Subject: Re: ARM: The Non-Evil Monopolist
From: "Stephen J. Turnbull" <stephen@xemacs.org>
Date: Mon, 12 Jul 2004 15:14:35 +0900

>>>>> "Ben" == Ben Tilly <btilly@gmail.com> writes:

    Ben> My concern with monopolies is that they can leave someone
    Ben> with both the position and incentives to be abusive and stand
    Ben> in the way of further progress.  If you got to where you are
    Ben> by giving a better product at the right price, the more power
    Ben> to you.

Be careful: I could draw the inference that if they turn around and
abuse their position now, that's OK.  You implied that the question is
"are you going to continue to give a better product at the right
price, or not?", and that is the right one.  Neither current share nor
past behavior necessarily gives a good prediction.

What you want is to have a market structure that encourages good
price/ quality offerings.  It looks (from some of the commentary on
Slashdot) like ARM has deliberately gone out of its way to tie its own
hands on that issue, by licensing to many fab companies on liberal
terms (which means it is always competing with its own older designs),
and by providing specs for integrating ARM into a design, which makes
it possible (but presumably still relatively expensive) for other CPU
manufacturers to get into the market.

Why would ARM choose to do that?  Because the market for new designs
is apparently expanding pretty much as fast as ARM can supply it, and
as fast as its customers can produce chips, even in the face of the
global recession.  By "being nice" it can maintain a dominant share of
an expanding pie.

What happens when the market stops growing that fast?  You have to
wonder what ARM's plans are for that contingency.

    Ben> ARM's monopoly

What monopoly?  You have to define what "market" they have "nearly
100%" of.  In a market growing this fast, most of the sales are in the
distant future (say more than 1 moore-doubling from now).  ARM may or
may not _have_ to be nice, but the rapid growth in their primary
markets means that there's very little cost to them in being nice if
that adds to investors' belief that they'll still be dominant in a
couple of years.  They can have their current revenue and a dominant
market position for the foreseeable future, too.  But that's not all
that far into the future, which is where "most of the market" is.

I conclude that short-run market share does not a monopolist make.  I
mean, if you split "short-run" by milliseconds, you'll discover that
the whole stock-market is continually cornered.  The question of where
to split the planning horizon (short vs long enough) is a matter of
judgement.  FWIW, my judgement is that one year is quite short from
the point of view of ARM trying to cash in on its flow market share.
Your judgement may be better (or simply different), of course.

    Ben> If their incentives changed, I'd bet that their behaviour
    Ben> would eventually follow suit.  Then you'd start hearing the
    Ben> complaints.

Exactly.  I think their incentives _will_ change, but it will take a
couple of years, and _because_ of the way they've hemmed themselves
in, they'll be more successful at monopolizing their segment of the
embedded market than IBM was in the PC market.  My point being that
it's not obvious at all that Apple has been as successful as IBM.

You also have to worry about "X inefficiency", which is the problem
that the successful monopolist will fail to work as hard as a firm
that faces rivals, so that everybody gets less for their money.

-- 
Ins