kinch@tesla.UUCP (02/08/84)
To continue my points of debate on the merits of the MacIntosh (mostly in
response to tesla!delph's reply to my review):
Peripherals on a serial comm port have to cost more than those on an parallel
synchronous (synchronous in the sense of working within processor cycles) bus.
I realize that we're debating the average number of peripherals per system,
typical performance requirements, etc., in order to arrive at an economic
conjecture as to the market potential of this machine. But the end of the
matter seems to be that Apple chose to make the base system as cheap as
possible (for the "mass market"), knowing that expansion would be more costly.
They could have chosen 1 internal slot, or 2, or N, but they chose zero, and
in consideration of the market and what most people buy computers for, this N
is lower than the optimal.
What cost slots? A serial interface for your typical peripheral, say an
Okidata dot-matrix printer, to convert from the parallel (read, bus-like) to
serial (read, MacIntosh-like) interface, costs $100. Not a $5 bill, as per
Jobs, for a microprocessor, although the Okidata interface uses one. I think
$75 might be a fair estimate of the extra cost to serialize a MacIntosh
peripheral's interface. Job's argument, though a simplistic cost estimate,
could be just as well applied to an internal parallel bus, "add a $5 bill [for
a card connector and printed circuit foil to extend the CPU and other signals]
to the motherboard, and then talk..." Note also that the daisy-chain pair of
serial connectors cost about the same as the bus-slot edge connector.
I don't know what the "Macintosh Finder" is, but there must be substantial
interplay of textual directives and icon-pointing directives. Icons are a
means to implement menus, and menus have this shortcoming. The theory here is
that if the task is simple the user-interface can be too; but if the task is
complex and requires discretion, detail, and finesse, a menu is unsuitable
(even if implemented in pictures). Most word processing is simple in this
way; software development isn't and everybody agrees Macintosh isn't targeted
at development. In between we have a debate. My judgment would be that the
profitable tasks yet to be computer-tooled will not be best implemented with
the Star/Lisa/Macintosh style. They are too entropic to be fitted with a menu
abstraction.
This limited applicability is in tension with the strong potential of
Macintosh in the school/college market, where flexibility and compatibility
are important. Apple has always been strong in low-end academic applications.
They're risking a lot by not meeting the academic pricing and demand until
late this year.
The software quality point is critical. Since the firmware/software is
graphic-oriented, it is especially difficult to commercially harden. If it's
not 99.99% bullet-proof, and at least the reviewers' and software developers'
models were not, then the product will be stuck with a bugginess stigma.
I admire the wealth of functionality and optimization in the user-interface
toolbox. But a few booby traps in there will turn off customers. The type of
customer who is least forgiving of bugs is exactly that customer who is
supposed to be attracted to the Macintosh: that naive soul who knows not the
nature of a software bug, how to avoid it, or how to work around it. Where
bugs are much embedded in firmware, the bug-killer will be hard to apply.
My objections as to response time delays due to big overhead are mostly based
on the Lisa experience. With her you waited 45 seconds to just delete a file
or start a new application. Having a 68000 and hard disk, it was difficult to
explain why the delay was that long. Now there's Macintosh with a single
floppy, supposedly using software descended from Lisa. Magazine reviewers and
non-computer people who don't know better don't think much about response time
(or at least the reviewers don't write much about it). But its a top-priority
item. I'm assuming that Macintosh is slower than Lisa.
Isn't it an error to introduce an ultimate word-processing microcomputer with-
out any letter-quality output device?
The argument that software compatibility isn't important because software
developers will supply all that's needed is circular. All kinds of software
is written for every computer made, but only successfully marketed computers
get software that is effectively produced, marketed, and supported. If your
machine is software compatible then you get the benefits of these products. If
a machine isn't compatible with anything, then nobody will market (with the
emphasis on market) software for it until a lot of them are sold. But not
many will sell without all kinds of niche-oriented vertical software, which
depends many machines being sold first, etc. Let's be realistic about
this. No software developer willingly ignores a market, but there will not
necessarily or initially be enough Macintosh computers in the hands of people
with money to spend on software to motivate software houses to develop
products AND spend the development costs five times again to get the
developments to market.
Writing and hardening software is only 20% of the software development job in
the microcomputer mass market. The up-front cost to publish, market, and
support a software product before any revenue is realized is much, much
greater.
This is one reason why so much software became available for the IBM PC.
Developers knew that the machines would be sold in quantity, and justified the
enormous investments in marketing required to sell a successful product.
I still think, after delph's reply, that service is going to be a sore point.
Apple dealer service may be great. But without free-market (that is, without
any interference from Apple) third party maintenance, your Macintosh will be
at the mercy of a maintenance monopoly. Good for Apple and their "authorized
dealers." Bad for customers. Companies and dealers that use the "authorized"
gimmickry are only trying to artificially eliminate competitors that provide
cost and/or quality options that Apple can't. Apple most certainly will not
give you a new computer if yours breaks. Would you buy a car that had a seal
on the hood latch, to be opened only by your authorized car dealer? The
analogy isn't perfect, but think of why the hood latch seal is objectionable,
and the same reasons apply to a computer's cover screws.
Calling dealers "authorized" and putting seals on cases are two old retailing
tricks. They are telling you that you are not qualified to judge the fitness
of a repair organization. They and only they will decide who should fix your
computer (if you let them get away with so deciding, which you don't have to
). This socialist method is supported by middling-size businesses to raise
service income above the competitive rates without raising service quality or
costs. The joke is that it doesn't work. The seals become meaningless after
a few months since the technicians get tired of policing and applying silly
little stickers. You can't void a warranty or charge extra because of a broken
seal, anyway, unless the customer signed an explicit agreement to waive
warranty for a damaged sticker before the sale. Anybody can call themselves
"authorized" because legally anybody can try to fix your Apple computer if you
want them to, and it doesn't void the warranty to do so. Provisions to the
contrary in gilded warranty text are not valid in most states. All in all
it's distasteful marketing bunko. Tell that to your local authorized
seal-preserving dealer and see if he doesn't either misunderstand this or
get embarrassed.
By the way, if Apple tried to enforce pricing (strict or otherwise) on parts,
they would be in violation of antitrust law and either enjoined or out of
business. Kaypro is in court for dealer price-fixing now.
Summary: What's unique to the Macintosh? Not price. Not the bit-mapped
display (IBM PC, too). Not the software/firmware model (also available on the
PC). Not the mouse (...). Well, the TV and print advertising are novel. The
post-mortem on the failure of the original Lisa to sell was basically (1) too
expensive (2) too slow (3) too narrow marketing target. Think about it,
nobody could say those things about the original Apple II. My prediction is
that we will be saying it about Macintosh.
Richard Kinch
decvax!cornell!tesla!kinch
(607) 273-0222