When I click a link, it takes about three seconds for the page to load. When I click a link, then change my mind at the two second mark, and click another link, it loads in one second. Why?
Sheer perversity
Because 2/3 of the bits have already arrived in your computer, your browser just has to re-assemble them into the new page.
I don’t understand. Can you break it down Barny style? Are 2/3 of the bits from a given web site on different pages the same?
You won’t – there’s nothing sensible there to understand.
Control-z was making a joke (a rather bad one), which has no business in the GQ forum. And didn’t include any 
smilie so people might realize that he was trying for a joke.
Sorry.
He’s joking.
There are all sorts of reasons why pages might take a while to load. In general, what you are saying doesn’t make much sense. It could be coincidence, it could be the sheer perversity of the net. Loading the packets that represent one web page should have no relevance to another. If the second web page is created by the server dynamically (like a thread here at SDMB) none of the content has even been created when you click on the second link.
Well, control-z might be joking, but what he says isn’t complete nonsense; caching is one possible explanation for this behavior. It’s pretty common for web pages to be designed to include dozens of images, several JavaScript source files, a few CSS stylesheets, and maybe a Flash or two. Your browser has to download each of these, unless it’s cached, and each download requires its own HTTP GET request. For this six-post SDMB thread, all of the extras actually take more space than the HTML, even without the HTTP overhead; graphics-heavy pages will have a much higher ratio. But many of these extras are site-global and can be cached (at the SDMB, all of the extras except for the ad content are basically site-global), which will reduce later download times.
That’s not the only possibility. There could be some delay waiting for a nameserver to respond; these responses are also typically cached for a while, so later visits to the same site won’t require the nameserver query until that cache is emptied.
You can use a network analyzer like Ethereal to monitor the two requests and see if they generate about the same amount of network traffic, look for long delays between nameserver query/response packets, and so on.
3 seconds! Hmph, kids these days! When I was young we had to read along with the 300-baud text – and we LIKED it!
If the links are to the same domain name, then the second link might just consult the cache rather than asking the DNS to translate the domain name to an IP address.
Some ISPs (I’m looking at you, comcast) have notoriously overloaded domain name servers, which dramatically slow down web access. This might be intentional, as slowing down surfing reduces the overall bandwidth load.
Is a nameserver the server that hosts that straightdope.com means to go to xxx.xxx.xxx.xxx?
Would it make sense to say it was analogous to getting in line at McDonald’s, thinking I want a Big Mac (establishing my position in the queue with the first click), and then deciding I want a Quarter Pounder when I’m ready to order (second click)?
ETA: I think jokes certainly have a place in GQ. Except when they make me look dumb. 
Yes.
Maybe more like calling Directory Assistance to get the number for the local pizza place so you can order a large pepperoni, and then realizing you don’t have to call DA again when you decide to switch your order to a supreme–you’ve still got the number scribbled on your hand. Next week, though, you might have to call 411 to get the number again, since you’ve washed your hands.
Getting put on hold by the pizza place (in the queue at the HTTP server) is a separate issue, but I think that usually happens during a DOS attack, shortly before the server melts down. (If requests are consistently coming in faster than they can be processed, the server will eventually timeout on connections; otherwise the queue will usually be pretty short.)