This is an attempt to list all of the possible ways in which humanity may avoid scenarios where the average standard of living is close to subsistence, in response to Robin Hanson's recent series of posts on Overcoming Bias, where he argues that such an outcome is likely in the long run.
I'll start with six, some suggested by myself, and others collected from comments on Overcoming Bias and Robin's own posts. If anyone provides additional ideas, I'll add them to the list.
(I have a more general point here, BTW, which is that predicting the far future is very difficult. Before thinking that some outcome is inevitable or highly likely, it's a good idea to repeatedly ask oneself "This is all the ways that I can think of why it may fail to come true. Am I sure that all of them have low probability and that I'm not missing anything?" There may be some scenario with a non-negligible probability that your brain simply overlooked when you first asked it.)
Singleton
A world government or superpower imposes a population control policy over the whole world.
Strong Security
Strong defensive technologies and doctrines (such as Mutually Assured Destruction) allow nations, communities, and maybe tribes and families to unilaterally limit their populations within their own borders, while holding off hordes of would-be invaders and immigrants.
Non-Human Capital
Maximizing the wealth and power of a nation requires an optimal mix of human and non-human capital. Nations that fail to adopt population controls find their relative wealth and power fade over time as their mixes deviate from the optimum (i.e., they find themselves spending too much resources on raising humans, and not enough on building machines), and either move to correct this or are taken over by stronger powers. (I believe that historically this was the reason China adopted its one-child policy.)
Unlimited Growth
We don't completely understand the laws of physics, nor the nature of value. There turns out to be some way for economic growth to continue without limit. (Robin himself once wrote "I know of no law limiting economic value per atom" but apparently changed his mind later.)
Selfish Memes
Memes that manage to divert people's resources away from biological reproduction and towards memetic reproduction will have an advantage over memes that don't. On the other hand, genes that manage to block such memes will have an advantage over genes that don't. Memes manage to keep the upper hand in this struggle (or periodically regain the upper hand).
Disease, Warfare, Natural Disasters, Aliens, Keeper of the Simulation
One or more of these come along regularly to keep the human population in check and per capita incomes above subsistence.
The big one:
Memetic instability, aka Hanson's assumptions are just wrong.
It's quite easy for an armchair evolutionary psychologist to say, "X group reproduces more than not-X. Therefore, by 2150, everyone will be Mormon because they'll be selected for!" Yes, I'm exaggerating a bit, but the actual argument is about as absurd. There aren't any large population subgroups with extremely high fertility as a subcultural norm. Given the substantial hedonistic sacrifice involved in high fertility, this is not surprising. It seems unlikely, barring kids being a lot easier to raise, that large subgroups of the population will maintain a high-fertility norm. This is even more true if people measure social status in material terms - high fertility inhibits material acquisition, and people tend to be pretty serious about pursuing status.
The fact that group X, which is 1.7% of the population, has high fertility, does not mean that their great-grandchildren will when they are a larger subset of the population. In fact, mere changes in geography could corrode the memes of such subgroups. The fact that, if every subgroup kept their current fertility for 1000 years, most of us would be Mormon/Muslim/what-have-you does not mean that this will actually happen. Assuming the future will be exactly like the present but more so has been systematically wrong for at least the last few centuries. Memes can change very, very rapidly, and their change is not directed by a meme-designer trying to beat all the other meme-designers; it's directed by human desires and cultures.
This also ignores the fact that most of the growth that occurs is in the third world, and most indicators show that their fertility falls as child mortality and women's education improves. Because women's rights are, generally, rather ratchet-like (they rarely get massively scaled back once they advance), these gains are likely to be sustained. As people no longer think they need 8 kids to support them, they'll start realizing they don't want 8 kids, and the old cultural standards will decay in a few generations. It seems like this is already starting to happen. Rather than actually refuting this apparent demographic reality, we get, "People who have more kids and teach them to have more kids will take over in several dozen/hundred generations." That's assuming some incredible memetic stability. I do not think that memes have shown that much stability historically, and they are particularly unstable these days.
Granted, it's entirely possible that technology will make child-rearing all play and no work, in which case Western demographic trends may reverse. The argument doesn't only cut one way; it does require that people be generally hedonistic and that kids cramp their style. However, there really isn't good reason to take Hanson's theory as the default position.