The idea is perhaps best explained by a diagram:

The large, yellow regions represent phenotypes.

G1 is the primary genetic substrate, and G2 is the secondary one.

Arrows within organisms indicate paths of genetic expression.

A simple organism with genetic substrate G1 produces substance G2 as a component of its metabolic processes.

G2 is inherited - and comes to carry heritable information.

Gradually, G2 displaces G1 as the primary genetic material for the organism.

This effect may be responsible for driving the adoption of the secondary genetic substrate - if this allows more convonient access to superior phenotype-construction technologies.

Consequently, it seemed likely that the original genetic material was related to the modern genetic machinery such that the two could be linked by a series of gradual changes.

The mechanism of genetic takeover allows for the possibility that a secondary genetic material can arise not as a modification of the primary one, but rather from molecules synthesized under its control.

The resulting liberation allowed Cairns-Smith to propose that the first living organisms were in fact extremely different in their construction from the type of organism we are now familiar with.

In fact, he proposed that our ultimate ancestors were actually clay mineral crystals.

Regardless of whether his theory of crystalline ancestry is correct, it still seems very likely that genetic takeovers were involved in the origin of today's genetic machinery.

Initially the second genetic material is not critical to the organism, and may thus be subject to variation and natural selection - without changes necessarily resulting in a non-viable organism.

A sceptic might suggest that this is just as difficult as the secondary genetic material arising from scratch out of inorganic material - but this is not the case.

The secondary substrate can be a thermodynamically unlikely object, whose very existence would be implausible in the absence of an existing evolutionary process and natural selection. This alone means the primary genetic material is important.

In much the same way that a spaceship traveling into space requires a booster rocket to get it out of the atmosphere, so early life required a boost from a different mechanism to those which act as genetic machinery in modern organisms. This is because the modern genetic mechanisms are too sophisticated to plausibly form under pre biotic conditions.

In the same way that a spaceship discards its booster rocket as dead weight, so the initial genetic machinery appears to have been completely discarded now that a superior, high-technology alternative has been evolved.

The primary genotype is responsible for providing an environment in which the second genome evolves.

Since the components in the original phenotype are replaced gradually, approximately on a one-at-a-time basis, there is likely to be a functional relationship between the original components and their replacements.

			The Baldwin effect analogy
			The information-transfer con be usefully compared to that which occurs with the Baldwin effect. In the Baldwin effect, learned behaviors become fixed genetically (over many generations). Information acquired by organisms in the course of their lifetimes eventually finds its way into the gene pool. With the Baldwin effect, the learned behavior creates an environment in which genetic evolution (leading towards the organism displaying the learned behavior instinctively) can occur more rapidly. Similarly, in a genetic takeover, the primary genome creates an environment in which evolution of the secondary genome can occur more rapidly in certain directions.
			Endosymbiosis analogy
			There is another interesting analogy, with the process of endosymbiosis: In endosymbiosis information can flow between the genomes of the organisms involved, as they "compete" with one another to perform each other's functions. Unfortunately, horizontal gene transfer between endosymbionts and their hosts can occur by the direct action of their DNA being physically mixed together, partly obscuring the effect. It appears that a likely result of endosymbiosis will be one of the participants completely absorbing the functionality of the other, through a mixture of horizontal gene transfer and copying of the supplied technologies.

It is important to emphasize the fact that there can be ways to transfer information from the old genetic medium to the new one - even if there is no specific apparatus that can read the old medium and write to the new one.

Consequently, the idea that the new life is (genetically speaking) the mortal enemy of the germ line of its ancestors is incorrect.

In a very real sense, some of the genetic information of the ancestors can survive in their descendants.

However there are currently "various indications" that a genetic takeover is imminent:

			The rise of human culture
			An examination of the communication of high-fidelity information between organisms and their descendants reveals that almost all the information transmitted over the history of life on earth has been via nucleic acids. However very recently, various new methods of transmitting information to descendants has arisen. These have the high-fidelity of replication necessary to be able to support evolutionary processes, and are capable of transmitting large volumes of information. These mechanisms are usually though of as transmitting "cultural" information - but there is no fundamental distinction between genetic information and cultural information in this context. Until human beings came on the scene "cultural" transmission of information existed - but was very limited in volume. For example, a bird's offspring may inherit the songs of their parents - but probably only a small number of generations will pass before it is not possible to identify the parents from the songs of their descendants. By contrast, human beings have brought with them the written word and - more recently - books, CDs, DVDs and other optical, electro-magnetic and electronic storage media. The result is a large volume of heritable, high-fidelity information which is not transmitted through nucleic acids. The ultimate effect of these types of new information storage media on biological evolution could be extremely far-reaching.
			Problems with the old medium
			Nucleic acids are a form of molecular nanotechnology - as such they are a compact, concise and surprisingly reliable form of information storage. It has been sufficient to act as the primary information-storage medium for life for the past four billion years. However, currently, new storage media are frequently used in preference to DNA for a number of reasons: Accessibility Today reading and writing information in nucleic acids is a laborious rigmarole. Sequencing a single human genome took years to complete. Write access Nucleic acids were "designed" to be replicated - but not written to. Modern information storage demands the ability to be easily modified when circumstances dictate this is appropriate; Random access Nucleic acids are long, stringy molecules. While it may be possible to grab them at a specific point, they're not designed with this in mind. They are like a tape - and lack the random access features you would get with (say) a disc. in order to get at the information you're looking for it is often necessary to do a lot of fast-forwarding or rewinding. Speed Even copying a single nucleic acid chain is a tedious and slow process. In principle it might be possible to break it up and parallelise the process - but for rapid access to the information, it is not currently an attractive option; Only usable in a limited range of environments Nucleic acid chemistry is effectively dependent on the presence of liquid water, and a number of (admittedly fairly common) organic compounds. There is no fundamental reason why life should no exist at very low or high temperatures; Dimensionality Nucleic acids are essentially one-dimensional molecules. In three dimensions they tend to coil up. Random access considerations mean that future storage media are likely to be two- and three-dimensional. It's true that - in many respects, if you compare them to DNA - most modern storage media appear to be sub-standard - large, wasteful, and ugly. However, having said this, nucleic acids are themselves a hacked-together, evolved solution - which is more likely to be a local optimum than an ideal method of information storage for organisms. It is virtually unthinkable that modern design and technology will eventually fail to provide storage media alternatives that are superior in almost every useful way to DNA.
			Advantages of the new media
			Human cultural information can be expressed in phenotypes in many diverse ways. It can control the design of automobiles, or result in animated images or music. Its volume can be very large, and it can be replicated with high fidelity. Human culture is subject to conventional Darwinian evolution - but is also subject to Lamarckian processes, and to intelligent design. These demand an information-storage medium with an ability to write data conveniently. Today's electromagnetic, magnetic and optical storage devices seem likely to be replaced by nanotechnological devices - or possibly mechanisms based on individual atoms in crystalline lattices. There may well eventually be advantages to the new media in terms of stability, expense, access speed and the ability to modify the information. Some of these were not significant criteria when the current genetic machinery was selected, but they are important in storage media today.
			New phenotype technology
			If the original genetic takeovers were driven by the ability to direct new technologies (e.g. proteins), it seems likely that a modern genetic takeover will be driven by the same type of effect. DNA is currently only used to influence phenotypes by directly controlling the synthesis of some twenty amino-acids. The resulting protein machinery is very flexible and can direct a large number of other types of chemical reaction. Despite this, nature has failed to master some relatively simple and obvious mechanical technologies - for example, notoriously nature makes very little use of the wheel. At the moment it appears that a whole raft of new technologies is driving the adoption of new genetic materials. Most programmable computers are currently constructed from specifications that are not stored in DNA. Also, DNA appears ill-placed to directly control the synthesis of the important Fullerene molecules. If a single area of chemistry was to be nominated as signaling the end of the DNA/protein era, I would nominate Buckminsterfullerene, nanotubes - and their chemical relatives. The significance of these molecules may be primarily structural - or they may eventually come to play important information-processing roles. Either way it appears that the relatives of diamond and graphite are likely to be important components of the organisms of the future because of their unique chemistry.

DNA is too limited in terms of its means of phenotypic expression, and in its ability to offer random access and read/write access to its data rapidly.

While it may persist for an extended period, its eventual complete replacement appears likely.

Certainly the era of its virtual monopoly on heritable information is already over.

Note that as far as I know, the possibility of a modern genetic takeover was first suggested by the robot builder, Hans Moravec.

The fact that the takeovers near the start of life are very distant from us has obscured their significance as evolutionary events.

Whenever evolution develops new information-storage devices there will be an immediate "pressure" for genetic information to migrate into them. This process will continue for as long as superior information-storage technologies exist. It seems likely that the next genetic takeover will not be the last.

The information storage aspect of the medium will probably become increasingly abstracted from the replication mechanism, and from the phenotypic expression of the information.

In the light of the fact that we may now be facing the first genetic takeover for several billion years, some interest in the mechanism behind them would appear to be appropriate.