Since this is a part of the defining sustainability series, I want to throw add my own approach as well.
It is true, as EnergyNumbers points out, that localization does tend to add some inefficiencies. If you compare a small bakery to a large bread factory, there are some optimizations that can happen with the latter that can't happen with the former (for starters, the ovens, by virtue of being larger, will have higher volume to surface area ratios and this reduces accidental energy loss). Similarly centralized transportation requires less inputs than decentralized transportation. The farmer who takes 500lbs of produce in his pickup truck 200 miles to sell at a farmer's market uses up a lot of fossil fuels compared to what would happen if things were shipped in larger quantity.
Taken to its extreme I think one might be tempted to say that Walmart is good from a sustainability perspective because it saves us from farmer's markets and local bakeries. The problem is though this optimizes (and optimizes badly, I might add) for only one factor and thus, in my view runs the wrong way.
Before I say why I think that runs the wrong point it is worth saying that a lot of things cant be localized. Many of the higher technology components, like solar cells, really do require a certain economy of scale even to hit break-even points energy-wise. For this reason I don't think that things go all one way. This being said, I think sustainability requires to some extent decentralizing the economy from where we are today.
Localization and Agriculture
Our current system of food production, distribution, and processing is very energy-intensive. The majority of this is in indirect costs, such as pesticides and fertilizers. The massive use of fossil fuels allows modern large-scale farming techniques and there is no current substitute which allows downsizing our nation's farms without the large input of fossil fuels.
It's worth noting that the congressional report cited above notes that the majority of energy consumed in the production of most crops is in the manufacture of fertilizers and pesticides, not the use of diesel to plow land, harvest, etc. Cutting down on fossil fuels in agriculture thus means abandoning large-scale mechanized farming of hundreds of acres and a move towards smaller scale, where concerns over soil fertility, pest control, etc can be managed through biodiversity. Thus liquid biomass fuels don't really even approach a solution for agriculture.
The primary contenders to solve this problem both require downsizing farms and replacing cheap fossil fuel inputs with more expensive human inputs. These include smaller scale organic farming (note that draft animals are making a comeback as a reaction to high fuel prices. Organic farming by draft animal potentially gets rid of the vast majority of fossil fuel inputs.
The other possibility is permaculture, which tries to harness natural dynamics between species for support of plants and control of disease. Permaculture is small-scale, however, and while it is high yield at low inputs, most of the human-directed inputs are directly human.
I don't see any possibility of achieving sustainability without decentralizing farming. This means a shift of inputs from fossil fuels to people and animals. That has a lot of implications regarding lifestyles, and a need to significantly downsize energy consumption in order to make that happen (more people working in agriculture, fewer in energy).
Localization and Energy
Some materials in this area do not work well from a standpoint of localization and so manufacturing and distribution of these may not be able to be localized. However, I think that there are still very important benefits from thinking about the problems in terms of localization.
The first is that the most promising technology to replace fossil fuels for electricity generation is wind power, at least given current EROI figures. Wind has a very good EROI. The real challenge is that wind is fickle and unreliable, and so it is hard to generate baseline energy over a small area. One solution for this to have large numbers of geographically dispersed wind farms but this requires a power grid designed for a level of what might be described as "chaos" that they are not designed for today (where things are based on large, reasonably dependable power plants serving specific areas). Thinking in terms of localized production, done well, is an important aspect I think to more distributed energy production (and Wind requires thinking about energy production in a distributed manner).
A second point is that there have been installations of methane digesters at commercial dairy farms. Whether CAFO-modelled dairy farms may not be sustainable, the idea of smaller scale, local energy production off of material that would otherwise be discarded is very positive development sustainability.
Localization and Waste Management
The final area where I think that is important to consider here is the idea of local material cycles. Our current approach can be described as "factory to landfill" and the problem with this is that recycling becomes somewhat problematic and requires heavy centralization. While some areas of recycling can be centralized, things like organic material are harder to do so. We can't exactly ship sewage or compost effectively over the railway system and so the ability to take waste products and turn them into useful products is reduced by over-centralization.
Waste management is closely tied to energy and agriculture however. New York City is moving towards harvesting methane from sewage treatment for electrical power, and they are not alone. Some forms of fertilizer come out of wastewater treatment and there are many other products which can be seen as locally produced in this way as well.
Robustness vs Efficiency
With modern agriculture and global distribution of food, we still see famines as a result of civil war and other factors, where disruption of distribution, in the distribution network leads to catastrophic human costs. There are major issues here including the over-centralization of agriculture globally and this raises very significant social and economic sustainability concerns. Overreliance on long-distance trade raises concerns about robustness. This means that overcentralization trades efficiency for a lack of robustness.
TL;DR
In general, I think decentralization, re-ruralization, and localization, compared to where we are today, are necessary for attaining sustainability. This does not mean however, that everything can be decentralized. Some things may require greater economies of scale to be viable. However our centralization today is heavily subsidized by fossil fuels and to kick that dependency one must move somewhat in the other direction.