To answer the question I asked in a previous post, the Perl source code is at least 10MB, on top of how many gigabytes are taken up by the OS.
A Forth interpreter could be as small as 8k. Even full-blown development versions are only a couple of hundred kilobytes. Hence Forth can go anywhere there is a CPU and some memory. One can even buy CPUs with the instruction set being Forth. So think pervasive and lots of bang per byte. Employing the principal of "learn once, use everywhere" (hence I use bash, Vim and Perl), Forth looks like a more useful tool to learn than say VHDL or Verilog for FPGAs.
And I like the look of the new SEAForth 40C18 chip: 40 CPUs(with RAM and ROM) on a chip, all with in-built Forth interpreter. $21 in quantity. Could be quite interesting.