept hidden. The dealers have access the stuff the public doesn't, and even non-dealer mechs. This is on purpose to lock people into their market, and perhaps its expensive, not just because of the lock-in but perhaps because it'll incite the car owner to trade in their old car for a newer one, and thus feed the system more money. The fix would be to create an open source car manufacturer, where everything is accessible and well documented. But good luck with that. Still the latest Ubuntu was out just last week, and I gotta say, it's quite impressive. Maybe there's hope for this idea too. Maybe Mr. Shuttleworth can buy a bankrupt car factory and start it up. :-D But you need to remember that Microsoft is what drives Intel and other hardware manufacturers to improve so quickly. By producing bloatware, they generate a market for Intel to create and sell faster chips. If Microsoft goes away, Moore's law will slow down significantly. (Perhaps something else will replace MSFT's role in this, but the idea stands.) As for smaller sets of parts that do lots of things, sure, that's wonderful for the maker, but would they sell them that way? However, this depression we're in now (and yes, that's what it really is) has perhaps woken MSFT up that they can optimize a bit - or so people who I know that have played with Windows 7 have told me. (Disclosure: I've not played with either Vista nor Windows 7, so this is 2nd hand info.). We're also seeing a lot of interest (but not surprisingly, few sales) in "Net Books". (Let's see do I plunk down $300 to buy a tiny tiny notebook computer that can't do very much other than surf the web and has little storage, or do I buy a used/refurb full sized notebook for the same price that I can do everything I'd need?) Now that cell phones have killed off PDA's, perhaps the netbooks are for people who find cell phones not as useful, and wish for something bigger, but don't really want a full sized notebook? Even so, I see something like 30 posts a day on NewtonTalk, so the PDA might be dead in the eyes of industry, but not the need for one.