I don't think this is wrong at all. It's commonly *assumed* to be wrong, but I have explained at length why I think it is this assumption that is wrong [linky]

You only analyzed photovoltaics, which have the lowest EROEI of all the options.  Among the flaws of your analysis:

1. You ignore wind.  Wind has an estimated EROEI of up to 80:1 over a 20-year lifespan (payback time as little as 3 months).
2. The world produces considerable biomass which is not used to its best effect.  This is already being produced, so new energy inputs would only be required for processing and shipping.  Given that some of the products (e.g. torrefied biomass) have an energy content similar to coal but would be produced closer to some markets, the energy devoted to shipping may go down.
3. Even silicon PV is changing.  Historic energy balances have been for PV silicon produced as a byproduct of the semiconductor industry; we may soon see PV-grade silicon produced for $18/kg.

Forget PV for a moment.  If you can show serious problems with the proposition that a sustainable energy system is impossible in the USA (as an example), I'd like you to go over my "Sustainability" essay and show me exactly where I went wrong.  Do note that I assumed a lot less liquid-fuel production than the maximum possible (I devoted about 2/3 of the carbon for sequestration), and even if technologies like the DCFC turn out to be duds there are backup paths which will yield a European-level energy consumption.

Greater production (work, actually) from existing supply keeping up with declining production. This certainly seems valid for a while. However, there are fixed limits to how efficiently you can use a resource (100% efficiency, which won't be reached).

Let's suppose that I put a wind turbine on a 1/4 acre pad at the edge of a cornfield.  This is a 126 meter diameter rotor, spaced 4 diameters on-center crosswind and 10 diameters downwind to the next machine.  Total land area blanked by the turbine is 63.5 ha (157 acres).  Total blade weight is ~54 tonnes, of which perhaps half is derived from resins (the other half is glass fiber).  Assume fatigue lifetime of 20 years.

If 80% of that land is cropped in maize every other year at 150 bu/ac/yr grain and 2.5t/ac/yr stover, the land will produce 157 t/yr of stover.  If the stover can be converted to e.g. bio-oil at 70% mass-efficiency and the bio-oil to resins at 20% efficiency, each year's stover can make 22 tons of resin.  This compares favorably with the 27 tons of resin required to replace the rotor after 20 years (1.35 t/yr).

This is a recipe for rapidly increasing production (80%/year growth potential) from renewable supplies.  It would be able to yield both food and energy with more than 90% of the non-food products (both energy and chemicals) available for other consumption.  In short, it grows like a plant.

Even 14% efficiency from biomass to a resin product seems to be no obstacle to rapid growth.  The key is that there is some efficiency level where the cost of renewable material and energy inputs stops being a limiting factor on system growth.

As for keeping up with declining production, let's assume a 2%/year compounded decline in fossil-fired electric generation (2005 figure:  ~2000 billion kWh/year) and a 40%/year compounded increased in wind generation (14.6 billion kWh/year in 2005).  The fossil-fired decline runs about 40 billion kWh/year/year, while the wind generation increases about 4 billion kWh/year in 2005, 8 billion kWh/year in 2007, and 32 billion kWh/year in 2011.  The increase in wind exceeds the decline in fossil in 2012, and then it's uphill until the inflection point of the logistic curve is reached — perhaps another decade.

Annual wind energy potential from the continental 48 states is estimated at about 1.2 terawatts average (~10,000 billion kWh/year) from the continent, and about 900 gigawatts (~7500 billion kWh/year) from the continental shelves.  This is sufficient to supply all current end-use energy consumption.

More goods and services due to increasing competitiveness of renewables--this, too, runs into the "finite world" issue, and does nothing to address the diminishing marginal returns problem as the ultimate cause of collapse.

Please show me where those limits are in a renewable-energy scenario (don't forget to assume 25%-efficient PV at $1/watt within the next 20 years), and why they'd lead to a collapse instead of a "climax forest" scenario.