Subject: SELLING support contracts
Date: Mon, 14 Sep 1998 11:17:10 -0400

I have kept quiet on this thread -- partly because I was out of town
when it started.  I have a couple of thoughts about it all, and no
time to actively debate them at the moment.  I'm tossing these out
only because they *may* inform the discussion.

On Bit Copiers:

Bitcopiers who distribute on CD-ROM present no greater threat to
(e.g.) RedHat than people who use FTP to download the release.  It
seems likely to me that quite the reverse is true.  I'll use RedHat as
an example, but I think the point generalizes.

If the user is knowledgeable enough to FTP down the distribution, they
don't really *need* RedHat's support.  Our lab at Penn is an extreme
example of this -- we've used RedHat extensively since RH4.0 on almost
100 machines, and I don't think we have *ever* called for support.  We
generally buy two or three copies of the release CDs for convenience
of installation, but that's pretty much it.

A CD distribution, by contrast, makes the product accessable to people
who *do* require support to install it successfully.  These people are
eventually forced to go to the original vendor.

I am struck, however, by the fact that installation support is growing
less important as installation becomes easier.  Also, in the
bitcopying world, a large proportion of those who cannot install will
just give up, and these customers are lost.  It seems to me that
post-install support becomes very important in such a world, and I'm
not sure how much of that *anyone* can afford to do on a $45-$50

On pricing:

The debate about what the price of a libre product ought to be is
truly silly.  Nobody here is in a position to control the outcome.

My belief is that pricing is set by the surrouding incentive
structure.  One of the positive things (to me) about the open source
movement is that it is steadily increasing the importance of
reputation and recognition as positive incentives for distributing
good work.  Price becomes less important when it ceases to be the
*only* incentive.

That said, the incentive for most developers is still to get as much
money as they can.  This cannot, in the long term, be changed by
regulation.  Changing the incentive structure is much more effective.

Perverse thought: given that the libre terms allow free
redistribution, we should be encouraging people to think in terms of
how to build *expensive* products that people will actually buy --
even if they do not ultimately choose expensive pricing.  The only way
this will work is if the provider really provides added value, which
is something we want to encourage.

On identification:

There is a tension between branding and unrestricted distribution.  It
seems to me that a libre license can legitemately require a
redistributor to change any strings identifying the originator of the
distribution to protect themselves from people shifting the burden of
support costs.  This doesn't strike me as conflicting with the
*intent* of the open source license, and it provides one basis for
motivation supplier (and brand) loyalty.

On services:

There has been a trend in the proprietary world over the past decade
toward viewing software as a subscription business.  The marketing
push behind technologies like ``Windows Update,'' and the potential
for such technologies to make a real difference, suggests that this
trend is accelerating.

Just because the software is libre doesn't mean that the update
distribution mechanism needs to be....