> Why can't people simply publish negative results?

In addition to the other comments: because a negative result is usually much more likely than a negative one, it is very hard to be sure that the experimenter did not simply screw up the experiment.