My first experience of a web browser was on the Amiga - IBrowse, AWeb and Voyager.
Computers in those days were so slow that you could watch the images on the page decompress line by line.
I didn't actually have internet access at the time, so I was viewing web sites that came on a CDROM with a computer magazine.
Eventually we got dial-up internet access, at 33kbps.
Although the PC could now decompress the images quickly, they didn't arrive quickly from the internet.
I'd open several web pages in separate windows, so that I could read one page while the others were downloading.
Tabbed browsing made this a much nicer, as it was much quicker to open a tab than a window.
Now that we have broadband, the pages don't take long to download, but I still use lots of tabs with gmail and bloglines open most of the time.
Browser memory usage has become an issue, as the web browser is no longer an application that is opened, used and then closed. It's now open all day. When an application is using lots of memory, it doesn't all fit in RAM at the same time, so the operating system has to swap some of it to disk and read it back later. The disk is much slower than RAM, so this slows the whole computer down.
Firefox 3 has made some great improvements in memory usage, but as it's written in C++, it is unable to use a compacting garbage collector that moves objects around, so the memory gets fragmented.
Google Chrome is avoiding this problem by having a process per tab. Freeing all the memory used by a process is very simple, just terminate the process.
I hope that this hasn't resulted in it being slow to open a tab.
I can't wait to try Google Chrome when it is released for Linux.