Interesting article,

Jon 'maddog' Hall maddog at li.org
Fri Mar 5 14:10:28 EST 2010


O.K., I will wade in here. :-)

For the most part, Ben is right.  Vendors who completely control both
hardware and software can make the "best" products, if your definition
of "best" is a limited market of items, and you are willing to pay for
them.  MVS, VMS, Digital Unix.  Rock solid, stable, scalable.  REALLY
EXPENSIVE.

A lot of Microsoft's problems have to do with drivers that come from
different vendors, trying to control different controllers that fit into
buses that are not that well documented.  One mistake in a driver
(inside the monolithic kernel) and BAM!  Lockup and blue screen.  I am
amazed that Microsoft's eco-system can actually boot at all.

>> Apple could have
>> crushed MS by now if they had gone with the GPL attitude instead of
>> picking BSD so they could keep all their toys to themselves.

>Yah, I'm not buying that.  If all you needed was the GPL, Linux
>would already have crushed Microsoft.

Ben, I think you are under-estimating your own argument about inertia.

Inertia is all about acceleration, not really speed.  In 1991 Microsoft
was already going 50,000 mph and accelerating and Linux started from
zero, with almost zero acceleration.  In 1994 a lot of the vendors
seemed as if they were going to give the server market to WNT.  There
were a lot of people saying that "Unix was dead".

After twenty years Microsoft is still accelerating, but I think it is
accelerating at a slower pace, and FOSS is accelerating at a faster
pace, but has still not caught up.  Then there is distance traveled, or
"speed over time" (in this case, installed base).  It may take a very,
very long time before FOSS has the same installed base, much less
"crushing Microsoft".

Apple has existed for about the same time as Microsoft, and still has
about the same market penetration as twenty years ago.  Its acceleration
is a lot slower, and more or less allowed it to keep the same desktop
and server market share (or maybe lost server market share in that
time).

>(When Linux first came out, you also had a fleet of commercial Unixes,
>Novel, several BSDs, OS/2, BeOS, and all sorts of other bit player
>platforms.  Today it's all Microsoft, with Apple and Linux nipping at
>their heels.

Sadly, and from a "choice" and "research" viewpoint this is true.  But
the fragmentation meant that unless any of them reached critical stage,
they would be just what you said "bit players", and would have died
anyway.

>  Absolutely correct!  However, the fact that's it's a hard problem to
>solve doesn't mean it isn't a problem.  Indeed, the fact that it's a
>hard problem is why it hasn't been solved yet, and why the best idea
>anyone has come up with is sheer persistence over time.

Yupolutely!  And in May of 1994 I came back from meeting Linus Torvalds
and made a presentation to my Digital Unix management at DEC that had as
a final bullet on the last page:

                 o Linux is inevitable!

They asked me what that bullet meant, and I said that no one could stop
Linux.  My management laughed.

Now Digital Unix is dead and most of them work for Red Hat.

What I meant by "Linux is inevitable" was that the concept of designing
a FOSS ecosystem with community was inevitable.  "Linux" itself may
migrate, fork, evolve, etc. but the model is here, and it will
accelerate.

When I worked for Bell Labs in 1982 I heard someone say "I do not know
what the next operating system will be, but I bet it will be based on
Unix."  I answered "I don't know what the next operating system will be,
but I bet it will be called Unix", meaning that the operating system
would evolve maintaining the same name.  I was only wrong by a couple of
letters. :-)

md



More information about the gnhlug-discuss mailing list