[net.followup] AT&T vs. the toolkit approach

wm@tekchips.UUCP (Wm Leler) (06/12/84)

I disagree that most of the "new" computer users out there
that the 3B is aimed at won't be interested in the famous
UNIX tools.  Look at all the PC software out there that
people are using -- word processors, spread sheets, graphics
and such.  And now what are people looking at doing?  Why,
connecting them all together (like *tools*), of course.

UNIX *is* tools.  Without the tools, it is not really that
good an operating system.  It is old, flabby, has a terrible
user interface and funny command names.  The first "program"
I wrote on UNIX was "written" in awk, sed, and nroff (no C).
It was about 6 pages, took me 3 days to write (including
learning how to use UNIX), and replaced a fortran program
that was 3 inches thick.

AT&T seems to be doing a good job of shooting themselves
in the foot.  The question is, AT&T is so big, how many
times can they shoot themselves before they get hurt?

			Wm Leler

hammond@mouton.UUCP (06/12/84)

I think that AT&T is responding to the marketplace - not the hacker
marketplace, but the real world one they hope will buy their machines.

Essentially, MSD*S comes with very few programs and as therefore quite
cheap to license on a mass basis (I heard $8/CPU for lots of CPUs).
Everyone buying personal computers running MSD*S or equivalent then goes
out and pays extra for the programs they really want.  I think AT&T is
trying to structure their marketing the same way.

In some ways this seems to be a sign of a willingness of AT&T to help
it's System V resellers.  If only 3B's had System V then they could
give away the software if you bought the machine.  However, if other
companies port System V to other machines, then AT&T stands to gain more
by not making it too expensive to put on those other machines, hence the
MSD*S type strategy of buy what things you want.

Without a price list even this is conjecture, but no worse than anyone else's.
Rich Hammond    Bell Communications Research,    allegra!hammond

blenko@rochester.UUCP (Tom Blenko) (06/13/84)

	I think that AT&T is responding to the marketplace - not the
	hacker marketplace, but the real world one they hope will buy
	their machines.

Where are all these real-world people? Apollo has some 2000 units in
the field. Sun has a couple of hundred, I would guess. How many of
these are give-aways? How many went to hackers of one sort or
another? How many are actually used by these application-oriented
people? Would you put up with Unix for $20K with only binaries,
even though you DO know something about it?

	Essentially, MSD*S comes with very few programs and as
	therefore quite cheap to license on a mass basis (I heard
	$8/CPU for lots of CPUs).  Everyone buying personal computers
	running MSD*S or equivalent then goes out and pays extra for
	the programs they really want.  I think AT&T is trying to
	structure their marketing the same way.

OK, if you believe AT&T wants the OEM market only (that's not what you
imply, or what I believe). Otherwise, Johny Smith, Inc. can port all
those Unix commands, and undersell AT&T any day of the week. And AT&T
also isn't prepared to ship lots of end-user-type application software.

I've worked for two companies that shipped Unix boxes, and this
approach was discussed at both. It was strictly a martketing scheme to
get a little sugar on top (or make the basic machine look cheaper,
however you like).

	In some ways this seems to be a sign of a willingness of AT&T
	to help it's System V resellers.  If only 3B's had System V
	then they could give away the software if you bought the
	machine.  However, if other companies port System V to other
	machines, then AT&T stands to gain more by not making it too
	expensive to put on those other machines, hence the MSD*S type
	strategy of buy what things you want.

If System V resellers aren't selling 3B's, just cheap AT&T (system)
software, how does AT&T make any money? AT&T certainly isn't going to
guarantee that their Unix commands will run on everyone else's
hardware.  And if AT&T is charging a bundle for the commands anyway,
why wouldn't the reseller port the commands herself, and take the
profits instead of handing them on to AT&T?

	Tom

hammond@mouton.UUCP (06/13/84)

Note: These are my opiniions, I don't know what AT&T is trying to do!

    Where are all these real-world people? Apollo has some 2000 units in
    the field. Sun has a couple of hundred, I would guess. How many of
    these are give-aways? How many went to hackers of one sort or
    another? How many are actually used by these application-oriented
    people? Would you put up with Unix for $20K with only binaries,
    even though you DO know something about it?

First of all, I think the cost of the 3B2 is around 2x that of the
IBM PC/XT or $10k not $20k.  Second, AT&T is not selling UN*X for the
fun.  They are in it for profit.  While they will make all sorts of
claims about how nice it is, they are only trying to get people to buy
it instead of an IBM or DEC PC (at least for the 3B2) with the IBM
or DEC operating system.  The "real-world people" that I think AT&T
sees itself going after are the people who bought IBM PC's, with
a trivial operating system in binary only form.  They are not out to
put SUN out of business.

    OK, if you believe AT&T wants the OEM market only (that's not what you
    imply, or what I believe). Otherwise, Johny Smith, Inc. can port all
    those Unix commands, and undersell AT&T any day of the week. And AT&T
    also isn't prepared to ship lots of end-user-type application software.

I don't think they want the OEM market only, but they are interested in
it.  They would rather collect a license fee (even a small one) for
UN*X than have fee go to other vendors like microsoft.  The person who
ports the UN*X commands still has to pay a fee per machine they put them
on so AT&T still makes something.  AT&T is probably doing what IBM did
and getting people to put together end-user software for them.  IBM has
lots of software for the PC which IBM didn't produce internally.

    I've worked for two companies that shipped Unix boxes, and this
    approach was discussed at both. It was strictly a martketing scheme to
    get a little sugar on top (or make the basic machine look cheaper,
    however you like).

I've never denied that we're talking about a marketing scheme.  It would
have made it easier for your companies to cut the price if the license fee
varied with the amount of software included.


    If System V resellers aren't selling 3B's, just cheap AT&T (system)
    software, how does AT&T make any money? AT&T certainly isn't going to
    guarantee that their Unix commands will run on everyone else's
    hardware.  And if AT&T is charging a bundle for the commands anyway,
    why wouldn't the reseller port the commands herself, and take the
    profits instead of handing them on to AT&T?

AT&T collects money per system sold.  That's more profit than if the
operating system were MSD*S.  Further, the number of machines running
UN*X is important, AT&T wants a large UN*X machine market share, even
if not all the UN*X machines are AT&T's.  No, I wouldn't expect AT&T
to guarantee their commands nor even sell them directly to end users
of other hardware (except DEC VAXEN and 11/70).  The charging is per
machine sold and I suspect will also include per collection of software.
Hence AT&T gets a fee from the reseller unless the reseller rewrites
the commands herself and there are better things to do than rewrite 'cat.'

Rich

bob@islenet.UUCP (06/14/84)

AT&T may (or may not) be missing an opportunity to gain a market advantage,
depending upon how easy it is to get the full range of tools for their
systems.

The easier it is, the more developers will do it, and the more 2nd source
software will be available for the AT&T machines.  The smaller the machine,
the wider the market, and the more important 2nd source software becomes to
potential buyers.
-- 
Bob Cunningham   ..{dual,ihnp4,uhpgvax,vortex}!islenet!bob
Honolulu, Hawaii

guy@rlgvax.UUCP (Guy Harris) (06/16/84)

> 	I think that AT&T is responding to the marketplace - not the
> 	hacker marketplace, but the real world one they hope will buy
> 	their machines.

> Where are all these real-world people? Apollo has some 2000 units in
> the field. Sun has a couple of hundred, I would guess. How many of
> these are give-aways? How many went to hackers of one sort or
> another? How many are actually used by these application-oriented
> people? Would you put up with Unix for $20K with only binaries,
> even though you DO know something about it?

The Sun/Apollo marketplace isn't the 3B2 marketplace.  The 3B2 marketplace
is the Fortune 32:16 marketplace, the Zilog System 8000 marketplace, the
Altos Whatever marketplace, the Plexus Whatever marketplace, etc..  It's
the small multiuser box markeplace, not the single-user high-power bit-mapped-
display workstation marketplace.

	Guy Harris
	{seismo,ihnp4,allegra}!rlgvax!guy

ron@brl-vgr.UUCP (06/17/84)

And now for something really disgusting.  ATT will not support SysV on
the 3B20D.

-Ron

ron@brl-vgr.ARPA (Ron Natalie <ron>) (06/18/84)

Funny you should mention the Fortune 16:32 marketplace, Guy.
Look how sucessful they were when they went around charging extra
for PS.
-Ron

zrm@mit-eddie.UUCP (Zigurd R. Mednieks) (06/18/84)

AT&T if, in fact, botching their entry into the market with the
unbundled approach to software. 3rd party software vendors won't want to
deal with umpteen software configuartions that may or may not be
orthogonal to the hardware configuarations. If my precompiler relies on
sed being there, I don't want to have to hassle my customer about a
program he may not have heard of or be interested in finding out about.
Also, being closed mouthed about the hardware is a serious mistake of
even larger proportions.

AT&T, it seems, is cruising for a bigger bruising than the HP9000,
another fantasic piece of hardware that has been kept under wraps by
corporate Vogons. Compare the secretive approach of AT&T to DEC's
approach with the PDP11 and IBM's with the PC. Both those machines are
extremely well documented and have become classics in their domain.
Serious software people WANT to know what's on the inside and will place
machines with secret insides at the bottom of their priorities no matter
how many two page spreads they see in Bussiness Week.

If AT&T falls on its face, I wonder if IBM will realize that the follow
on to the PC had better be as well documented as the original PC.
Probably not.

-Zig

guy@rlgvax.UUCP (Guy Harris) (06/19/84)

Fortune may have had other problems that contributed to their sales
difficulties; do you have evidence that their unbundling of UNIX was
a major contributor to those difficulties?  I'd heard about software
delivery schedule slippages and some reliability problems, which are
more likely to upset the average non-computer-jock buyer than the
absence of "ps".

	Guy Harris
	{seismo,ihnp4,allegra}!rlgvax!guy

olson@fortune.UUCP (06/22/84)

#R:rlgvax:-202900:fortune:3500021:000:799
fortune!olson    Jun 21 14:44:00 1984

Enough already!  Fortune (admittedly) had a number of problems
with it's initial release.  Among them was the fact that ps was
not included.

Subsequent releases fixed many bugs and inadequacies, and have
all included ps (also pstat ala V7).  Multi-user systems
even get vmstat.  We may not be perfect, but we do try and fix
our mistakes :-).

	Dave Olson, Fortune Systems
	UUCP: {ihnp4,ucbvax!amd70}!fortune!olson
	ARPA: amd70!fortune!olson@BERKELEY

PS: No argument, a system without ps is very difficult to maintain
when things go wrong.

As to unbundling of software, as has been said, most users in the
market-place do NOT require all that Unix traditionally provides.
As to whether the pricing and packaging of our (or ATT's) software
is adequate and acceptable to users, only time will tell.

mats@dual.UUCP (Mats Wichmann) (06/27/84)

[ Line-eater bug available as a seperate but equal package ]

No PS?

A multi-user, multi-tasking UNIX system needs to have a way to unstick
wedged processes (which requires the process ID, obtainable by running PS). 
Especially programs that take over managment of the screen (word processors, 
for example) have a nasty tendency to die and leave a terminal wedged so it 
needs to be unstuck from somewhere. We can of course be nasty and say that 
anyone who tries to run several processes on a Fortune deserves what they get, 
but it IS possible to hook more than one terminal to their box....I consider
PS to be an essential part of the system.

On the other hand, Guy is probably right that Fortune has not lost a substantial
number of sales because of their unbundling. They lost the UNIX hackers
market (and consequently gave us many of our early sales; thanks guys!), but
they were not going after it anyway, so it can hardly be considered a loss.

As usual, there is no one answer to this - it depends on your market, your
support capability, corporate policy, and the usual things that make up
a business. 

	    Mats Wichmann
	    Dual Systems Corp.
	    ...{ucbvax,amd70,ihnp4,cbosgd,decwrl,fortune}!dual!mats

[ Still not related to Glenn "Rogue" Wichman, who works at Fortune and
  has no cute signoff message ]

mats@dual.UUCP (Mats Wichmann) (06/27/84)

[ Botch away, guys - the rest of us love it ]

Unbundling means that software vendors have to write software that will
run on a minimal configuration. This is not bad news, necessarily. One
of the places where BSD UNIX has traditionally gotten performance
improvements is in converting standard shell scripts, which depend on
executing lots of programs, to C programs that do the work themselves.
Rob Pike may be digging himself a grave so he can go turn over in it,
but the market is not seriously going to demand that the UNIX modular
concept be preserved over getting better performance. Remember, what
*REALLY* sells a machine is solutions, not potential.

Much worse is being close-mouthed over machine specs. The days of the
lockin are on the wane (not to say they won't come back later, though).
One of the reasons IBM made such a killing with the PC was their very
intelligent realization that they could expect better performing software
by letting the details out the the world, rather than trying to do
everything themselves, as they always had done in the past. If they
were to change this policy for their micro systems, they would only
do it if they were *DAMN* sure that they could provide all the quality
software needed all by their little selves, not becuase of AT&T's success/
failure. AT&T, on the other hand, is probably figuring that the wonderful
UNIX System will provide a means for developing software without releasing
hardware details. Seems to me if they let out the hardware details,
someone would write a word processor (for example) that ran exceptionally
well on a 3B, and thus provide more of a lock-in.

Time will tell who is right.

	    Mats Wichmann
	    Dual Systems Corp.
	    ...{ucbvax,amd70,ihnp4,cbosgd,decwrl,fortune}!dual!mats