David Murphy has a post at TOD with a good graphic showing the effect of declining EROEI on a Hubbert curve (the "Net Hubbert Curve") = The Net Hubbert Curve: What Does It Mean?.
Cutler Cleveland of Boston University has reported that the EROI of oil and gas extraction in the U.S. has decreased from 100:1 in the 1930’s to 30:1 in the 1970’s to roughly 11:1 as of 2000 (Figure 1). But beyond the fact that society receives currently around 11 barrels of oil for every 1 barrel that it spends getting that oil, What does this mean?

Well, first, it means that, if the trend of declining EROI continues, society will be spending an increasingly larger chunk of their remaining energy to get more energy. This cycle is positively reinforcing:

Declining EROI means that the net energy contained in each unit of energy delivered to society is decreasing over time, requiring the extraction of increasingly greater quantities just to meet societal demand →
decreases the quantity of energy remaining in the ground for future society →
makes it more difficult to find and develop the remaining bit of energy.

With every barrel we pull out of the ground we propel ourselves further down this path, creating a more difficult situation for future generations. (note: I assume that the “Best First Principle” applies to this scenario, i.e. society is using the best resources (i.e. oil fields) first, then the second best, etc…)

More importantly, declining EROI also means that the amount of discretionary energy available to society is FAR less than that predicted by a Hubbert curve (Figure 2). The Hubbert curve represents the total gross quantity of energy available, and, as it is calculated, there are equal quantities of energy available on the left and right side of the peak. This, however, is only true in a gross sense. The net energy available (i.e. discretionary energy) is less. In other words, declining EROI means that there will be much less net energy extracted post-peak than pre-peak on the Hubbert curve.