The idea of media-based modulation (MBM) is to embed information in the channel states via intentional perturbations of the transmission media. This article covers a broad range of topics regarding MBM, expanding on its benefits and reviewing relevant challenges, alluding to potential future research directions. The article starts by arguing how MBM differs from a source-based modulation; we highlight the key shortcomings in a legacy multiple-input multiple-output (MIMO) system that MBM sets out to address, including the issue of deep fades and MIMO diversity-multiplexing trade-off. The article further explains how MBM works in harmony with other index modulations and improves upon them by providing similar advantages with a more compact transmitter. Numerical results (simulation and analytical) are provided to support the claims on the discussed benefits. The highlights of numerical results include: 1) outage comparison with legacy MIMO systems; 2) comparisons with other state-of-the-art modulation schemes such as generalized spatial modulation; and 3) performance example of sending 32 bits of information in a single transmission with an excellent symbol error rate of $\mathsf{SER} \simeq 10^{-5}$ at \say{energy per bit to noise power spectral density ratio} of $\mathsf{E_b/N_0} \simeq -3.5$ dB. The article continues with methods to address the issues of receiver training and decoding for large constellation sets. A number of other research questions, such as pulse shaping to limit spectral growth due to the time-varying nature of MBM and the effect of forward error correcting codes on MBM diversity order are discussed. Finally, an RF transceiver structure is presented to generate independent propagation paths for embedding information. Fabrication and testing of the transceiver structure show close agreement between simulation and measurement.