In the 1970s Charlie Hall, an ecologist at SUNY-ESF, was studying salmon migrations and asking how much energy the fish expended versus what they gained from feeding. He realized the same accounting applied to civilization. The energy returned on energy invested — EROEI — of any source is the ratio of useful energy delivered to the energy spent extracting and processing it. Conventional Pennsylvanian crude in the 1930s ran around 100:1. By the 2010s, conventional global oil had fallen to 10–20:1; tar sands and shale oil sit at 3–10:1. The number is unfamiliar in mainstream energy discussion, but it bounds what a civilization can do with its energy.
EROEI is not the same as cost per unit, and that is what makes it interesting. A source can be cheap in dollars and have a low EROEI, or expensive and high; what the ratio captures is the net energy delivered to the rest of the economy after the energy sector has fed itself. At EROEI = 1 a society would have to devote all of its productive output to extracting its own fuel, with nothing left for agriculture, manufacturing, education, or leisure. The rise is non-linear: below a threshold often estimated near five-to-one an industrial civilization probably cannot sustain itself; above thirty or so the marginal gains shrink. The historical trajectory is a story of finding sources with progressively higher EROEI — pre-industrial agriculture, coal in the nineteenth century, conventional oil at its mid-twentieth-century peak — each step releasing labor and capital from the energy sector for use elsewhere, then a slow erosion as the easy oil was drained and unconventional sources became the marginal supply.
The argument matters most at the technology-substitution margin. If a society on twenty-to-one substitutes a new source at five-to-one, net energy outside the energy sector falls even if the substitution is complete by volume; the energy sector grows to compensate, and the rest of the economy contracts. This is the structural concern behind worries that renewable substitution might constrain growth in ways cost-per-kilowatt-hour does not capture. Critics — Murphy and Hall among the most-cited — note that the metric depends on where the system boundary is drawn, that solar's lifecycle EROEI is genuinely improving as panel manufacturing efficiency rises, and that the past-transition analogy is imperfect because sun and wind carry no fuel cost once installed. The rebuttal: civilization-scale substitution has never been attempted on a hard time-budget, and the transition itself demands large net-energy investments that show up on the wrong side of the ledger.
EROEI accounting is now standard in energy-systems modeling: IPCC Working Group III scenarios, IEA sustainable-development pathways, and academic integrated-assessment models compute net-energy yields rather than gross production. The batteries-and-storage question is where the contemporary numbers are most contested — lithium-ion battery EROEI is roughly 5–10:1 over a typical cycle life, and grid storage adds enough overhead that some critics worry the renewable-with-storage system EROEI may be lower than headline solar and wind numbers suggest. The honest position is that the system EROEI of a fully-decarbonized twenty-first-century economy is not yet known.