What seems to be the problem in recognizing these words as being separate concepts?  Everytime I look there's another abuse of the word "selfish" when "self-interest" is intended.  Ranging from Ayn Rand's The Virtue of Selfishness (which immediately redefines selfishness so that it really isn't), to Richard Dawkin's The Selfish Gene (which is redefined so that it really isn't), to the following quote from The Introduction to Economic Analysis which states that people in economic models are "entirely selfish.  (The technical term is acting in one's self-interest)"

Where did this absurdity originate from?  Is it a fragment of human psychology that refuses to recognize any behavior except cooperation and so needs some extreme usage to make the point?

Self-interested behavior is the only kind of behavior a human (or animal) can engage in.  It is a direct result of self-determination and most importantly .... it doesn't need anyone else around for it to occur!

In other words, self-interested behavior is what you always do.  When you eat, you're behaving in a self-interested fashion, when you sleep you're behaving in a self-interested manner.  There is nothing that one can do that isn't self-interested. 

When an interaction with another human (or organism) occurs, then that self-interested behavior can manifest as selfish or altruistic.  However, it cannot be those things without the interaction.  Therefore to attribute such a trait to any organism requires that the organism be non-adaptable to circumstances and behaves in a mechanistic way.

It appears that the sticking point may be in the concept of altruism, which has also been a much abused term with definitions that don't hold.  Typically, altruistic behavior is seen where an individual willingly sacrifices themselves for the benefit of someone else (they incur the cost, while the other reaps the benefit).

Once again, this simply isn't true.  Altruistic behavior does not necessarily incur a cost, nor does it necessarily convey a benefit.  Both exist only as potential outcomes, so the real question to ask is:

"What self-interested behavior can warrant the risk of incurring a high cost to oneself for the potential of benefiting someone else?"

In humans this understanding of altruism is so deeply ingrained in our social fabric we tend to assume that it's a requirement, but ultimately ALL self-interested acts are based on some trade-off in costs and benefits.  These can't always be articulated and may even be trivial in many cases.

There are clearly cases of behavior appearing altruistic because a benefit is accrued but the cost is minimal or even non-existent.  Such results are still altruistic because they promote cooperation.  In some cases, altruistic actions may fail in which case the cost is high and the benefit is never realized. 

Without getting into all the issues that describe altruism, it is certain that altruistic acts that are not the result of self-interested behavior are either coerced or instinctive.  They cannot be the product of free will unless the individual is recognized has acting in their own self-interest (i.e. self-determination).

Just as in game theory there are two behaviors; defector (selfish) and cooperator (altruist), let's not invent more terms that don't reflect reality.  As I stated in the beginning, selfishness and altruism require the interaction of others in response to a particular circumstance or encounter.  Self-interest is the action of an organism no matter what it is doing.  Let's stop confusing the two.