[plt-scheme] Re: Why "lambda"?

On Sat, May 30, 2009 at 09:38:49PM -0400, Marco Morazan wrote:
[...]
> Ok, I guess I will chime in at this point. I feel rather uncomfortable
> with the notion that using alpha and beta are more complex than using
> A and B. Frankly, we can use two different squiggly lines as far as I
> am concerned. The simple fact is that the use of a variable is a form
> of (low-level) abstraction. Who cares if the symbol used for the
> variable representing the abstraction is latin, greek, russian, or
> mayan? If you understand the abstraction it does not matter, period.

I completely agree, and was surprised that this wasn't said immediately
 (and then hesitated myself, thinking I might be missing something).

> Why use greek? Well, it is that little thing called convention very
> much like English is a convention. When I read a paper and see, for
> example, alpha-conversion or beta-reduction my focus is not on the
> alpha or the beta. It is on the abstraction that is being represented.
> That said, I agree that it is important to explain and understand the
> abstractions. Nobody understands or fails to understand what
> alpha-conversion and beta-reduction due to the alpha and the beta.
[...]

And conversely, which I think you allude to with "convention", there is
 no benefit to choosing a different symbol (syntax) that also has no
 intrinsic connection to the concept (semantics), and it makes it harder
 for those who have traditionally associated the concept to a particular
 symbol, so the best choice is the conventional one.

For some readers, it may further strengthen the impact of the argument
 to notice that no one arguing for lambda suggested it was bad or irrelevant
 to try "-<" instead of "amb" in a recent example of pedagogy, presumably
 because "amb" though conventional was meaningless to the intended audience
 (and unlikely to appear in the near future work of almost all of them) and
 the form of "-<" suggested some semantics.

BTW, thanks again to the PLT team: I'm enjoying using unicode, and the
 latex keybindings, very much, when the symbols are conventional or their
 shape suggestive. It's also been yet another way to quickly impress and
 interest a few people in PLT, around the CS department here (and beyond,
 e.g. a math grad).