Language Market Survey

Mark S. Miller
Thu, 19 Nov 1998 15:04:13 -0800

As previously discussed on the list, inspired by Bjarne, I'm making many
language design decisions based on my (admittedly bad) marketing intuition
regarding my potential audience.  However, I'm not purely shooting in the
dark -- just mostly.  Here's my market survey technique:

Go to a technical bookstore -- Stacey's, Computer Literacy, Stanford on or
off campus, etc -- and look at their language wall.  Bookstores may
overstock during or immediately after some new hype wave, but 6 months
later I'd bet they're allocating their precious shelf space based only on
what books are moving.

The proportion of books being bought on different languages does not seem
to be a good indicator of the current proportions of language use, but I
believe it's a good leading indicator of future proportions.  Why?  It says
where people are currently investing their precious learning time.  Using
this indicator, I could see earlier than most how wildly successful C++
would be.

The current approximate measures at Stacey's in Palo Alto:

(1 case is usually == 6 shelves)

C & C++:				3 cases
Java, JavaBeans, Javascript: 		3 cases
Visual Basic:				11 shelves
Perl & Python:				1 shelf

I suspect there's something wrong with the above Perl & Python proportion,
so I'd be especially curious what people find at other stores.  Even adding
in Tcl & csh would have left it at about 1 shelf.

It's a shame that Javascript is irretrievably mixed in with Java, as it's
no more related to Java than Python is.  Ignoring Javascript, we see that
the C tradition is way bigger than the VB tradition, but *among scripting
languages*, the VB tradition is way bigger than the C tradition.  This
ambiguity is why I'm so interested in hedging this bet.

Btw, this technique doesn't just apply to languages.  There's more shelf
space devoted to JavaBeans than to ActiveX and COM combined.

	--MarkM, marketing amateur