Switch On: To start expressing a protein-coding gene — synthesizing its product — you have at some stage to synthesize its mRNA — transcribe it.
Switch Off: To stop expressing a protein-coding gene it makes sense to stop making more RNA, as the poster suggests. But that may not be enough! Many eukaryotic mRNAs have half-lives ranging from hours to days, so that merely stopping transcription will not immediately stop the protein being produced. If it is important to stop synthesis immediately, something else is required. Nucleolytic degradation of the mRNA initiated by microRNAs is one way to do this.
More than one way to kill a cat
Why is this not more widespread? The answer is that this is only one of a variety of means to regulate gene expression. Others include having mRNAs of shorter half-life (as in bacteria), controlling mRNA translation, inactivation of the protein (if it is an enzyme) or degrading the product by proteolysis. Which method is employed depends on evolutionary chance, the nature of the system, whether it is advantageous to be able to resume activity quickly etc. So systems regulated by microRNA need to be considered on a case-by-case basis.
And although I am not competent to write on this (additions solicited) I am led to believe that mathematical modelling shows that the existence of a variety of regulatory mechanisms can allow a smoother and more buffered response to a variety of stimuli.