Speed is a pretty important reason, and users not caring is also a pretty important reason. Users do care about speed, and don't care about absolutely ideal line breaking; so why trade off something they do care about for something they don't?
Well, it was an important reason in 1990, maybe, or even 1995. But we’ve had a couple of orders of magnitude improvement in computing speed since then. There’s no reason to favor some marginal speed advantage (tenths or hundredths of a second) to better layout.
Tenths of a second can be pretty damn significant. Every tenth of a second extra load time leads to approximately a 1% drop in sales, according to research done by Amazon.
One of the big points of competition between browsers these days is in speed. Part of the reason people switch from IE to one of the more modern browsers is because they are just that much snappier.
Maybe. Though the only time there’s going to be a noticeable slowdown is when you’re dealing with huge amounts of text. Like an amount that will take 20 minutes or an hour to read. At that point, an extra 10th of a second to render, in return for a much more pleasant reading experience seems like a no-brainer trade-off.
On a typical Amazon page the difference is going to be unnoticeable.
You can just do the fast algo. first, and do the more sophisticated one, when your browser is idle. (The text will jump around a bit, but it does so anyway while loading.)
