Actually, I do read his blog, one of the few from MS.  It usually hurts my
brain, and the sense it makes is only flitting.  I went back and read the
ones that I thought might be relevant, and here is his quote:

//Why is this broken? Because it should always be legal to put a Turtle into
an array of Animals. With array covariance in the language and runtime you
cannot guarantee that an array of Animals can accept a Turtle (animal->reptile->tutrle)
because the backing store might actually be an array of Giraffes (animal->mamal->giraffe).
//  Parens are mine.

So basically, I have this _thing_, and there is nothing anywhere that I can
convert it to in order to call methods on it.   I can't convert it to the
base class, and call base class methods.  I recognize the similarity, but
something in me won't see why it is the same case.  Is it not allowed because,
somehow at some point I could write to that object (which I used the generic
for) with a type that would be contravaint?, ie, I could theoretically try
to write a turtle into my giraffe, and the compiler can't tell that I am
not going to do that, so it just doesn't allow the conversion at all?

Sorry for the muddled thought.  It makes sense just enough that I get it,
but I can't write it properly.  In any case, I am not doing any thing dangerious
in this class, the only New or assignment happens in the constructor basically,
but the compiler can't guarantee that will be the case in the future.  Arg.

I read I can do it by poking at Ms.Vb.CompilerServices.Conversions, but I
don't think thats a good idea.

Thanks for the re-pointer =)

//Andrew