The "scary" technologies/breakthroughs

[Gilthan]

Even you, I'm sure, wouldn't suggest we replace injection molds able to mass produce household plastic bottles of multi-liter size for ~ $0.01-$0.10 each with instead slower, more complex nanotech scale assembly lacking the economics of self-replicating tech?

That primarily depends on the cost of transportation. Right now, transportation is cheap relative to production, a fact that has dictated global trade patterns. If transportation becomes more expensive (which is a frankly likely outcome of oil depletion), localising production becomes more attractive.

Oh, here is where we have a huge difference! For one thing, the energy cost of manufacturing the typical product from raw materials to completion is literally multiple orders of magnitude above transportation energy requirements. For instance, it takes 12000 kWh to produce 500 kg of aluminum from bauxite, yet the 1600 kg 2011 model Chevy Volt only requires a mere 8 kWh to travel 40 miles.

If electric-powered rapid transit is the future, as probably so, increasingly automated, we're talking of around 0.1 kilowatt-hours per ton-mile transported. At $0.05 per kilowatt-hour from existing nuclear power plant economics or up to $0.10 per kilowatt-hour sometimes with distribution expenses, that corresponds to under $0.01 per ton-mile transported. Needless to say, that's extremely cheap. Assuming this is the future, there's probably no human delivery driver to pay, so even adding in other expenses, the potential to be actually significantly cheaper than today is very high.

If you can afford to buy the product at all, you can afford to get it shipped 50 miles just about as easily as 0.5 miles.

The main reason we spend a much more substantial amount than that on transportation fuel is: (1) mostly we shuttle people and their vehicles back and forth, like someone commuting to work hundreds of times a year; the actual cost of shipping all their rarer non-food manufactured purchases to them is pretty low in energy expenses (2) the average gasoline-fueled vehicle is 10+ times less efficient, while gasoline also costs more than nuclear electricity per MJ.

You are also completely ignoring the inherent customisation advantages of local direct assembly, relative to mass production where varying the output even slightly is difficult and costly. Correctly marketed, this is an extremely attractive advantage.

Customization can be done whether the factory is 0.5 miles or 50 miles away, on a web page, or donning your virtual reality / telepresence headset if you want to get really fancy. I suppose greater distance would make it more a nuisance if you wanted to physically be on site in person, but I don't need that to specify the color of my toaster, if I care. Of course, I won't deny that there are some items for which for which a local custom manufacturing facility within a fraction of a mile would be convenient and potentially relatively practical in the near-term, like household decorations, hobbyist models, prototypes for small businesses, et cetera.

We apparently disagree, however, over how readily custom small-scale fabricators would compete on price and performance with larger-scale factory mass production of challenging items from laptops to trucks, and we seem to disagree heavily over whether bulk macro-scale fabricators can affordably be nanotech without self-replicating technology.

Oh, you're referring to an integrated circuit chip with much larger and orders of magnitude less complex circuitry than a modern PC CPU. That's pretty far from nanotech.

A typical phone CPU has one to two orders of magnitude fewer transistors than a desktop core. This is a small gap relative to the difference in component counts between a processor and a nanofabricator.

A nanofabricator is a 3D MEMS with moving physical parts, whereas current CPUs are 2D aside from at most a handful of layers, a really huge gap in complexity for a given minimum feature size. Can you point to any particular company seriously attempting to build MEMS for a really low cost per square centimeter and cubic centimeter, as opposed to tiny units?

Oh come on, a trivial search would turn up something like this, an 8-bit microcontroller that costs 34 cents each. That's a sticker price for 100 units, I guarantee you that a manufacturer negotiating a contract for 100,000 chips will be paying half that.

Fair enough. I just used the first major distributor I looked at for a quick example. However, the reason for that is because it barely matters for the big picture. The $100000+ per cubic liter is an approximate order of magnitude estimate, yet there is still a huge price compared to macro-scale product manufacturing whether tiny microcontrollers are $0.34 or $1.80 each.

But cost per unit area / volume isn't decreasing according to Moore's Law, rather cost relative to the number of transistors within a given area.

Of course the cost per area isn't decreasing (no, actually it is; modern CPUs have larger dies despite being cheaper) when the feature size keeps decreasing. If the feature size is held constant, then the cost per area will decrease. You can see this quite clearly in that older semiconductor processes are used for progressively cheaper chips.

However, I don't think it is anything like a factor of 1000s reduction in cost per unit volume, not the needed orders of magnitude. What's the best example you can give for lowest cost per unit area / volume of micron-scale circuitry?

At an extreme large-scale size, I'm pretty sure you'd agree with a preference for bulk manufacturing. After all, nobody would argue that tiny assemblers produced by micron-scale lithography are becoming remotely as cheap per ton as a giant dump trunks for making large concrete pavements, at least not short of self-replicating technology. So perhaps this debate should be thought of in terms of what is the perceived cut-off scale where such nanotech assemblers become competitive?

Maybe we should better specify this. What range of products is your hypothetical replicator envisioned to produce at a cost and performance competitive with more specialized, centralized manufacturing factories?

I am not hijacking this thread any further, start a new one.

Up to you.