Establishing an extensive and highly durable, long‐term, seafloor network of autonomous broadband seismic stations to complement the land‐based Global Seismographic Network has been a goal of seismologists for decades. Seismic signals, chiefly the vibrations from earthquakes but also signals generated by storms and other environmental processes, have been processed from land‐based seismic stations to build intriguing but incomplete images of the Earth’s interior. Seismologists have mapped structures such as tectonic plates and other crustal remnants sinking deep into the mantle to obtain information on their chemical composition and physical state; but resolution of these structures from land stations is not globally uniform. Because the global surface is two‐thirds ocean, increasing the number of seismic stations located in the oceans is critical for better resolution of the Earth’s interior and tectonic structures. A recommendation for a long‐term seafloor seismic station pilot experiment is presented here. The overarching instrumentation goal of a pilot experiment is performance that will lead to the installation of a large number of long‐term autonomous ocean‐bottom seismic stations. The payoff of a network of stations separated from one another by a few hundred kilometers under the global oceans would be greatly refined resolution of the Earth’s interior at all depths. A second prime result would be enriched understanding of large‐earthquake rupture processes in both oceanic and continental plates. The experiment would take advantage of newly available technologies such as robotic wave gliders that put an affordable autonomous prototype within reach. These technologies would allow data to be relayed to satellites from seismometers that are deployed on the seafloor with long‐lasting, rechargeable batteries. Two regions are presented as promising arenas for such a prototype seafloor seismic station. One site is the central North Atlantic Ocean, and the other high‐interest locale is the central South Pacific Ocean.