Re: [cobalt-users] the GUI interface

[Dan Heller <argv@xxxxxxxxxxxxx>]

I knew that what I said about the GUI interface would be misconstrued,
but it could not have been helped. Brent's reply was a gentle sort of
"I disagree with you" message, and he validated it with, "Cobalt is
doing a great thing by their marketing", and noting that their product
can by used "by anyone's mother."

The part that's being misconstrued is the perception that I have a
problem with the fact that there *is* a GUI. Nonono... that's a _great_
thing, and I wholly support it! In fact, I don't even mind that there are
limitations of the GUI. My problem is with Cobalt's policy that the GUI is
the *only* supported interface for doing system-administration on the box,
and the assumption that anything can be done (and therefore should only be
done) using the GUI is seriously flawed. So much so, that they are limiting
their own future potential with this policy.

At some point, any serious administration of a unix system requires
knowing the details of the applications that do the administration.
Thus, you're going to need to read the man pages to see the command-line
parameters and side effects, or related programs, associated with any
given task. Yet, this de-emphasis of non-GUI tools has caused their
main infrustructure of the system to begin to erode. For example, many
man pages are missing, incorrect, or out of date. There's not even a man
page for *csh* for cryin' out loud.

There is a fundamental design of good system architecture, and the GUI
is at the top of the pyramid. It's a UI-element, and its purpose is to
provide convenient access to lower-level functionality. However, Cobalt's
policy towards the GUI is analogous to Microsoft's and Apple's: that the
GUI is *part* of the operating system, rather than a discrete element
on top of it.

The Cobalt box is a fine value for the dollar. The way Cobalt markets
it is reasonable and understandable, given the early stages of where
the internet is today; there are early adopters, and people are flooding
into the market rapidly. There's no question that the Cobalt boxes are
useful and reliable by comparison to anyone else. If their marketing
gets them ahead of everyone else, more power to them. But, as I used
to tell my marketing staff, "be sure you know where the bullshit ends
and the reality begins: keep your eye on the longer-term strategic
direction so as to keep short-term marketing 'hype' in check." Cobalt's
success is going to depend on the same principle, but it'll only be
successful if they have a fundamentally sound technical strategy.

Microsoft gained monopolistic power NOT because of their operating system,
but because of the business-critical apps that run on top of them. They
maintained that monopoly by limiting competitive app companies from develop-
ing viable alternatives technically by closing off the OS, and by keeping
the "bundle" price of the OS/apps package so low that it wasn't viably
possible for competition to get their foot in the door. The connection
with Cobalt here is that MS tried to link the GUI portion of Windows to
the operating system itself in order to fend off competition, but Cobalt
can't do that. They don't own unix, and won't keep out competition simply
because they have a gui front end for system-administration. It might give
them an "edge" over competition, but they'd have to be very good about it.
And the one critical thing in the UNIX world is understanding the that
reality of real system administration requires flexibility for the admin's
benefit. Let him use scripts and hand-editing of config files as needed.
If they don't do this, they'll be leap-frogged by the company that *also*
provides a nice-convenient GUI to system-administration, but doesn't require
it.  In the UNIX world, good marketing only takes you so far -- without good
technology underneath it, you don't last long.

-- 
	--dan

Photo Gallery:  http://www.danheller.com/