P*rl in Latin, whither Python?

Alex Martelli aleaxit at yahoo.com
Mon Nov 13 23:44:50 CET 2000

"Martin Christensen" <knightsofspamalot-factotum at mail1.stofanet.dk> wrote in
message news:87wve7afwb.fsf at fangorn.stofanet.dk...
> 1) Esperanto is a nice idea, and it would be great if everybody spoke
>    it.
> 2) Not many people speak Esperanto, so it's not immediately useful.
> 3) Because it is not immediately useful, I will not learn it.

Right!  This tends to be the dilemma of innovations, if they play
in a field where 'network-effects' (in the economics sense)
dominate.  Many new technologies, such as CDROMs and DVDs,
meet it at their outset.  And, you know what?  The non-early-
adopters, the ones which wait and see, sometimes save the
expensive mistakes that the early-adopters make (think Beta,
laserdiscs, ...).  Some technologies (CDs, DVDs, ...) offer
good enough price-performance ratios (compared to already
widespread alternativesa) to eventually overcome this "self-
perpetuating argument" (CDs and DVDs are pretty well
established by now), others don't (the ones that validate
the wait-and-see attitude:-).  "Think of it as evolution in

> deadlock where we were still using imperial measurements and

Isn't that what the US population is more or less doing in
practice, whatever the rest of the world & its governments
may think about it...?-)

> Roman
> numerals because we were waiting for everybody else to adapt the
> smarter system!

'Arabic' numerals (actually from India) were quickly seized by
shrewd Italian merchants (and used for another new and quite
revolutionary technology, double-entry bookkeeping, which
may have been imported with them, or, more likely, invented
by some clever Italian) because networks effect mattered less --
using them gave *competitive advantage*.  Sure, there are
positive network effects too (if arabic numbers are widespread,
you'll be able to use them when exchanging numbers with
suppliers and customers; hiring accountants will be easier...);
but, if your competitor is taking 3-4 times as much as you
are for accounting, and is more likely to make mistakes in it,
too, then you're excellently well placed to eat his lunch.

The parallels with programming in Python are striking, aren't
they?  With normal human herd-instinct, we're all here trying to
gain new users (and there are positive network effects as we
do -- books, more tools, ...) -- but, being the only firm to
use Python in a given niche while your competitors struggle
on with lesser languages is *BIG*... "eat their lunch" time!-)

Human languages, like many other technologies centered on
communication, are iffier that way.  Maybe a Swiss-Sweden
combine, for example, would have a competitive advantage
if using Esperanto -- but none, AFAIK, has yet bet a huge
multinational firm on it (Asea/Brown-Boveri, just such a
combine, has adopted English as their internal language,
and have dove pretty well with it, I hear).  English, being
already-widespread, does have strong network effects in
its favour (even between countries with no English mother
tongue speakers).

> We see the same thing in the computer world. So many people think that
> Windows and MacOS are crappy operating systems, but until everybody
> else has converted they'll just stay with it.

A situation with a delightful balance of network-effects
versus competitive-advantage.  Note the difference between
the spreading of Linux and BSD variants on servers (where
network-effects due to the OS are modest, and there was
actually some early bias in favour of Unix variants) and on
the desktop (training of end-users, "where's my [insert
favourite can't-do-without professional/game app", ...:-).

Most people who care to look also loathe Intel-compatible
32-bit architectures, but, there, the network effects are
just overwhelming (here, mostly, due to "learning effects"
typical of the semiconductor industry)... Windows NT
variants running on other CPU's just about died, and by
far most Linux (and BSD variants) installations are on
Intel-compatible CPU's, too...


More information about the Python-list mailing list