Eric Drexler’s “Radical Abundance”
There is an issue with the viscosity of water.
We are at the edge of where the rules change.
Temperature is a strained metaphor here.
Friction forces have nowhere ‘lower’ to deposit dissipated energy.
A pile of models of water molecules has an angle of repose meaning that it can support shear stresses while in a static configuration.
I think that it is conventional to say that shear stress in a static pile of water vanishes because individual water molecules can ‘tunnel’ to lower energy states unlike the pile of models where friction rules differ.
I think that nano-bearings may enjoy similar low friction benefits.
Gears must be designed to minimize such tunneling events lest they ‘slip a cog’.
However the mathematical bridge that links this deepest physical model to everyday physical predictions charges an impossibly steep tool in computational complexity and cost, and so scientists and engineers instead use shallower models of physical law and avoid using levels deeper than they need.
This observation is true, important and seldom reported.
I believe that there remain surprises here.
Indeed it may be that the ‘shallower models’ are more nearly correct than the deep models.
The shallower models have been better tested with observation.
The toll serves as too convenient an excuse for not holding the deep models up to observation.
I do not see that this dilemma threatens Drexler’s thesis.
It may threaten dreams of quantum computing.
Drexler presents a long and interesting discussion on differences between science and engineering.
There are several interesting differences, some obvious but new to me.
The notion of recursive levels of assembly seems sound in most cases but there is a question of how to retain atomic precision at larger assembly stages for smooth interfaces.
We adjoin two decimeter units at a smooth interface, we are faced with aligning about (109)2 bonds.
This might be called the fastener problem.
On page 184 Dexler discusses engineered DNA structures that I think I have seen described as RNA structures.
I don’t know how important it is to distinguish RNA from DNA.
Page 192, concerning steriotaxis:
Practical efforts must start at today’s level (already far above what was once the ground), and then build higher levels, layer by layer, with blocks made of available materials.
Identifying suitable architectures is a task for exploratory engineering that addresses a series of fabrication constraints; each level of capabilities—each emerging class of systems—must be implemented using capabilities made available by the level below.
On page 218 there is a short section, “The Road Ahead in Molecular Engineering”, which I take to be the book’s main message and for which the rest of the book provides history, context and justification.
Implicit there are the conflicts of vision that explain much of the recent history.
Drexler posits a long road ahead without gaps with intermediate payoffs which leads to APM technology.
The road is now only vaguely foreseen.
Clearly there is some doubt about the existence of this road, or the ability to find it.
Page 233: Drexler discusses agriculture without broaching a post biotic era.
One battle at a time.
Drexler also avoids the AI contingency.
Page 244: Drexler characterizes the alternative paths towards APM.
Perhaps Kauffman’s term “the adjacent possible” would be useful for its connotations here.
Page 254:
I leave to readers to consider potential winner-take-all outcomes and whether they would want to live with the aftermath, almost certainly not sitting at the top.
As for what disproportionate winners might claim is their right, by the rules of one game or another, every such claim must be placed in perspective.
Each advance in technology adds only an increment to a tower of achievements built over millennia by millions of human hands and minds, and that tower is, by practice and age-old principle, the common heritage of humankind.
Each generation contributes, and no living person can rightly claim ownership of more than a tiny and transient part.
I worry about this even without APM.
The market value of unskilled (and much skilled) labor is decreasing perhaps even in absolute terms.
I also worry about the need to feel useful, preferably by being useful.
APM makes the problem extreme.
Chapter 17 is about adversarial APM—military uses.
A historical analog is crypto which is highly technical and reliant on secrecy.
Current NSA news (2013) could easily be rewritten for a hypothetical secretive government APM project.
Drexler has not discussed flow of APM technology between institutions and nations, nor a role for patents.
What are prospects for reverse engineering in APM?
On page 267 Drexler addresses the reasons why states would want to restrict some aspects of APM, mainly to protect against rogue non-state actors—terrorists.
Where to establish such a fence must wait until the outline of the path to APM emerges.