[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[arch-announce] Re: very old thread and the problems

From: Larry McVoy
Subject: [arch-announce] Re: very old thread and the problems
Date: Fri, 9 Aug 2002 21:04:40 -0700
User-agent: Mutt/

>   From: Michael Tiemann <>
>   > I think my message is pretty clear: I don't wish you any ill, and won't
>   > stand in your way, but neither do I feel any compulsion or
>   > responsibility for your situation.  I would appreciate it if you dropped
>   > me from this doesn't concern me.
> I soon shall, then, but no - your message is *not* clear.  It sounds
> like you are covering something up or not speaking forthrightly and,
> in my experience, when it becomes really important, those are unlikely
> courses for you to take.   So I am faced with this apparent
> contradiction between what I believe to be your very honorable values
> and your very dishonorable actions.
> Can you explain to me briefly why it should not concern you?  It
> relates to your nominal responsibilities as the CTO of Red Hat, and
> in awkward (hopefully less important) ways, to your past actions,
> during our past associations.
> It genuinely seems to me that helping me should be a no-brainer for
> you, for at the very least, the reason that the technology I have
> offered you is (easy for most any engineer to determine) very useful
> to Red Hat's advertised business goals.
> In that context, you seem *very much* to "wish me ill".

Come on Tom, you are jumping off the deep end.  I won't claim to, but
I think I speak for all of us when I say you have our respect, but you
are getting weird.  It's *your* job to come up with a business model that
works for your projects, not Mike or the kernel group or the subversion
team or anyone else.  Either you can make it work or you can't.

Everyone has to provide value in a way such that people are willing to
pay for it.  RedHat provides value by grouping together 100's or 1000's
of packages, all at least somewhat tested and very much supported, in
a way which is easy to install and manage.  The value is clear, they
charge us $50 or $100, whatever, but the amount of work they've done is
clearly more than $100 worth of anyone's time.  So it makes sense.

It's worth noting that a version control system is a tiny fraction of 1%
of the value that they provide.  Nobody would pay them $100/copy for CVS
or Arch or Subversion, it's just not worth it when you can download it,
type configure, type make install, and you're done.  See the difference?
While any of us could download one package and get it to work quickly,
very few us (if any) think it is less than $100 worth of our time to
put together a distribution, that's way more work and not very fun work
at that.

Your problem is that you don't want to face up to the fact that it
costs a huge amount of money to make a decent SCM system and there is
absolutely zero chance of ever recovering the development costs if it
is free software.  If it's free, noone will pay you anything for it.
Don't believe me, go ask Kingdon, he tried, we spend more in a month on
salaries than Cyclic grossed in a year.  And we're tiny, go check out
Rational's salaries, they're public.

If you recall, on the kernel list, I pointed out quite a while ago that
it would take about $12,000,000 for you to catch up to where BitKeeper
or ClearCase is today.  You accused me of defaming arch, all sorts of
nonsense like that.  I wasn't disrespectful of arch at all, quite the
opposite.  You can certainly take arch to the point of a real product.
It will cost you at least $12M.  The problem is getting that money back.

You aren't going to get it back if you give away the product.  Does that
mean that giving it away is the wrong answer?  Not necessarily.  It
depends on what you want.  If you value freeness more than completeness,
then you should go the free software route.  You've been screaming that
the arch project is in crisis.  It is NOT in crisis, what is in crisis
is your financial situation.  Arch is GPLed, so you could slap it on
sourceforge or whereever and let it go down the path of free software
everywhere.  Either people will work on it or they won't, but it is in
fine shape, it's free.

If you want a production quality product, then you may have to give up on
the free software aspect.  That's not universally true but it certainly
seems to be the case the more obscure the application.  And I think we
can all agree that SCM is pretty obscure.

My advice to you is the same as it was 6 months ago: decide what
you want.  If you want arch to be a contender, take it closed source,
get some funding and go for it.  If you value freeness more, then let
arch loose, let other people drive it forward and see what happens.
Either way, there is no crisis, only opportunities.
Larry McVoy              lm at   

reply via email to

[Prev in Thread] Current Thread [Next in Thread]