Intuitively it seems like the "fear factor" approach *should* work.
However there have been studies showing that this doesn't have the
expected result.  Malcolm Gladwell in "The Tipping Point" talks about
an experiment at Yale in the 60's where the experimenter distributed
booklets to students giving information about why they should get a
tetanus shot:
http://jludvik.net/weblog/2005_01.html
One version of the booklet depicted the disfigurements and other
possible consequences of not getting a tetanus shot, and it was
expected that more students who read this pamphlet would visit the
clinic -- but they didn't.  The rate of students who got the shot was
only 3 percent in either case.  But when the pamphlet included a map to
the clinic and a list of times when shots were available (even though
this information was easily available to all students who wanted the
shot), the rate of students getting the shot went up to 28 percent.
This suggests that if you want results, then scaring people is not as
important as making it easy for them to follow your advice.

But you could certainly try the "fear factor" approach; maybe it's
different for people when they're warned about a problem that they
already have, not a long-shot risk like tetanus.  The final verdict
about what really works, comes from the numbers after you give people
the advice and see what happens.  All I'm saying is that to anyone
concerned about the obesity epidemic, that's the game you should be
playing: finding what advice works best averaged across all people who
hear it.

It seems the problem is that even among people who are qualified to
figure out what would work, nobody really has any incentive to give
advice that scores well on this measurement.  A doctor's first priority
is to make sure they give advice that it safe and won't get them sued;
whether you follow it is not up to them.  A government researcher's
first priority is to keep their own job; in a perverse way, it may
actually be better for them if obesity is on the rise, since their
agency gets more money!  And some people just get a kick out of giving
"advice" that puts down other people, implying that they eat too much
or exercise too little, and if you ask them if people are born with
different metabolism rates, they look at you like you're from Mars.

But if you create a contest like the one I described, then for the
first time ever you've given people an incentive to give advice that
helps people lose weight.  It would be interesting to see what people
come up with.

You didn't ask, but I'm about average weight and happy with it, so it
doesn't much matter to me personally.  It just seems like this is a
problem that lots of people talk about but nobody has tried this
approach.