When contemplating future scenarios of energy contraction, there are some questions that come up over and over. They appear to be simple and straightforward at first, but on further examination they often turn out (as far as I can tell) to be unanswerable. Consequently, they also tend to be sources of endless and usually fruitless debate.
1. How much energy is required to make a [thing]?
The easy question that can be answered is "how much energy is currently used to make a [thing]?" That's fine, but that easy question does not always provide a useful answer to the harder question.
There are some processes for which the minimum required energy as set by physical law is reasonably close to the actual amount of energy currently used for the process. And some of those processes are extremely important: reducing iron and aluminum, making ammonia, propelling a vehicle of a given type at a given velocity. Iron, aluminum, ammonia, plastics, and similar materials are therefore exemplars of the concept of embodied energy.
But bicycles, chainsaws, and computers are not among those tractable examples. Nor is the actual transport of most (non-perishable) products, for which the value of velocity (and hence, the choice of transport methods) is determined entirely by abstract economic factors separate from the actual cost or value of the products transported. The embodied energy of a computer or bicycle is necessarily greater than the sum of the embodied energy of its basic materials (which is relatively small), but how much greater, I don't believe anyone can say. For most products, Question 1 is somewhere between difficult and impossible to answer.
Manufacturing processes are designed to minimize total costs (including capital costs) based on present or anticipated near-future costs of materials, labor, and energy. If the cost of flushing a part with 1,000 gallons of fresh water is a penny less than the cost of achieving the same results by hiring an employee to wipe the part with a damp rag, the former method will be chosen. If running newly painted parts on a production line through a 10,000 BTU/hour oven to dry the paint costs less overall than having employees put the parts on drying racks and shuffle the racks through a warehouse to allow them to dry for 48 hours at ambient temperature instead, then the oven will be there.
For most products, generating a realistic curve of cost per unit (including retooling costs starting from the current configuration of the factory) versus energy price would require an extensive and expensive analysis of a wide variety of alternative methods on all scales (from the overall manufacturing process chosen, to individual details such as the cleaning and paint-drying examples above) by a team of engineers possessing among them a wide assortment of specific expertise.
We cannot expect journalists to do any such analysis, or even, in most cases, be aware that it could make any difference. Instead, they can only report "X amount of A and Y amount of B are needed to make a C," usually in the context of an impending shortage of A or B, implying that a shortage or radical price increase of C is inevitable. In many cases that turns out to be true, especially when all considerations are in the short term. But in many other cases it does not, and the expected price spike or shortage does not occur. It's not a reliable chain of reasoning.
2. What is the true EROEI of a given process (e.g. manufacturing and using PV cells), taking all relevant factors into account?
Since the energy cost is a necessary component of EROEI, the frequent intractability of Question 1 above already presents a formidable and sometimes insurmountable difficulty for answering Question 2.
But there's another important problem as well: there is no straightforward way to draw a clear envelope encompassing all the relevant factors. If a product requires a mined and smelted metal, for instance, it's obvious that one must include the energy required to mine, smelt, and transport the necessary quantity of the metal. But what about, for instance, the energy expenditure required to maintain law and order along the transportation corridor? On the one hand, if the road to the mine is impassible due to bandits, the metal will not be available. So it can't be left out. But on the other hand, maintaining civil order has many other benefits; if one charges the entire energy cost of doing so to the single product under consideration, that cost will be far overestimated. (Ultimately, that error, made repeatedly and systematically, will burden every individual activity within an interconnected economy with the total cost of all activities throughout that economy. A vast system of invested capital (let's say, fifty trillion dollars worth) underlies your ability to walk to the hardware store and buy a wood screw. But that doesn't mean the "true cost" of the wood screw is fifty trillion dollars.)
To get a useful answer about the contribution of "overhead" to the EROEI of a product, one would have to carefully determine the proper share of each of the innumerable overhead factors toward the product in question. Doing that, though, requires knowing (at least in some sort of summary form) all of the other economic activity going on that share those same factors. That's difficult even for the present day when the quantitative details are at least theoretically knowable; it becomes quite impossible under speculative future scenarios. In essence, one cannot conclude anything about any one industry, until all the details of every other industry are concluded.
Those who assume present day levels of availability of transport, expertise, capital, civil order, and other infrastructure components in future scenarios are rightly faulted for doing so. However, those who assume that any or all of those things will be completely absent are making just as strong an assumption, and thus are equally unable to make any useful predictions.
That problem leads to Unanswerable Question 3:
3. What technologies/products will and will not exist in a given energy-contracted future scenario?
For me, any predictive methodology for answering this question has to pass what I call the whale oil test. It must "postdict," given early industrial age conditions as an input, that highly complex sailing vessels with large crews including numerous highly trained specialists could be built, outfitted, and sailed halfway around the world for years at a time, all at great expense, facing unpredictable weather and pirates and the fallout from ongoing naval warfare, to risk life and limb hunting enormous animals from rowboats with hand weapons, in order to obtain oil for lamps to light rooms. If your heuristics for deciding on the feasibility of technically challenging endeavors do not assign whale oil in the early 19th century at least a reasonable possibility, then you cannot trust it to predict whether or not LEDs will be made (clean rooms, rare earth metals, and all) and used to light rooms in a post-industrial future.
The thing is, it doesn't take the entire industrial output of the modern world to make a clean room. A clean room requires a few different things, each of which can be done in a wide variety of ways. An electric motor is currently the most practical way to turn a fan to move air through filters, but a steam engine, horses, water wheel, or human crew could do it instead (working from outside the clean room, of course). Tyvek garments are currently the most practical outerwear for clean room workers, but workable substitutes (perhaps silk, paper, or shaved nudity) probably exist. Solar concentrators can, at least intermittently when weather permits, reach the necessary temperatures to serve the purpose of electric furnaces. Electric lighting is very practical, but sealed glass skylights would not compromise a clean room (although one would hope that an LED factory, at least after being in operation for a while, would itself be LED lit if it had to operate at night). If the presently used air filter materials were not available, substitutes could probably be found. (No, I can't say what those might be. Coffee grounds? Spun sugar? Present day engineers have off the shelf tested solutions available, and so are not likely to research substitutes, until it becomes necessary.)
Such possibilities would be irrelevant if LED lights were an energy extravagance that the future world could never afford anyhow. But the life cycle energy efficiency of LED lights (powered by any available mechanical energy including human muscles if necessary turning dynamos) far exceeds (as best anyone can tell given the problems discussed above for Questions 1 and 2) that of any other artificial light source including candles, vegetable oil lamps, and wood fire. So the question isn't whether the future world can afford such extravagance, but rather, what forces would prevent the people of the future from exploiting the most economical option, forcing them to resort to the greater extravagance of candles etc. instead.
It's quite plausible that such forces could arise. If the knowledge of how to make LEDs from raw materials is lost; if everyone is reduced to such a state of subsistence survival that capital for even a highly profitable project is unavailable even to the ruling class; if taboos against technology become codified in the prevailing religion; if clean-room-looting barbarian raiders roam the land unchecked... then there will be no LED manufacture.
Those are extreme scenarios, though, and not popular among long-descent proponents, for good reason. (Certain taboos depicted in Greer's novel Star's Reach excepted...) The more prevalent viewpoint seems to be the missing-ingredient theory: the notion that some specific bottleneck, some particular unavailable product or substance, must eventually stop the enterprise short. That's a reasonable hypothesis, but only if possible work-arounds and alternatives are carefully considered and ruled out. To point to any arbitrary element -- polymer fiber filters or electric furnaces or Tyvek suits or electric fans -- and decide that its likely unavailability would doom the prospect, is to make the assumption that no work-arounds or alternatives can exist. It risks claiming the equivalent of, 19th century whaling was impossible because there was no steel cable, GPS, radio, sonar, or fiberglass sails -- all of which would have been used and considered indispensable by whalers if they had been available. That's the Indispensability Trap: indispensable does not really mean indispensable. Except when it does.
Because of course, one cannot say that a decisive bottleneck, an indispensable ingredient, cannot possibly arise. Available resources and human ingenuity have limits. If there were really no such thing as an indispensable part, such problems could always be circumvented, and we'd all have flying electric cars that fold up into briefcases. Perhaps LEDs really do require a mineral that can only be obtained from one specific mine in China. (As opposed to, that mine in China merely being the cheapest present-day source, and therefore the only source that anyone in the present day world economy bothers to draw on.) Perhaps no filter medium on earth can replace whatever special high-tech polymer is currently used for clean rooms, and still get the air clean enough to make LEDs. Those types of possibilities must be considered.
The problem is, in many cases only experts in those particular specialties can consider them, and it might take a lot of work even for them to come up with reliable answers. (Assuming that even they can manage to get the answers right. I recall many predictions, claimed to be from qualified experts, that whole industries would be crippled by the withdrawal of "irreplaceable" Freon from the market.) We need, in the first example, a geologist with a particular sub-specialty, and in the second example, probably a dust filtering expert, an expert on the polymers currently used, and an expert or experts on the properties of a range of possible substitutes.
The general principles of history cannot answer in the experts' stead. The general principles of economics cannot answer in the experts' stead. Narratives about Easter Island or the fall of the Roman Empire cannot answer in the experts' stead. The general principles of ecology cannot answer in the experts' stead. Magic cannot answer in the experts' stead.
There is one thing that can answer in the experts' stead, though: empirical research. Trial and error. We can go looking for other sources of rare minerals. We can test different kinds of filters and find out whether any available alternatives work well enough. But no one's going to do that, until someone whose job it is to get LED manufacture working (or to keep it working) faces the necessity of finding those alternatives.
It also means that the details of the future depend -- big surprise, huh? -- on the results of R&D. That's a roundabout way of saying that they cannot be predicted at all.
