4,677 hours of failure in 4,344 hours of time means that at any given time, an average of 1.07 locations were offline.

There are 131 DMV offices in Virignia; I don't know how many other Department of Transportation locations are included in the same bucket. If we assume that it's *only* the 131 DMV offices, 1.07 failures at any given time means that the system means that 130.3 locations are working, meaning that this statewide patchwork of network connections is 99.45% reliable.

If your 'redundant' connections cut the failures in half (which they wouldn't), you'd have 99.59% reliability at more than twice the cost for the network.

Adding 'redundancy' would more than double the network cost (since presumably currently they're using the lowest bidder), and in most places it wouldn't add any real redundancy anyway. Getting actual network redundancy is *fiendishly* difficult, even when you're spending a lot of money and siting a facility in a place that's well-served for networking. In small-town Virginia, you're almost certainly going to wind up paying for having redundant wires hanging on the same poles.

Nah. The OS X girlfriend is very low-maintenance, except for the fact that she'll only consent to live in a house with everything already built in at construction time: a built-in coffee maker, microwave, etc.

Sure, a built-in coffee maker costs $300 while one that sits on the counter is $30. Windows likes the cheap coffee maker, but when it breaks you frequently find that you wind up having to tear down the whole house and build a new one around a replacement coffee maker.

The built-in coffee maker (almost) never breaks, and when it does it's easy to fix. However, it appears to be totally incapable of making cappuccino. You're not really sure, though, because the documentation just pretends that cappuccino doesn't exist. You seem to remember hearing that Steve Jobs thinks that cappuccino is stupid.

The Linux girlfriend waits for someone to deliver coffee for free. Amazingly, this actually works, every morning When you find out that what's in the cup two days out of five is actually tea, you figure that that's close enough.

How about they fire him and he leaves with NO money? That makes even BETTER business sense. I can pretty much guarantee that if I royally screw up at my job I won't be receiving any bonus and I'll be looking for work again.

Yeah, well, that would seem to make sense. But when most people at that level sign on for the CEO job, part of the contract provides for a large payout if they're fired. During the hiring phase, everyone is happy with one another, shaking hands, slapping backs, and toasting one another in the boardroom. I don't think a lot of these boards really think about the fact that, if they eventually come to fire this guy, it'll be because he has screwed up.

If you look at it charitably, you can say that the CEO is putting his own business reputation on the line by taking on the top job, and that the golden-parachute clause is insurance against this. Most geeks have had the experience of taking a job or contract with a company and having things fail miserably through no fault of their own: I once had a contract at a place where they told me that Perl was an 'insecure language', and that thus all their software was to be written in VB, for example. This kind of idiocy impacts the bottom line, and thus the CEO's performance as well -- and there isn't a whole lot he can do about it. If when you were fired, or left a job in disgust, it was going to be discussed in the Wall Street Journal, you'd probably want the contract written to provide for some kind of payoff as compensation for the company's idiocy.

Or you can say that this kind of compensation is itself part of the company's idiocy. The truth is probably that it's both. I don't believe that competent CEOs are anything like as scarce as their paychecks would seem to suggest.

But scarcity isn't the only reason, or even the main reason, why CEOs get paid so much -- it's also because their work is easily measurable: either the stock price goes up, or it goes down. Their job description is 'increase the value of this number', and so everything they do for good or ill comes down to that value. It's easy to write a contract that says, in essence, 'if you increase the shareholders' value by a billion dollars, we'll pay you 1% of that'. $10 million is still a lot of money, but 1% is what you get as a green salesman off the street; so you might consider 1% as a really cheap rate for a competent, experienced CEO.

On the other hand, it's very hard to measure the value to shareholders of a competent sysadmin, which is why sysadmins get paid less than the arguably less-crucial marketing and sales guys. Again: marketing and sales are easy to quantify. A lot of geek work is notoriously hard to quantify: not only is most of it preventative in nature, but it's also invisible if you do it right.

If you want to make the big bucks, find a situation where you can do what you do in a way that the value you add is measurable and quantifiable.