The scientific method has several well defined steps.
1. See something interesting.
2. Make up a story about it.
3. Change or embellish the story to make it sound more plausible.
4. Try things out to see if the story explains anything real.
5. Throw the ridiculous fantasy out and start over, or publish the story, along with all supporting experimental and theoretical data, as a possible fact.
Some sciences are better at verifying their stories than others, and some stories seem to be extremely useful, but we are still basically making up stories to explain the mysterious world around us. As long as we don't skip any steps, we end up with some pretty reliable stories. Going straight to publication from Step2 is where we run into trouble.