Comment 14 by kriton

Zeuglodon, I'm not sure, are you agreeing that bacteria can make genetical copies of themselves?

I think you're confused. Our aim is to identify what the unit of replication is. This requires that the unit is unambiguously capable of creating atomic-scale identical copies of itself. In fact, a bacterium's ability to split is itself a consequence of the replication process of this unit. Without genetic mechanisms to regulate and time the mitosis of the cells, you wouldn't get the division of cells in the first place. After all, a cell body deprived of its genes does not create copies of itself. It would simply sit there and do nothing.

Either way, I'm wondering about this r part. Is there empirical evidence that says those numbers 1/2, 1/4 and so on present themselves in nature?

Yes. The process of meiosis and genetic recombination in sexually-reproducing organisms is a basic fact of biology. Between parents and offspring, for example, the probability is on average 50% that a parent provided any particular gene. The 50% threshold applies mostly to those genetic loci that still vary within the population, for instance for variables that can, by the process of pleiotropy, also express themselves as eye colour. Either you got your eye colour gene from your mother or from your father. Averaged out, 50% of your genes will come from any one parent. Of course, there will be cases where a person may have more genes from one parent, but these are statistical anomalies to be expected as part of the recombination process. To the rest of the family tree, you simply implement an extension of this basic fact.

I read the Wikipedia article, but that only said... simulated populations of robots isn't really the stuff I'm interested in.

Why not? Simulations may not always map onto real life, but to dismiss them out of hand is to stack the deck. You forget that kin selection was not only an answer to a known phenomenon, but needed to be mathematically verified. In any case, the empirical evidence comes from, for example, genetic studies of animals like social insects.

After all, if we were to sequence my genes and my sisters genes, and compare the sequences, we would find that they were much more than 50% similar. Genes are conserved, and my parents would have lots of genes in common that are identical even if they are not related in the everyday sense.

If they are identical at the same genetic loci, then they are practically copies. In any case, the 50%, 25% and 12.5% relatedness figures are more usually invoked for the set of genes that are not included in a "species threshold". Say, for instance, 99.9% of human genes were identical. The remaining 0.1% would be where the relatedness figures would be present. It isn't hard to see why: obviously, if the genomes were 100%, genes would have nothing to compete with because whichever individual wins means the genes win anyway. You have to remember that the genetic competition is between alleles competing for the same spot on the genome, not genes fighting any old gene.

Suppose I have a half-sister, and we look at a gene where the sequence is identical between us. A gene has no clue were it came from, so from a genetic point of view, what does it matter if my half-sister got an identical gene from the same parent as I did, or from another parent?

This is one place where I think you're really confused. The altruism of an individual is a product of the combined phenotypes of several genes, each of which is either competing with rival alleles or had to compete with them at one point. Imagine two alleles competing for the same spot. One allele makes its host organisms behave more altruistically to its genetic relatives. The other does not. Under certain conditions - for instance, when organisms would benefit from staying in isolated groups - the altruistic allele would simply become more widespread in the population until it reached fixation (i.e. choked out its rival). The mathematical rules of kin selection theory would shape both the survival of further mutations and the direction of natural selection. The population would consist of increasingly more "intelligent" altruists who would be willing to do altruistic things when faced with certain situations. This might take many speciation events before it reached pitch perfection, and several versions at varying stages could be found during the same time interval (hymenopterans are a particularly rich source of examples, from lone digger wasps to huge colonies of bees and ants).