[Sussex] Worth a read

[Geoff Teale]

On Sun, 2003-11-09 at 23:48, Mark Harrison wrote:
> Interesting read, but I found it a bit flawed.

As I found your argument I'm afraid Mark.  I agree totally with the
"engineering" view point, and that systems should be built by said
engineers, but going back to what Dobbo said, there is a compelling
argument for some type of programmatic interface rather than WIMP, of
which a command line is a fine example.

The thing is there are levels in all these things.  Very few people who
know at least one programming language (or indeed work as a programmer)
have any idea how an Operating System really works, let alone a
computer.  Just as very few people who can change a spark-plug or a fan
belt really understand why a pressurised mixture of air and hydrocarbons
ignites with a spark, and how in turn that ignition applies forces on a
piston head.

However, what has happened in the last 20 years is that in order to
promote wider use of the tool (but more importantly to sell more units)
we've "dumbed down" the interface to the tool, and in doing so made the
tool they use fundamentally less useful.  

This problem is linked in with the "power user" problem in businesses -
it's rooted in this:


People and businesses want to use the power of computing technology (or
rather they need to because competitors will do.  Traditionally very few
people actually _want_ to use computers).

Most powerful functions of computing require a level of expertise.

Expertise is time consuming (and to some people, boring) to acquire, to
a business this time can be expensive.

Software companies cottoned on to the fact that it was easier to make
money delivering a tool to do a complex task by providing rigid or
over-simplified implementation that was easy to use, than it was to make
money selling a tool that could do the job properly.

When it took off in the early-mid 1980's this principle extended (in
fact it actually really started) with the interface to the OS and really
came to the fore with the emergent concept of the office suite.  Price
kept it out of the business market for a while (Apples were expensive
and less "business like" than PC-clones running DOS), but when Microsoft
started to get somewhere with a GUI sitting on top of DOS people got
interested in the marketing line that it would mean anyone could use a
PC, that they'd need less training (and less salary) and be just as
effective as the high paid boffins in the "Data Processing" department.

Apple, Microsoft (and pretty much all the companies that came in their
wake) have traded on the idea that they can make a complex job easy. 
They do it with metaphors from the real world, visual queues and
naturalistic interaction.   All of this is very noble, and great for
business, but ultimately staggeringly inefficient.

What's happened here is that we've taken a staggeringly powerful tool,
and shaped it around the way we did things before it came about, and in
doing so have made it a minor aid to our day to day business that costs
massive amounts to implement, requires a huge support team, is wholly
unreliable and the biggest black hole for data in most modern
businesses.


-- 
Geoff Teale <tealeg at member.fsf.org>
Free Software Foundation