These comments are available on the web at http://www.foresight.org/hotnews/1999SFWkly.html
Peter Byrne recently claimed that quantum mechanics implies that molecular machines can't work. Oops.
The most generally accessible argument that molecular machines are feasible is to observe that biological molecular machines exist. Enzymes can copy DNA with remarkable fidelity. Bacteria are propelled forward by flagellar motors (enter "flagellar motor" into your favorite search engine for more information on this fascinating subject). Ribosomes are programmable molecular machines that make proteins using instructions read from mRNA.
In 1992, K. Eric Drexler published Nanosystems, which directly addressed the feasibility of molecular machines. He concluded that molecular machines were indeed feasible within the framework of generally understood and accepted physical law. As part of this analysis, he proposed some specific examples (a useful exercise: if molecular machines are feasible, then we should be able to design specific molecular machines). Byrne would have us believe that (a) Nanosystems did not address even basic quantum mechanical issues and that (b) the proposed machines would not in fact work for this reason.
Unfortunately, Nanosystems directly addresses the issues raised by quantum mechanics (section 3.2, "Quantum theory and approximations") and summarizes the basis of modern molecular mechanics: the Born-Oppenheimer approximation. This approximation is discussed in every introductory text on molecular mechanics, for it forms the basis of the field and the foundation for much of modern computational chemistry. A brief and somewhat simplified summary follows.
Atoms are made of nuclei and electrons. The nuclei are much more massive than the electrons (the nucleus of the carbon atom is over 20,000 times as massive as an electron). As a consequence, the quantum "fuzz" or uncertainty about the position of the nucleus is very small compared with the uncertainty in the position of the electron. It is therefore reasonable in most situations of practical interest (almost all of chemistry) to treat the nucleus as a classical point mass, subject to classical forces created by the quantum mechanical electron cloud. However, the shape of the electron cloud is determined by the positions of the nuclei (assuming the electrons are in the ground state: true for much but not all of chemistry, and true in particular for the molecular machines proposed in Nanosystems). Therefore, the forces acting between the nuclei are determined by the positions of the nuclei. This leads us to the key conclusion and the foundation of molecular mechanics: we can often accurately model the behavior of a molecular system using purely classical forces between the nuclei and without solving for the electronic structure. Under these conditions, analysis of molecular systems can be (and often is) treated using Newtonian mechanics.
As a consequence, accurate modeling of molecular systems (including proposed molecular machines) is entirely feasible, and has been carried out with great success on an incredible variety of systems. To take but a single example, the simulation of the planetary gear proposed in Nanosystems (see http://www.zyvex.com/nanotech/gearAndCasing.html) by Caltech's Materials and Process Simulation Center (http://www.wag.caltech.edu/) was based on standard and widely accepted principles.
Byrne's claim that "...respected nanotechnologists are reluctant to be critical of the Foresight Institute" because of delicate and subtle concerns is incorrect (and, actually, quite amusing). They are reluctant to be critical because they have no substantive criticisms of Nanosystems. Nanosystems was published in 1992. No significant errors have been found since it's publication. Not a bad record.
As for Scientific American: even they can't defend their story (which was published in 1996, not 1999). David Jones' book review (the only published argument against nanotechnology that they found worth citing) is grossly in error, a fact disputed by neither Scientific American nor Jones (see http://www.foresight.org/SciAmDebate/ your web article seems to have accidentally omitted any link to this very relevant discussion about Scientific American's story). Carl Feynman's letter to Scientific American said he was "dismayed" at their misuse of his father's essay on "Cargo Cult Science."
In summary: no substantive arguments have been advanced against the feasibility of molecular machines, including those proposed in Nanosystems. In contrast, there is now a (very large) body of evidence, both experimental and theoretical, which supports the feasibility of such machines. This is reflected in statements in Business Week, Time, Newsweek, and other publications (that did a better job of checking their facts than Byrne).
The questions the rest of the world is interested in are (a) how long will it take to develop nanotechnology and (b) how do we do it?