I tend to double rhythms guitar tracks. As Bristol says, it "Introduce(s) subtle timing/pitch discrepancies" which is what fattens the track up.

As for the Haas trick (or copy and nudge), it can produce great results. I use channel tools to add a delay of less than thirty msecs to one of the tracks.

BUT, there is a caveat. Be sure to check it in mono, as it can cause comb filtering problems in mono. It is possible to lessen the effect, but it may take a bit of tweaking - turning down the copied track, EQ etc.

Sidroe
Another secret is to copy the original track to a new one. Then slightly shift the entire new clip just milliseconds out of time with the original. If you shift it too far it will sound like a digital delay. Shift it just enough to hear the difference between the two while panning hard left and right. If you want a different texture use the EQ. This technique makes for rock solid syncing of the performance on the two tracks. Quite often you will not get EXACTLY the same performance by manually doubling the track. But that is sometimes what you want. The little discrepancies make for a more live feel. 

Tony Banks, the keyboard player for Genesis, uses this technique for his keyboards.  I believe the widening threshold is something like 15 milliseconds or less.
 
I am a keyboard player that plays some guitar, and am currently experimenting with splitting an electric guitar signal so that one signal path goes into a direct box, then the 1/4" output from the direct box is fed to a guitar amp, and its line-out is fed to a line-level input on my audio interface, while the XLR output from the direct box is fed to a tube pre-amp - who's output is then sent to another line-level input on the interface.  
 
For the above, both signals will be independently tracked, with two separate instances of modeling and effects, additionally allowing the addition the widening technique to really get a rich wide sound.
 
Hope it works, it's on my to do list for today.
 
Bob Bone