Virtual particles in the Nobel lecture by Frank Wilczek
-------------------------------------------------------

The Nobel lecture by Frank Wilczek at
   http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1150826/ 
talks near the beginning about the traditional virtual particle picture:
''Loosely speaking, energy can be borrowed to make evanescent
virtual particles''. 

Note his qualification ''Loosely speaking'', which indicates that this 
cannot be taken seriously. He also says why - because one encounters 
divergences once we try to take them seriously. 
Then he gets more serious and shows how renormalization fixes the 
problems, though he does not say that this comes at the cost of making 
the virtual particles infinitely heavy, and hence again physically 
meaningless. But this can be easily deduced from the description of the 
renormalization procedure in any textbook on QFT.

Later, he slips back into the traditional jargon since it
provides a vivid intuition about Feynman diagrams -- especially
for the many nonexperts in his audience, but again he does so
with a careful, explicit caveat:
''(I'm being a little sloppy in my terminology; instead of saying
the number of virtual particles, it would be more proper to speak
of the number of internal loops in a Feynman graph.)''

Towards the middle he mentions lattice discretizations, and how they
cope with the problem in a nonperturbative way by not invoking virtual
particles (i.e., formally correct, a loop expansion) at all.