There's a baseball-geek stat called equivalent average (EqA), which basically works on the same scale as batting average (i.e., .300 is good). Baseball Prospectus 2008 projected Stewart for a .275 EqA; Johnson was pegged to put up an EqA of .260. Stewart was also due to make less money.
Those projections are park-neutral, but were made on the assumption Stewart would be playing in Oakland. He might gain a few points of EqA yet since he hits a lot of doubles. Rogers Centre is more generous for two-base hits than Oakland's stadium (larger outfield, natural grass, ocean air that keeps balls up in the air a split-second longer).
The drop-off on defence isn't as big as some people would have you believe. Stewart got tagged early in his career as a sketchy outfielder since he couldn't throw and was compared to Devon White standard (in typical price-of-everything, value-of-nothing Toronto-fan fashion, people probably figured every centrefielder should be like Devo). Last season in Oakland, though, his Revised Zone Rating, .898, was a smidge higher than the .891 Johnson posted when he was healthy in 2006. Besides, a left fielder's throwing arm is about 5% of his job.
In other words, Johnson can take solace that plenty of National League teams will need a fourth outfielder. Hey, Curtis Granderson is hurt in Detroit.
(UPDATE, 3:30 p.m.-ish: Drunk Jays Fans has a good thread going; on a holiday, no less.
UPDATE, 4:30 p.m.: Jeff Blair's take: "Johnson was auditioning to be a fourth outfielder — not a fourth line winger on an NHL team — so stow all that nonsense about grit. Stewart's as capable as Johnson of getting hit by a pitch ... If you're counting on your left-fielder to throw people out, you're in big, big trouble.")