What design is: 911 vs. Fleetwood

[Paul Graham <address@hidden>]

I think the smartest programmers have a lot more power in
dictating language popularity than most people realize.
Remember Ada and C?  As I heard it, the DoD required all
their software to be written in Ada, and you had to get
an exemption to use anything else.  But the best hackers 
insisted on using C, and the DoD had to grant so many 
exemptions that Ada soon became a dead letter.

I think if you want average dufuses to use your language,
you can do it by getting the geniuses to use it.  Then
the geniuses write Unix, and all the dufuses learn your
language in order to hack it; or the geniuses start 
Microsoft, write all the initial software in your
language, and then hire dufuses to write more in the
same language.  

The result is that the dufuses often end up using a
language that wasn't designed for them, often with
disastrous results.  Something similar happened with
the Porsche 911.  It was basically a racing car.  The
serious car people loved it.  The dufuses noticed this
and so they wanted 911s too, which was a problem 
because the car had major oversteer.  A lot of wannabees 
got killed as a result.  

Just like a lot of wannabees ended up writing stack
overflow bugs in C, which was certainly not designed to
be the language for "average programmers" that it became.
Not necessarily ideal, but I fear that is how the world 
works.

Our plan for Arc is, design the Porsche 911.  If you do
that, eventually all the drug dealers will want it too, 
even though the Cadillac Fleetwood was the car designed
for them.

(It's more *fun* designing the 911 too!)

--pg

--- Christopher Barber <cbarber@curl.com> wrote:
> | I think the solution here is to admit that continuations
> | are hard, but give up the idea that you're trying to
> | design a language for people who can't deal with hard
> | stuff.  It is an amazingly liberating axiom.  You can just
> | concentrate on whatever makes the language better, instead
> | of getting sucked into the black hole of trying to guess
> | what the  "average programmer" can deal with-- that's
> | marketing, not technology.
> 
> No, that's design.  Of course, if you do it right, you don't simply
> guess, but
> make efforts to actually measure the usability of your design for
> your target
> audiences.
> 
> | Psychological hurdles are a real problem, like Dan says,
> | but you can choose not to solve that problem.
> 
> Not if your target audience expects you to do so!
> 
> We could not adopt that approach when designing Curl because we
> wanted to
> support a much broader audience than advanced programmers.
> 
> | (My guess is that if you do end up designing a language
> | that all the best hackers like, everyone else will get
> | dragged along, psychological hurdles or no.
> 
> That would be nice, but I doubt that would happen unless you could
> come up with
> a language so wonderful that all talented programmers would adopt it
> and refuse
> to work with anything else.
> 
> - Christopher
> 
> 


__________________________________________________
Do You Yahoo!?
Check out Yahoo! Shopping and Yahoo! Auctions for all of
your unique holiday gifts! Buy at http://shopping.yahoo.com
or bid at http://auctions.yahoo.com