Subject: Re: Larry Ellison on FSBs
From: "Ben Tilly" <btilly@gmail.com>
Date: Tue, 25 Apr 2006 11:06:19 -0700

 Tue, 25 Apr 2006 11:06:19 -0700
On 4/25/06, Stephen J. Turnbull <turnbull@sk.tsukuba.ac.jp> wrote:
> >>>>> "Ben" == Ben Tilly <btilly@gmail.com> writes:
>
>     Ben> Clayton Christensen would disagree.  His finding is that by
>     Ben> doing everything that you're supposed to do (paying attention
>     Ben> to your customers, etc) a company will be successful in
>     Ben> ordinary business situations but will fail miserably at
>     Ben> dealing with disruptive innovation.  Because it is the nature
>     Ben> of disruptive innovation that your customers will tell you
>     Ben> that they are not interested in it until it is good enough
>     Ben> for their needs, at which point it is too late for you
>     Ben> because you have entrenched competitors whose cost model you
>     Ben> can't compete with.
>
> Yes, that part of Christiansen's thesis is hardly disputable these
> days.  But who's entrenched?  Oracle.  Who can't compete with whose
> cost model?  MySQL cannot compete with Oracle's cost model, because
> whatever MySQL has put together at great expense, Oracle has too, for
> free.  Ellison said, sure, we recognize that open source is a threat
> to our software product, and we have no intention of competing with
> it; when it's good enough to hurt us, it's good enough for us to adopt
> it.

What are Oracle's profit margins?  What are MySQL's profit margins? 
As long as MySQL is structured to accept thinner profit margins, they
have a competitive advantage against Oracle.  This is why, for
instance, Sun is unable to successfully enter the Linux market, even
though it is theoretically all free for them.

Furthermore MySQL doesn't put together its improvements at any expense
to itself.  My understanding is that the improvements it makes are
generally paid for by customers who want those improvements.  Given
their dual licensing model, as long as MySQL can maintain control of
the software base, they are the only people you can get those
improvements from.  Conversely if Oracle wishes to enter that market,
they have to fork the software base.  Which will be far from free for
Oracle, both in money and reputation.  Doubly so since Oracle doesn't
understand what people like about MySQL.  (Or if they do understand
it, they can't go there without going directly counter to the
interests of all of the Oracle DBAs out there.)

This is not as secure a barrier to entry as outright ownership, but it
is still a barrier to entry.  (The more significant one is the profit
margin issue.)

> "That's hard!" says Mr. Anderson.  I beg to differ.  If it was hard
> enough to be a problem, this list wouldn't exist, and neither would
> the FSF.  Nobody would care about FLOSS, nobody would bother with
> enforcing copyright, and even patents wouldn't be a big deal, because
> *existing* software would be a rival resource, embodied in the
> particular programmers who are the only ones able to extend it.  Sure,
> it's not easy, either, but the problems that Oracle faces in adopting
> an *existing* "good enough" open source product can be addressed with
> money.  After that, they're in the same place they are now.[1]  Note
> that Oracle can fund an army (1000 or more) of software engineers just
> on the interest on cash-in-hand.

You have way overstated the case.  What is accurate is that if it was
hard enough to be an *INSURMOUNTABLE* problem, then the rest of what
you say would be true.  But it is not.

It is, however, hard enough to be a problem.  And one approach to free
software is how to use that problem to your advantage.  (Dual
licensing and being responsive to their community enables MySQL to
make it a bigger problem than normal.)

Note that if Oracle funds an army of software engineers to tackle the
problem, then for structural reasons that CC goes into in detail,
Oracle is guaranteed to get a business that is structurally incapable
of operating at the thin margins that they will need to to compete. 
Money isn't the answer for all problems.

> Oracle moving toward open source databases could easily kill MySQL.
> But if Ellison isn't just grandstanding in that interview, Oracle is
> already over the hump; they're not dependent on selling technology.
> Yes, they need to *offer* technology, but it's not clear that they
> even *want* it to be proprietary.[2]  And Ellison (says he) knows that.

See my response to Keith.  When you look at the books, licensing isn't
the bulk of their revenue.  But it is much harder to figure out how
much of their business depends on the acceptance of their database.

> Is it really that easy?  Of course not.  Timing is always difficult.
> It's one thing for a database company to throw away its webserver
> product; it's another for it to throw away its database product.  But
> *Christiansen's "incumbency is a burden" thesis doesn't apply* to open
> source threats, not if the incumbents are paying attention to it.

As long as Oracle is structured to need fatter profit margins than
MySQL, Christensen's analysis still holds, IMO.

[...]
>     Ben> There is exactly one situation that I'm aware of where paying
>     Ben> attention to your customers becomes a losing long-term
>     Ben> strategy, and that is the situation that Oracle's core
>     Ben> business is in right now.
>
> eh.  "The customer is always right" is not what I'm saying, and it's
> not what Oracle et al are doing.  Have you heard of "XP"?  Well, what
> Oracle and SAP and IBM are trying to offer their customers, with a lot
> of success, is "XB".  What MySQL and Red Hat are not yet in a position
> to offer their customers is "XB".  But whatever MySQL and Red Hat
> offer their customers in terms of technology is available to Oracle to
> offer its customers, well before their subscriptions run out.  That's
> what Ellison is saying.

Please explain your acronyms.

I have heard of many things called "XP".  None of them seem to fit
this paragraph.  And I definitely don't know what you mean by "XB".

> Ellison is also saying that developing software for sales is not
> Oracle's core business any more.  Sounds plausible to me; I know that
> Oracle (and SAP and IBM) are what scares my former students now
> working for NTT Data and Nomura Research, two of the leading Japanese
> business process consultancies.  Do you think Ellison is just blowing
> smoke?  (Not a rhetorical question!)

I think that Ellison has a history of grandstanding and overstating
his case.  I don't doubt that he has a basis for his case.  However I
believe that he has significantly overstated it.

As I've said before, I'm aware that Oracle is in auxilliary
businesses.  Unlike Microsoft, they seem to be succeeding outside of
their core business.  (Microsoft, by contrast, consistently loses
money on every business line except Office and operating systems.)

>     Ben> Furthermore new businesses or not, it still hurts Oracle to
>     Ben> see their cash cow evaporate.
>
>     >> Companies like Oracle don't drink milk, they eat steak.
>     >> Unexpected visitors?  Kill the fatted calf and have your
>     >> barbeque a week early.
>
>     Ben> Strangely enough this kind of confidence has not helped any
>     Ben> other company that I know of which has faced disruptive
>     Ben> innovation.
>
> LE on open source: "So it is not disruptive at all - you have to find
> places to add value."  But let's just ignore that. :-)

Of course he doesn't want people to believe that it isn't disruptive. 
But I've seen how much work it takes to administer Oracle versus a
similar amount of data in MySQL.

> How about System/360?  You don't think Thomas Watson did that because
> he thought risking his company would liven up the talk at his
> investors' cocktail parties, do you?  The story may be apocryphal, but
> supposedly Andrew Carnegie tore down a nearly completed, multi-
> million-dollar steel mill and started over because his engineers
> offered him a three cent/ton saving.  Disruptive innovation is nothing
> new, and history shows that sometimes the big boys get the answer on
> the first try, and some times they don't.  Christiansen is only
> reports on the latter, IIRC.

If we're going to talk about Christensen, then we need to use the
words in the way that he does.  Disruptive innovation in his thesis
doesn't mean "stuff that causes lots of work to be done".  It means,
"established businesses in competition with businesses whose product
is comparatively crap, but which can be expected to improve faster
than customers' needs".

If a new product is better by the standards that an industry uses to
measure such things, then it is not disruptive in Christensen's
meaning.  I believe that this applies to explain why both of your
examples are not counter-examples to Christensen's thesis.

> Any good MBA course on strategy will start and finish with the
> importance of cutting your losses, and adapting to the environment.
> It's not always possible.  But it's not always impossible, either.

"In theory we all know how to dance.  In practice we don't do it." 
Paraphrased from memory from a manager looking at an analysis of how
many critical tasks they were skipping when trying to launch new
products.

Where things get very difficult is when it is obvious that your
long-term interests are best served by doing A, while your short-term
interests are best served by doing B.  In that situation it is hard to
do A, and harder still for organizations than for individuals.

> What does this mean for FSBs?  Same old same old: take care of your
> customers, and they'll take care of you.  And to quote LE himself:
>
>     So it's all very interesting.  You can build a sustainable
>     business [in open source], you just can't charge a lot
>     for it.  There's brand value - there's real brand - there's
>     people, and that's it.

By his terms that is absolutely right.  You can't charge a lot in open
source, and you certainly can't charge as much as Oracle is structured
to need.  Unless Oracle can restructure itself to not need that
(organizationally very difficult), it needs to seek business
opportunities elsewhere.  (Of course, like IBM, those opportunities
might be found in complementary businesses to open source ones. 
Indeed I suspect that Oracle is going that way.)

That's been the downfall of established companies in other cases of
disruptive innovation.  Of course it is obvious to them which way the
market is going.  But they simply can't afford to compete head to head
with the innovators because the innovators are structurally able to
create a profit at margins which are way below what established
competitors can possibly make money at.

> Footnotes:
> [1]  Of course MySQL will know their database better, which gives them
> a small technical edge.  But Oracle has the same edge with *their*
> database.  The technology is a horse race.  So what is going to tell
> is the service.

Actually the technology is NOT a horse race.  Take 2 machines that are
installed and ask two competent people to install a database on it. 
MySQL will be installed faster, and requires  lot less competence to
install.

By contrast Oracle installations that I've seen are using features
that simply aren't found in MySQL.  (Or weren't until recently. 
Though plenty are using features that still aren't in MySQL.)

> [2]  Consider the cost savings, for one thing.

We'll see how it works out for them.

Cheers,
Ben