While I started out to write a topical piece on current controversies, biomedical or economic, I decided others were doing a more than adequate job of venting hot air without me. Instead, one fallacy which consumes a great deal of scientific effort stands out as in need of explanation. It is at least as old as the idea of finding a "missing link" in evolution, if not the "great chain of being" (scala naturae).
We have, on one hand, Virchow's principle that organisms, from cells on up, originate only from other organisms ("omnis cellula e cellula", medical Latin which would baffle an ancient Roman thinking about 'little rooms'). This was certainly an advance over spontaneous generation (for those unwilling to experiment with entire planets and wait billions of years -- scale matters). The problem we have is the difficulty of actually being there at the right place and moment when a particular organism gives birth to another organism of special interest. In nearly every case, we are not going to observe the precise event, but a (hopefully) closely-related event which allows us to infer that the event in question actually took place. Even the most pragmatic researcher must venture into the realm of multiple competing hypotheses.
We want to see the entire process as a single chain leading inevitably to the pinnacle and center of the Universe, our personal selves. The dismal truth is that most lines of descent don't lead anywhere. Even when you look at historical records of human ancestry, considering only those who survived long enough to acquire names, there is a serious problem -- roughly a third of all lines of descent end in each generation. Those who will pass their genes through two more generations amount to roughly 4/9 of those living at any given time, less than half. More generations mean lower odds, and this is for a species which places high value on individual survival. Had we been talking about blue crabs (callinectes sapidus) the odds would be more like 100 million to one against any individual egg surviving long enough to reproduce.
Where we want to find a single chain in evolution, on close examination we keep finding not just vines with a few tendrils branching to the side, or even tall trees, but bushy shrubs with large numbers of twigs that don't rise far at all from their roots. The rationale for this distorted emphasis is that the survivors are the only ones that matter. True, in one sense, but misleading. The environment in which evolution takes place is largely determined by living things, and most of these will be just as unlikely to produce surviving lineages as described above. Pathogens form a major part of the fitness landscape, and pathologically lethal pathogens which will not be around for the long term are a major feature with which survivors must cope. No matter how well adapted you are to some future environment if you do not get there adaptation is irrelevant.
Much of the above primarily concerns organisms which use sexual reproduction. This, despite individual attempts to introduce variety, is a pretty systematic process. I would liken the sexual rearrangement of genes to the cut and shuffle of a fair card game. Viruses don't play by these rules. They are quite willing to keep cards up their sleeves, and mix cards from standard and pinochle decks, if not Tarot decks and Mahjong tiles. They are even willing to tear cards apart and paste them together to create new cards. Their game may not be fair, but we don't make the rules, and have had little success enforcing our own.
Even when it is possible to find the single lineage devoutly sought by biologists those numerous short branch points leading away will still exist. If the main lineage benefits from maintaining a low profile and slow replication, all the easy leads to its existence will come from those tiny twigs which break cover. This may likely cost them their existence, but these are also the cases in which they may take the host with them in a recognizable pathology.
Far more likely than a simple chain of one fixed genome is a small cloud of very similar genomes which pass considerable genetic information back and forth via recombination with close relatives in the same tissues and cells. In this case our metaphor of a single chain or string must be replaced with a kind of braid or cable of vines, each with many tendrils. The problem of inference here is to reconstruct the essential parts of that structure from the divergent sequences which arise through errors in normal operation. Successful replication in healthy contexts won't give you the rapid exponential increase in sequences biological research has grown to depend on.
Even if the viruses could be said to know exactly what they are doing researchers do not. The resulting reasoning is more akin to inference in quantum mechanics than conventional pragmatic inference. The distinction between a classical mixed state and a true quantum superposition of states is academic with so much uncertainty. You would need far better tools than we have to reach the point where the difference is apparent. Confusion is a fundamental aspect of the field, and claims of certainty are likely spurious.
The literature on quantum mechanics illustrates very well the length of time it took to deduce even simple interactions in the presence of fundamental uncertainty. Progress in understanding genetics and molecular biology has been painfully slow. Both subjects are fairly esoteric, but there is one example of such reasoning close to the fingertips of everyone reading this.
Modern computer networks are prominent examples of distributed processing. While these machines are designed to individually be as close to deterministic as possible the collective behavior is not nearly as easily predictable. Getting a global picture of the state of such a system is virtually impossible. Even if this were not so, reasoning about the state of the system must constantly proceed with incomplete information. (As it happens, I once saw a room full of PhDs arguing vociferously about what one part of a distributed system "knew" about the state of another part.)
My conclusion is that, despite many determined attempts to impose order and explain everything, we humans simply aren't built to deal with this form of reasoning under uncertainty. I have seen bright people complicate simple systems into messes where it was almost impossible to eliminate fundamental errors -- while thinking, and reporting, that they were constantly improving the system. I have seen tiny systems which functioned correctly replace huge complicated systems which sort of worked, most of the time. Despite many advances I have not seen a decrease in examples of fundamentally flawed design. (Perhaps this is due to clever marketing by companies which have figured out how to profit from selling bugs. Read the End User License Agreement carefully to find out how little they promise their product will do, or avoid doing.) More and more, development of computer software resembles biological evolution instead of any rational process. Individual parts of design may employ the most sophisticated reasoning imaginable, but the end result continues to be opportunistic survival of the fittest in an environment dominated by chance.
The main difference in reasoning about molecular biology is greater complexity, higher parallelism, higher uncertainty and very limited human control of design. Programmers call logical errors "bugs". Biologists may talk publicly about hunting microbes, but they too sometimes refer to their quarries as bugs. (The linguist roots of this practice run very deep, back into the world of animism, where human beings felt they really understood what was going on. Control was simply a matter of figuring out which spirits to propitiate or exorcise.)
When intelligent human beings have been banging their heads against a problem for a long time without solving it two things to check are assumptions, stated or unstated, and ignored anomalies. The collection of biological anomalies tied to assumptions of simple viral lineages is getting rather large.
If you are determined to avoid learning anything from anomalies you can get into real intellectual contortions. One aspect of current problems is the extraordinary preference contaminants seem to show for human cell lines derived from neoplasms of unknown etiology. How do they know? Because all reputable virologists seem to start from the assumption these are mouse viruses we have an hypothesis of recombination origin based on two preexisting sequences found in some laboratory mice.
The problem with this is that they were found in a database, only one sequence was found in the actual samples leading to the claimed recombination. This leads to the statement that the other must have been present, but below the detection threshold. The idea that a complete sequence requiring no recombination might have been below detection threshold is rejected on grounds that appear scientifically obscure.
The critical feature of the resulting sequence is a deletion, plus the presence of the two overlapping subsequences. Deletions of any size can take place very easily. If they remove elements which prevented rapid replication, the sudden appearance of detectable virus is a perfectly natural and predictable outcome. No special pleading is required.
When another sequence without the deletion is isolated from a prostate tumor on the other side of the world, this is described as another, unrelated virus. Clearly, these researchers are mainly interested in following leads to a point at which their significance can be dismissed. Meanwhile, prostate cancers continue to appear in increasing numbers of men.
A long history of similar contamination battles has left veteran virologists grizzled and scarred. The remarkable preference of contaminants for particular human cell lines, tissues and organs, and a distaste for anything resembling the animal host assumed responsible for the contamination has yielded a string of anomalies over a period of four decades.
I'll offer a rather trivial observation, successful pathogens regularly modify their environment to benefit themselves. If particular types of contaminants favor particular pathological cell lines the idea that a closely-related pathogen is behind the pathology should be given added weight.
We have, on one hand, Virchow's principle that organisms, from cells on up, originate only from other organisms ("omnis cellula e cellula", medical Latin which would baffle an ancient Roman thinking about 'little rooms'). This was certainly an advance over spontaneous generation (for those unwilling to experiment with entire planets and wait billions of years -- scale matters). The problem we have is the difficulty of actually being there at the right place and moment when a particular organism gives birth to another organism of special interest. In nearly every case, we are not going to observe the precise event, but a (hopefully) closely-related event which allows us to infer that the event in question actually took place. Even the most pragmatic researcher must venture into the realm of multiple competing hypotheses.
We want to see the entire process as a single chain leading inevitably to the pinnacle and center of the Universe, our personal selves. The dismal truth is that most lines of descent don't lead anywhere. Even when you look at historical records of human ancestry, considering only those who survived long enough to acquire names, there is a serious problem -- roughly a third of all lines of descent end in each generation. Those who will pass their genes through two more generations amount to roughly 4/9 of those living at any given time, less than half. More generations mean lower odds, and this is for a species which places high value on individual survival. Had we been talking about blue crabs (callinectes sapidus) the odds would be more like 100 million to one against any individual egg surviving long enough to reproduce.
Where we want to find a single chain in evolution, on close examination we keep finding not just vines with a few tendrils branching to the side, or even tall trees, but bushy shrubs with large numbers of twigs that don't rise far at all from their roots. The rationale for this distorted emphasis is that the survivors are the only ones that matter. True, in one sense, but misleading. The environment in which evolution takes place is largely determined by living things, and most of these will be just as unlikely to produce surviving lineages as described above. Pathogens form a major part of the fitness landscape, and pathologically lethal pathogens which will not be around for the long term are a major feature with which survivors must cope. No matter how well adapted you are to some future environment if you do not get there adaptation is irrelevant.
Much of the above primarily concerns organisms which use sexual reproduction. This, despite individual attempts to introduce variety, is a pretty systematic process. I would liken the sexual rearrangement of genes to the cut and shuffle of a fair card game. Viruses don't play by these rules. They are quite willing to keep cards up their sleeves, and mix cards from standard and pinochle decks, if not Tarot decks and Mahjong tiles. They are even willing to tear cards apart and paste them together to create new cards. Their game may not be fair, but we don't make the rules, and have had little success enforcing our own.
Even when it is possible to find the single lineage devoutly sought by biologists those numerous short branch points leading away will still exist. If the main lineage benefits from maintaining a low profile and slow replication, all the easy leads to its existence will come from those tiny twigs which break cover. This may likely cost them their existence, but these are also the cases in which they may take the host with them in a recognizable pathology.
Far more likely than a simple chain of one fixed genome is a small cloud of very similar genomes which pass considerable genetic information back and forth via recombination with close relatives in the same tissues and cells. In this case our metaphor of a single chain or string must be replaced with a kind of braid or cable of vines, each with many tendrils. The problem of inference here is to reconstruct the essential parts of that structure from the divergent sequences which arise through errors in normal operation. Successful replication in healthy contexts won't give you the rapid exponential increase in sequences biological research has grown to depend on.
Even if the viruses could be said to know exactly what they are doing researchers do not. The resulting reasoning is more akin to inference in quantum mechanics than conventional pragmatic inference. The distinction between a classical mixed state and a true quantum superposition of states is academic with so much uncertainty. You would need far better tools than we have to reach the point where the difference is apparent. Confusion is a fundamental aspect of the field, and claims of certainty are likely spurious.
The literature on quantum mechanics illustrates very well the length of time it took to deduce even simple interactions in the presence of fundamental uncertainty. Progress in understanding genetics and molecular biology has been painfully slow. Both subjects are fairly esoteric, but there is one example of such reasoning close to the fingertips of everyone reading this.
Modern computer networks are prominent examples of distributed processing. While these machines are designed to individually be as close to deterministic as possible the collective behavior is not nearly as easily predictable. Getting a global picture of the state of such a system is virtually impossible. Even if this were not so, reasoning about the state of the system must constantly proceed with incomplete information. (As it happens, I once saw a room full of PhDs arguing vociferously about what one part of a distributed system "knew" about the state of another part.)
My conclusion is that, despite many determined attempts to impose order and explain everything, we humans simply aren't built to deal with this form of reasoning under uncertainty. I have seen bright people complicate simple systems into messes where it was almost impossible to eliminate fundamental errors -- while thinking, and reporting, that they were constantly improving the system. I have seen tiny systems which functioned correctly replace huge complicated systems which sort of worked, most of the time. Despite many advances I have not seen a decrease in examples of fundamentally flawed design. (Perhaps this is due to clever marketing by companies which have figured out how to profit from selling bugs. Read the End User License Agreement carefully to find out how little they promise their product will do, or avoid doing.) More and more, development of computer software resembles biological evolution instead of any rational process. Individual parts of design may employ the most sophisticated reasoning imaginable, but the end result continues to be opportunistic survival of the fittest in an environment dominated by chance.
The main difference in reasoning about molecular biology is greater complexity, higher parallelism, higher uncertainty and very limited human control of design. Programmers call logical errors "bugs". Biologists may talk publicly about hunting microbes, but they too sometimes refer to their quarries as bugs. (The linguist roots of this practice run very deep, back into the world of animism, where human beings felt they really understood what was going on. Control was simply a matter of figuring out which spirits to propitiate or exorcise.)
When intelligent human beings have been banging their heads against a problem for a long time without solving it two things to check are assumptions, stated or unstated, and ignored anomalies. The collection of biological anomalies tied to assumptions of simple viral lineages is getting rather large.
If you are determined to avoid learning anything from anomalies you can get into real intellectual contortions. One aspect of current problems is the extraordinary preference contaminants seem to show for human cell lines derived from neoplasms of unknown etiology. How do they know? Because all reputable virologists seem to start from the assumption these are mouse viruses we have an hypothesis of recombination origin based on two preexisting sequences found in some laboratory mice.
The problem with this is that they were found in a database, only one sequence was found in the actual samples leading to the claimed recombination. This leads to the statement that the other must have been present, but below the detection threshold. The idea that a complete sequence requiring no recombination might have been below detection threshold is rejected on grounds that appear scientifically obscure.
The critical feature of the resulting sequence is a deletion, plus the presence of the two overlapping subsequences. Deletions of any size can take place very easily. If they remove elements which prevented rapid replication, the sudden appearance of detectable virus is a perfectly natural and predictable outcome. No special pleading is required.
When another sequence without the deletion is isolated from a prostate tumor on the other side of the world, this is described as another, unrelated virus. Clearly, these researchers are mainly interested in following leads to a point at which their significance can be dismissed. Meanwhile, prostate cancers continue to appear in increasing numbers of men.
A long history of similar contamination battles has left veteran virologists grizzled and scarred. The remarkable preference of contaminants for particular human cell lines, tissues and organs, and a distaste for anything resembling the animal host assumed responsible for the contamination has yielded a string of anomalies over a period of four decades.
I'll offer a rather trivial observation, successful pathogens regularly modify their environment to benefit themselves. If particular types of contaminants favor particular pathological cell lines the idea that a closely-related pathogen is behind the pathology should be given added weight.
