I am reading The Ultimate Resource 2 by Julian L. Simon (Princeton, 1996). He makes the point that commodities prices tend to decline over the course of the last couple of hundred years compared to the cost of human labor. The average US wage in 1999 dollars is growing at about 2.1% per year geometric mean from 1900-1999. The real interest rate has been about 1% (in the UK anyway).

This means that if technology were constant and reserves were constant, the real price of a commodity would rise 1%. Otherwise, it would make sense to mine as much oil as possible and put the money in the bank, or leave the oil in the ground and wait for the price to rise.

In fact from 1860-2000 the price of kerosene (see figure 5) has dropped about a factor of 6. If you look at the service kerosene was providing (lighting), it has dropped a factor of 40 from £125 to £3 (in constant 2000 £) per million lumen hours. The lighting was overwhelmingly provided by gas and kerosene in 1900 and electricity in 2000. That technology change was the equivalent of an increase in the availability of light with no new resources being discovered.

As we discover new resources, new extraction technologies emerge, and new cheaper substitutes narrow the uses for the expensive commodity, the price tends to drop. With the price dropping over the long term, that leads us to the corner solution of mine everything you can right away. The race is to extract the resource before its price drops. There are of course temporary shortages that spike the price and help spark technology change.

The good news for the species is that if the past is any guide, we can expect to see energy prices continue to drop in real prices and even more in terms of purchasing power. That is substantially more optimistic than this piece I filed last September. So my daughter in 30 years ought to be able to afford 6 times as much energy service as I can today. As the spaceflight price drops to where the fuel is a significant component of the cost, declining energy costs are good news.

Fischer-Tropsch is our friend.

As energy gets cheaper relative to labor, and we gets richer, more energy will be consumed.
This process should set an upper bound on how long it will be before there is a strong economic incentive to move into space, since there's a fixed limit on how much waste heat can be radiated from Earth.

If artificial intelligence is developed, it might cause the growth in power consumption to accelerate (since the doubling time of an AI server farm is probably

There may come a time when AIs and other powerful computers will have to be placed in space, where their waste heat can be radiated without interference.

I talked to a Dell manager at dinner and after raising the point he said that using a computer to heat a house might be in our future. I don't think we are going to be like Niven's Ringworld Puppeteers and generate enough heat that we don't need the Sun anymore.

I think we are more at risk from poor carbon policy leading to greenhouse than sheer heat generation of human activities being an issue. We need a factor of 10 increase in consumption to get to 2000 kW per person worldwide. Hopefully we will be enscounced in space well before we hit 500.

IIRC, the per capita primary energy consumption in the US is at a rate of about 10 kW/person. This includes all energy used in the economy, not just that used by individuals for themselves and their families. Europe was about half this, and the developing world even less.

Right. To get the rest of the world up to US levels would require a large increase in consumption.