Several of my colleafues at PGP have studied hydrofracturing and fluid intrusions in the Earth's crust. Such intrusions of magma are sometimes 3-dimensional structures (magma chambers), 2-dimensional (sills and dikes), or 1-dimensional (volcanic channels). From elastic solids we know fractures to be essentially 2-dimensional and from fluids we are familiar with fingering instabilities causing breakup roughly speaking from 3 to 1 dimension. In order to complement field studies and numerical simulations by my colleagues I decided to study fluid intrusion in granular materials with appreciable cohesion between the grains.~\cite{Nermoen2010a} Preliminary studies of fracture patterns in quasi 2-dimensional cells using video and pressure sensors gave insights about the intermittent character of the intrusion process. This insight was important on a field trip to Mali to study a once again active hydrothermal vent system that had been described in scientific literature. Because the propagation mode of the smoking holes seemed more similar to a burning front than an intermittent bursting intrusion from a deep source I hypothesized that the structures we observed in Mali were not hydrothermal vents, but a subsurface fire. The hypothesis was verified the next day by performing the experiment: dig a hole in the ground. The article~\cite{Svensen2003a} about subsurface fires was editor's choice in Science and I was interviewed in Nature and numerous other science magazines, newspapers and radio programs about the article.t