Not only is Hamilton's Rule not a rule, it isn't even a strong suggestion.  The relationship c < rb (1), doesn't begin to qualify as a meaningful description of anything.

The first problem one encounters is that all the variables are highly subjective.  It is also problematic in describing what a cost and benefit actually is in a quantifiable way.  In a book describing the research of white-fronted bee-eaters, this helpful piece of information was provided:

"For example, imagine an individual with nine units worth of aid that it can dispense to relatives. Suppose that this individual consistently interacts with one sibling (r = 0.5) and one uncle (r = 0.25).  Since siblings share an r value twice as great as that between uncle and nephew, the proportional altruism model predicts that six units of aid should be dispensedto the sibling and three units of aid should be dispensed to the uncle."

The Altruism Equation: Seven Scientists Search for the Origins of Goodness by Lee Alan Dugatkin

Well, who would've thought that cost and benefit could be broken up into such convenient units.  More absurdly, this description suggests that reproductive cost/benefit is a discrete enough value that it can be quantified in such a way.

Similarly J.B.S. Haldane's famous comment that "I would lay down my life for two brothers or eight cousins" (based on r = 0.5 for brothers and r = 0.125 for cousins) clearly misses the point completely.

Haldane is invoking cost/benefit as life or death, while the definition provided by Hamilton is reproductive cost/benefit.  There is nothing in this "rule" which is so absolute and more importantly, the values being discussed can't accurately be quantified. In addition, costs and benefits aren't predictable, so minimally, there must be a probability associated with any cost or benefit to determine the possible outcomes.

Consider what it means to incur a reproductive cost.  Does this cost go down if you've already reproduced?  How much by how many offspring?  After all, this plays directly into the issue of grandparents caring for their grandchildren which predicts that grandparents should be more inclined to help their children rather than their grandchildren.

How is a reproductive benefit conveyed, especially if the animal is too young to breed?  How does a parent weigh their own reproductive potential (r = 1.00) against that of their offspring (r = 0.5)?

All of these problems are exacerbated by the definition used for altruism, which states that "altruistic acts are those that benefit the recipient but harm the actor".  This definition places us in the absurd position of only considering acts that result in definite harm and benefit.  So exactly the same action would be excluded if it didn't result in harm, which is effectively "cherry-picking" the data.  The oft sited example of giving alarm is used as if the death of the alarm-giver is a foregone conclusion.

Once again, we also have to put up with the abuse of language by hearing how helping family members is a "selfish" act.

"Helper birds postpone opportunities to breed in order to help family members," says Cornell University biologist Stephen T. Emlen. But the behavior is genetically "selfish" because it helps young relatives survive, thereby perpetuating the family's genes, Emlen says

This statement clearly refutes Hamilton's Rule since it predicts that the original actor's cost must be LESS than the benefit conveyed based on the degree of relatedness.  There is no benefit that can be conveyed to the family members that is greater than the benefit to the original bird if it reproduces.  Clearly there is an additional mechanism at work that elicits such behavior, but it is not examined.

An additional quote provides even more insight from Nature:

"We report here that older male white-fronted bee-eaters (typically fathers) actively disrupt the breeding attempts of their sons, and that such harassment frequently leads to the sons joining as helpers at the nest of the harassing father. Calculation of fitness costs and benefits to the various participants helps to clarify both why parents engage in such 'recruitment' behaviour and why sons frequently do not resist."

http://www.nature.com/nature/journal/v356/n6367/abs/356331a0.html

The following quote even more strongly suggests a behavior aspect to the white-fronted bee-eaters that may revert to a "helper" role when their own reproductive success may have become compromised.

"When a male loses its mate and young, perhaps to a marauding snake or other predator, it frequently will rejoin its parents or brothers and help feed their new brood."

http://ngm.nationalgeographic.com/geopedia/Bee_Eaters

So it appears that the helping behavior isn't quite as altruistic as first conveyed, but appears to be more coerced. This clearly suggests a behavioral model rather than a genetic trait.

Perhaps even more telling is the following:

"Nearly half of unpaired bee-eaters Emlen studied, however, did no volunteer work at all, and those that did were extremely picky about whom they served. "In 90 percent of the cases where helpers had a choice," he says, "they chose to help the closest available relative."'

"A nesting bee-eater will pitch any egg that shows up in her nest before she begins laying her own. Once she has started, though, she will accept interlopers' eggs."

"The daughter has the inside track on how to sneak in an egg," says Emlen. "The extra burden on the nesting pair from that one egg is great, but if the parasite is a daughter, they at least get her help."

http://www.nationalwildlife.org/nationalwildlife/article.cfm?articleID=1232&issueID=90

So, it appears that we have Hamilton's Rule being applied to about 50% of the birds.  In addition, it appears that the birds have no ability to recognize genetic relatedness.  It is clear that the best recognition that can occur is that of visual cues before egg-laying begins. Given that often the daughter actually introduces outside eggs into the nest, I can't even begin to fathom how Hamilton's Rule applies.

In truth, it appears that this entire study illustrates that under harsh conditions, there is more benefit in cooperating, since an opposing interpretation could readily state that if the males were too insistent on breeding, then it could result excessive inter-family competition which risks all members.

Even answering the basic question of the conditions necessary to observe altruism are missing from these defintions.  For example, is it ever observed between species?  Does it only occur with social animals?  Does it require animals that actively rear their young?

Another example relates to bee colonies where it is considered that worker's altruistically give up their reproductive capability in deference to the queen.  However, if the queen dies, then workers will begin to take on the role of egg-laying to replace the queen.  If a new queen is introduced within several hours of the original death, the works will cease their activities and revert back to normal behavior with a genetically unrelated queen (which is a bit problematic for Hamilton's Rule).

In fact such behavior by bees is rather intuitive, when we consider that even in human society the loss of a leader is considered far more dangerous than putting someone in that position that was not duly elected.  So instead of assigning a fabricated "r" value, it is actually more realistic to suggest that the survival of the colony requires a queen and so workers behaviors will be triggered to begin competing for that position.  If a new queen is introduced, the hive is essentially safe and such activity no longer needs to occur.  It has no relation to the relatedness of the bees, but rather the survival of the hive itself that dictates events.

More problematic is the value r which requires the ability to recognize genetic relatedness.  Since this is highly unlikely and has never been observed to occur, the "Green Beard Effect" has been postulated as a means around this.  However, this simply creates a bigger problem since one cannot argue about the genetic benfits of kin recognition/selection and then postulate that as long as they share phenotypic similarities then that is sufficient.

In general, the problem fails to distinguish between familiarity and proximity during rearing versus true relatedness.  Many animals "imprint" which immediately creates a bias towards a relatedness which cannot be independently assessed.  In other words, one cannot use the animal's behavior as an indicator of "r" or relatedness.

Some experimental evidence that seems to be missing would help establish:

• 1. Are animals capable of recognizing genetic relatedness or do they merely employ various cues of their rearing?
• 2. Between animals that exhibit "kin selective" behaviors, is there a change in behavior if siblings are separated at birth and reintroduced as adults?
• 3. Are purported acts voluntary or coerced? (In other words are there potentially other unseen or unknown consequences from within the social group?).
• 4. Do all members of a group participate?
• 5. Are "kin selective" behaviors taught?

There is little question that basic recognition of similarities, cooperation, and numerous other strategies can improve the chances of survival and fitness.  It is much more likely that reciprocal altruism is the dominant means by which such strategies occur (which was also recognized by Hamilton).  The concept of "gene selfishness", or the gene-centric view may be a useful tool for the biologist, but that doesn't make it a biological reality.  The simple truth is that Hamilton's Rule provides paltry support for the highly questionable idea that kin selection is the basis of altruism.

For additional articles see 'The Problem of Kin Selection Theory" and Steve Davis' "Hamilton's Rule, or Hamilton's Folly"

(1)  c = reproductive cost to the actor
b = reproductive benefit to the receiver
r = degree of genetic relatedness