Because of the evidence.
Uncontrolled observation and experience are not good forms of evidence, because people are very good at seeing causal connections where there are none.
To take homoeopathy as an example, while there are many anecdotal accounts of self-evident effects, when experiments that control for bias are carried out these effects disappear.
Additionally, homoeopathy is wildly implausible because its principles contradict things that we already know well enough, for example, for computers to be built and to operate reliably enough for us to be having this discussion. Chemistry, physics, and engineering all work reliably enough to be relied upon on a routine basis.
