Using 'setTimeout' with string syntax internally makes the javascript engine 'eval' it. Whenever the browser encounters an 'eval' anywhere in the code, it cannot do many of the optimizations (and hence disables them) just because anything can get into eval.
Optimizations such as caching variables cannot be done with 'eval' existing in the code because 'eval' may introduce new variables which will get ignored during the compilation phase of Javascript (where it detects all the declarations).
The second syntax is faster because it will just call the function after the delay and you won't get into the evils of 'eval'.
