I did this for my NAS, but it was more expensive than an HP microserver with a similar form factor. The only reason that I did it was that I wanted to be able to use the machine for XBMC so I wanted a slightly better GPU. The only bit that I'm likely to upgrade is the disks, and even then I had to make some compromises (the case has 4 removable disk bays and a slimline optical drive bay, but I couldn't find a motherboard that had everything I wanted and more than 4 SATA slots, so I can't use one of the disk bays).

I used to think desktop computers were upgradable, but it's not really true. Sure, you can bump the RAM and the disk easily, but by the time a new CPU is worth the bother, the socket and chipset have changed, so you need to buy a new motherboard. The new motherboard takes a different kind of RAM. The hard disk might still work if you're lucky (although you may find that the interface type has changed) but it's probably going to be the bottleneck in the new system so you probably want to upgrade it too.

The last time I upgraded a desktop, I kept the case and optical drive (which I replaced a bit later). I kept the hard disk, but added a second one and eventually stopped using the smaller one. After the next upgrade, I had enough parts to build a completely new desktop. If two upgrade cycles means that you've replaced every single part, then it's simpler and easier to just lengthen the upgrade cycles a bit and by a completely new system.

I guess you missed the huge amount of regulation that has come in regarding pollution in waterways in the last 50 or so years then? Or do you think that this prediction would still have been wrong if factories had been allowed to keep dumping waste into rivers? In fact, maybe you should just try visiting some of the parts of India and China where they've managed to build an industrial base without such regulation and see how the water tastes. The entire point of making such predictions is so that we can avoid them happening.

Arguing the facts doesn't appear to work. Read the posts above yours. A number of them are full of assertions with no citations backing them, followed by responses citing data showing that they're wrong. In a world full of rational people wanting to have an informed debate, that would be the end. Now go back to the last story about climate change on Slashdot. You'll see the same assertions being made, by the same people, and being contradicted then too. At some point, you have to just accept that either these people have some vested interest in denying the evidence and so can't be convinced by more evidence.

It's more common in the sciences, where the convention is to provide a LaTeX stylesheet. In subjects where manuscripts are submitted in Word format, the typesetting by the authors tends to be so bad that you need to have a professional redo it.

Moore's law applies. The reason the Mac was so much cheaper than the Alto was that it was a decade later. The Alto was also heavily designed for experimentation. Programs were compiled to a bytecode with the bytecode interpreter implemented in CPU microcode. This made it very easy to change the instruction set and find one that was well suited to the requirements of the software, but for a commercial product you'd have wanted to sink a lot of that logic into the hardware.

Let me guess. You were reading at +2 back then?

No, this is what Slashdot has always been. I started regularly reading Slashdot around 2000, and back then there were posts just like yours decrying the state of Slashdot today, pining for some golden age. And yet, looking at the archives, not much had changed.

And neither mentions the CPU architecture, but if you go to the product brochure then you learn that they're Intel Xeon E5s (which doesn't narrow it down much). Interesting that they're using E5s and not E7s, but perhaps most of the compute is supposed to be done on the (unnamed, vaguely referenced) accelerators.

There are also costs associated with taking cash. Having to store large amounts of cash, having to audit tills more often, having to transport it to the bank, increased security needed as a result of being a much more attractive burglary target, and so on. For large stores, these tend to be more than the cost of accepting cards (next time you're in a supermarket, imagine if every transaction was cash. Think about how much they'd have on the premises by the end of the working day.

The difference is that the costs of accepting cash don't scale linearly, whereas credit card fees do. For a small shop with a low turnover, cash is often a better deal, but for a large shop it isn't. I came across a paper a few years ago that compared the two and was quite surprised by how much handling cash costs even small businesses.

They don't hate you, but you're probably not their favourite customer. The most profitable people are the ones that have high income and poor impulse control, who will buy an expensive thing periodically and then take a few months (at 10+% interest) to pay it off.

The people who put through a lot of purchases and pay promptly are the next best - they're charging the merchant 1-3% of the total purchase price to lend you the money for a month, which is a pretty good interest rate for the lender. I put a load of work expenses on my card and so last time I had an issue with a fee that I disputed, my card company immediately and without quibble cancelled the fee and added a good-will payment to my card, because the fee was about 5% of the profit that they make on me in a year.

For some card companies, the people who massively overspend are a good long-term investment. They get a (relatively cheap) court judgement against these people, which requires them to pay back a small amount each month for a very long term. It generally works out to 5-10% annual interest, but pretty much guaranteed over a 10-25 year period. The risk is very low and they're a steady stream of income for an up-front capital investment. This is why you get a lot of card companies advertising to students and other low-income groups.

While technically true, there are some corner cases. In the UK, Jeremy Clarkson discovered this when he tried to prove a point by posting his account number and sort code online. He then discovered that it's possible to set up a direct debit with just this and a signature. Anyone that accepts direct debit payments is bound by the code of conduct that requires them to return anything taken by that mechanism if it's disputed and they're easy to cancel, but he did end up being signed up to donate money to a few charities. You'll get the money back in the end, but there's a lot of inconvenience possible until you do.

I know it's hard to remember the first statement in a post by the time you get to the end, but I'd recommend trying before you write a snide reply.

I remember having a Quake directory that was 500MB, when the original game was about 50MB. 90% of it was user-generated content. This included a load of maps and a load of mods. I don't remember the names of all of the maps, but I do remember that almost all of the time we played was on third-party maps, not on ones that came with the game.

If your system uptime relies on individual nodes having high uptime, then you're doing it wrong.