Philosophy != Science, but both have their place.
There are plenty of avenues for creativity, discussion, unproved hypotheses etc, but peer-reviewed magazines are not one of them. That way, everyone can distinguish solid, confirmed results that can be relied upon, from unproved assertions or tentative conclusions that might be right - or might not.
Scientists are free to follow hunches or interesting leads; nobody is stopping that. But there has to be a clear indicator of the reliability of information, and solid peer review of methodology is the best method we've found of determining that.