Robert Freitas

Robert Freitas is an American researcher who has spent three decades producing detailed engineering analyses of machines that do not exist. His four-volume Nanomedicine, of which two volumes have appeared, works through the physics of micron-scale devices intended to operate inside the human body: power, heat, navigation, sensing, communication, and what happens when a rigid manufactured object meets blood. The best known of the designs is the respirocyte, an artificial red cell. None of the devices has been fabricated, and Freitas has been consistent that his work is not a prediction that any of them will be.

§Overview

Freitas works in a genre he calls exploratory engineering: analysis intended to establish that a proposed device violates no physical law and to bound its performance if built, deliberately setting aside how anyone would build it. The output is a specification and a set of numbers rather than an experiment. Judged as engineering analysis the volumes are unusually thorough; judged as evidence that Medical nanorobots are coming, they establish nothing, because a design consistent with physics is not the same as a design that can be made. Most of the argument about his work is an argument about which of those two things it is being read as.

§Career

Freitas took degrees in physics and psychology at Harvey Mudd College and a law degree at Santa Clara University. His early work was in exobiology rather than medicine: a long treatise on the scientific study of extraterrestrial life and civilization, written in the 1970s, and a proposed scale ranking information-processing systems by their rate of computation per unit mass. With William Gilbreath he co-edited the report of the 1980 NASA/ASEE summer study on advanced automation for space missions, whose most discussed section is a design for a self-replicating lunar factory.¹

He began the Nanomedicine series in the mid-1990s at the Institute for Molecular Manufacturing in Palo Alto, where he is a senior research fellow, and spent the early 2000s as a research scientist at Zyvex, a Texas company then attempting instrument-based approaches to atomically precise fabrication.

§The Nanomedicine volumes

Volume I sets out basic capabilities: how a micron-scale device would draw power from blood glucose, shed waste heat without cooking the tissue around it, know where it is, and talk to its neighbours.² Volume IIA is devoted entirely to biocompatibility — what the immune system, the complement cascade and the protein corona that coats anything injected into blood would do to a rigid diamondoid object, and what surface chemistry might reduce it.³ Volumes IIB and III have been announced for many years and have not appeared.

The device studies are the part of the work that circulates. The 1998 respirocyte paper describes a pressurised gas vessel about a micrometre across that would carry far more oxygen per unit volume than haemoglobin does, competing with the oxygen carriers whose repeated clinical failures are described under Artificial blood.⁴ The microbivore is an artificial phagocyte that would trap a bacterium, digest it to amino acids and sugars, and discharge the residue, on a cycle timed in seconds.⁵ The clottocyte would stop bleeding faster than platelets; the chromallocyte would enter a cell and replace its chromosomes with corrected copies.

What the numbers are

Figures such as a respirocyte's oxygen-carrying advantage over a red cell are outputs of a physical model of a specified object. They are neither measurements nor forecasts, and they carry no information about when or whether the object could be made. The distinction is stated plainly in the papers themselves and lost in most secondary accounts of them.

§Self-replication and global ecophagy

Freitas has also worked on the theory of machines that copy themselves, a thread running from the 1980 NASA study to the book-length review he wrote with Ralph Merkle in 2004, which catalogues the physical and mathematical literature on kinematic self-replication and proposes a classification for it.⁶

That interest produced the most-cited thing he has written outside the medical designs. His 2000 analysis of global ecophagy asked what would physically limit a runaway population of biomass-consuming replicators — the Grey goo scenario — and concluded that waste heat binds before materials do: converting the biosphere at the rates the scenario requires would release enough energy to destroy the machinery doing it, and would produce a thermal signature detectable long beforehand. He treated this as grounds for monitoring and for design restrictions on replicating systems rather than for dismissal.⁷ It is a rare case of a technology's proponent producing the quantitative risk analysis, and it is cited in the Existential risk literature by people who reject the rest of the programme.

§Diamond mechanosynthesis

Everything in the volumes assumes a manufacturing capability that does not exist. The devices are specified in stiff diamondoid components with atomically precise surfaces, which only the positional chemistry proposed by Eric Drexler could produce. Freitas's response has been to work on the smallest tractable piece of that problem: mechanosynthesis of diamond, in which a positionally controlled tip places carbon atoms on a diamond surface one or two at a time.

With Merkle he has published computational studies of candidate tooltips and reaction sequences, evaluated with quantum-chemistry methods, and the two have coordinated a loose collaboration aimed at a first experimental demonstration. A patent covering a tool for positional diamond mechanosynthesis issued in 2010, and the 2009 Feynman Prize for theory recognized the tooltip work. No experimental demonstration of controlled diamond mechanosynthesis has been reported, which is the same gap that has stood open since Richard Smalley's objections to Drexler in the early 2000s.

§Cryonics

Freitas has argued at length that cell-by-cell repair of cryopreserved tissue is a well-posed engineering problem, publishing a book-length treatment through the Alcor Life Extension Foundation in 2022 that works through failure modes in preserved tissue and the operations a repair device would have to perform.⁸ It is the fullest statement of the biological revival route described under Cryonics, and it inherits that route's dependence on machinery nobody has built. That dependence is part of why others in the field prefer the informational route, in which the preserved brain is scanned rather than repaired: the programme pursued under Brain preservation and Whole brain emulation. Alcor's own literature endorses neither route.

Whether the analysis can be checked

Supporters hold that a quantitative design that survives physical scrutiny is a stronger claim than an assertion that revival is impossible, and that the burden has not been met on the other side. Critics hold that an analysis of surfaces, tissues and failure modes that have never been produced cannot be checked at all, and that unfalsifiable engineering is not engineering.

§Reception and legacy

Chemists and clinical researchers have largely not engaged with the programme. The medical successes at the nanoscale arrived by other routes entirely: Lipid nanoparticles and antibody conjugates from formulation chemistry and immunology, DNA origami devices with no onboard power or computation, magnetically steered particles that are steered from outside rather than autonomous. Freitas is cited far more in futurist, transhumanist and cryonics writing than in the clinical Targeted drug delivery literature, and a substantial part of his output appeared in books from a small press, in institute reports, or in journals associated with the movement rather than in mainstream biomedical venues.

The case for the work is that it is specific enough to be wrong. It names dimensions, materials, power budgets and cycle times, which is more than most speculation about medicine's future offers, and the biocompatibility volume takes the strongest objections seriously instead of waving at them. The case against is that specificity about an object nobody can make is not the same as evidence, and that thirty years of increasingly detailed designs with no fabrication attempt is the signature of a research programme that has substituted analysis for experiment. Both readings survive the record as it stands, which is why the argument has not moved much since Nanomedicine Volume I appeared.

§See also

• Respirocytes
• Medical nanorobots
• Molecular assembler
• Eric Drexler
• Grey goo
• Cryonics
• DNA nanotechnology
• Artificial blood

§References