[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Jay Wurts speech at Slug-90
I "accidentally" left my portable tape recorder turned on
during Jay Wurts opening speech at the SLUG-90 conference
this past week at Stanford. Due to a long and boring plane
ride back to Florida, I had time to transcribe his speech
into this mail message. He had some very interesting
comments and I know a number of people were interested in
getting a copy of what he said. I don't know if anyone will
get upset by this, but here is a transcript of Jay's entire
speech (minus the parts that he actually said "don't quote me
on that"). Some of the contents have already helped me
convince upper management that we're on the right track with
Symbolics. Hope it can help others as well. Please excuse
any spelling errors, there's no Meta-X Spell Region on this
machine and the Symbolics laptop isn't out yet.
Jim Dumoulin
NASA / Kennedy Space Center
Symbolics Management Presentation Speech by Jay Wurts,
President of Symbolics Inc. Presented to the Symbolics Lisp
Users Group (SLUG) at Stanford University on June 20, 1990.
Hello, I'm Jay Wurts and I'm glad to be here. I'd like to
welcome everyone to SLUG. I wish we had weather like this in
Boston all year around. Paul (Pangaro) has introduced a few
of the senior people that are here from Symbolics but I just
want to make doubly sure that everyone knows them so I'll
introduce them again.
I think he mentioned Ken Tarpey, Ken will you please stand
up. We were extremely lucky to get Ken. Ken has been Vice
President of International Service at Computer Vision and
Vice President of Service at Prime Computer (which is a 2
billion dollar company that has been thru some pressure also)
and it was an excellent opportunity for Symbolics to get some
very very experienced management talent in the service area.
He got the service area brought under control very quickly
and in a couple of months was promoted to be in charge of the
entire Genera business unit. Jack Slaven, who most of you
know very well. Jack is an expert of getting customers out
of trouble, so if there's anyone in trouble either partially
or completely brought about by Symbolics, Jack likes to fix
those sorts of things. So, don't sit and stew, call
Symbolics.
Also on the front row is Ken Sinclair. Ken has been one of
our very senior developers and when we reorganized very
recently to put much much more attention on software
delivery, we put Ken Sinclair in charge of it because he had
the best record of anyone in the company of delivering a
software project on time. He was in charge of the UX-400
team and he got that project done in a small number of months
with a very small team. We wanted to take no chances on
software delivery on CLOS and CLIM so we brought Ken into the
effort. Also, who joined that effort 2 or 3 months ago,
because of the extreme importance of it was David Moon in the
back. As many of you know, we are extremely serious about
software delivery and one of the ways to be serious about it
is to put some of the very best of our talent on the project.
Also here is Howie Shrobe, Vice President of Technology and
Teri Carilli, recently promoted to manager of Software
Support. Not here, but of interest to you is Mark Truelar
who has been in charge of all of our hardware operations and
has been recently promoted to be in charge of the Genera
Software running on proprietary hardware. Also, Ron Benanto
is our CFO who has done wonders to keep things running
smoothly thru turbulent times and even though he isn't here,
he is another key member of the team.
To start out, I'd like to talk about the goals we've achieved
during this past year. In spite of our last quarters
financial performance, which looks fairly disappointing, we
really have achieved everything else we set out to accomplish
this year. I don't think we've ever had a year in our
history where we've managed to get everything except for one
item done. We shipped the UX-400, on schedule last summer
and we're shipping the XL-1200 this month on schedule. The
XL-1200, we were absolutely delighted to have met all of our
design goals and its currently running at 3.0 times the speed
of an XL-400 or 5 to 6 times the systems performance of a
3640. We've had announced press releases to that effect and
for some strange reason, none of you believed us, but for
those of you following SLUG mail you don't have to take our
word for it because you've seen some of the benchmarks that
have been coming out on that machine. Particularly the MIT
benchmarks done by the robotics lab, one of the ones I was
very excited about because they were very through. Not only
did they benchmark the XL-1200 against other Symbolics
machines, but they ran them against all of their Sun machines
at the same time, and the numbers were pretty clear. System
performance for a large application, 250 seconds on a 3650,
110 seconds on a SPARC Station 1, 80 seconds on a SPARC
station 4/370 and that 370 had 56 Mbytes of main memory and
55 seconds on an XL-1200. That's a real application, that's
system performance, that's MIT guys and they were skeptical.
At the end of 2 hours of wringing the machine out, their
final comment was "Well, it does what you said it did." He
was a little unnerved at the beginning of the first demo
because when we started an application up first on the 3650
and the 3650 was about a third into the demo, we started up
the XL-1200. This was one of the first 1200's built, and it
went into the debugger. And one of the guys jokingly
mentioned "if you managed to get it out of the debugger
before the 3650 finishes, we'll be interested." Well, we
went into the debugger, fixed the problem, got out of the
debugger and finished the benchmark and the 3650 was still
computing. So we're not talking 40% improvement, we're
talking about a significant improvement in system
performance.
The next major milestone we achieved was the shipment of the
Frame-Thrower image computing board and intelligent frame
buffer. This is a project that has been a glint in Tom
McMahon's eye for 4 years and cost 2 million dollars for the
gate arrays alone. It has 11 custom gate arrays on it. It
has 1024 bit wide data paths. Its is an extraordinarily high
performance image computer and frame buffer. 4 channels in,
4 channels out, 100 Mbits/sec. These are performed in
parallel, one each for red, green, blue and alpha. It fits
on a 9U VME board and runs in one slot in an XL-400 and will
pretty soon be available for other machines. Its based on 4
years worth of experience with the PIXAR and our feelings
that there really was a better way to do it, and we
incorporated many of these ideas into the processor. One of
the most exciting features of the board is its genlock
circuitry which is entirely software controlled and we have
the only device in the industry which can perform NTSC video,
PAL, HDTV or 601 Digital, and the output of these are all
under software control. You can start with one and upgrade
down the road as you need to. In the graphics market, where
we're marketing this product rather early at the moment,
peoples reactions at the trade shows has been, first "How can
you do that, Nobody can do that" and the second comment has
been "How can you sell your whole bundled machine with that
processor in it for $79,000 when the nearest competition is
over $200,000". This board can also support single color in
the XL-400 and the XL-1200 and has in addition to external
sync on this board, a 3 dimensional geometry chip that isn't
currently turned on now but is producing some very very good
results in our lab in producing high performance color
geometry manipulation on the screen. Its running about the
speed of a Silicon Graphics Personal IRIS in 3d geometry even
though that's not the primary purpose of the board. So its
an impressive piece of hardware.
For those of you who have been following Symbolics over
the years, we've come out with many machines in the past
couple of years. We've come out with tall machines, short
machines, thin machines, fat machines but all about the same
speed. We now have a machine that is very very much faster
so you may want to take another look. We've broken tradition
here.
The next major accomplishment has been Genera 8.0 which
has already been shipped and those of you under software
maintenance have it by now. The major purpose of this
release was to re-unify Genera, which we had to divide into
pieces to provide support for the large number of various
hardware products. We managed to get it all back together
again in one compiled system, the same binaries should run on
all ivory machines. I hope this is enough to forever put to
bed 7.0 in peoples memory. It probably won't be, the scars
seem to be fairly deep there, but we're really working hard
on it.
In addition to re-unifying Genera, we added a high
performance version of CLOS and with it we have a Common Lisp
developer which has been generalized at your requests a year
ago to not only support development for CLOE on the 386 but
to support development under ANY Common Lisp platform. This
should keep you out of trouble and allow you to deliver your
application anywhere. Also included, was service support for
multiple ivory boards in Sun servers and we've simplified the
installation process on the Sun boards. We had very very
ambitiously hoped to include CLIM in Genera 8.0 but given how
ambitious CLIM is and that we also decided to break with
tradition and work co-operatively with other Lisp vendors in
the market place, it proved to be too much to get 4 or 5
companies working together to get a product that wasn't even
fully spec'd a year ago into shipment at this point. We have
it running, it is alive, it is being used by IBM, FMC and
Northern Telecom and will be included free of charge in our
next release of Genera 8.1 so the rumors that CLIM will never
exist might have been based on sensible data a year ago but I
can assure you at this moment, the product exists and we're
very excited about it. The team of participating companies
are growing (not shrinking) and we're finally getting the
kind of cooperation needed. This should have happened years
ago and perhaps I should apologize for Symbolics role in the
past which was to frankly make it as difficult as possible
for our competitors to work together with us. We changed our
mind on that and decided since Lisp is the thing, that we
really needed to pull together to keep this great technology
going. This is a very different strategy than what we had a
few years ago and CLIM is the cornerstone for that.
The final goal we achieved during the past year, the most
important one of all to those of us in the Lisp community, we
managed to get some more successful customers. Lisp has been
under a lot of attack these days, and in my mind with
management, the biggest issue is who has used it successfully
to produce useful applications and the list wasn't very long
a few years ago. There was a long list of successful
research projects but not a long list of companies that
really made their investment in Lisp. We are really pleased
the list is longer this year.
Houston Power and Light is saving 14 million a year with
an interactive planning and scheduling system based on
Statice and Genera and they are absolutely delighted. Dick
Alfred is an older gentleman with white hair, general manager
for maintenance at Houston Lighting and Power. If you talk to
him he starts out by saying that he personally has been doing
power plant maintenance for the last 200 years and Symbolics
was the first computer and software people that he's worked
with that he hasn't wanted to throw out of his office. In
particular, what they've done, has been to collect all their
scheduling data off their mainframes, with our networking
ability, and put it into Statice. Looked at it every way
from Sunday, using about 5% AI and the rest traditional
programming, have it put into schedules in real time using
Genera and the presentation system to display it on the
screen. The interface is so good they can display it in a
War room on a wall for their work foremen to come in and
negotiate the schedule. Based on that, they do their
maintenance planning for several years in advance. It used
to take them 3 weeks to do their maintenance planning and the
first time they used the system, they managed to do the
planning in one day. The biggest advantage, though has been
that they've never had before very much buy-in among their
people as to whether or not they were going to make their
schedule. In fact, their previous track record for making
the schedule was 15% of the time and that record has now gone
up to 85% because of this system. The reason for that is
because the schedule can be viewed in real time, on the wall,
with graphical interfaces so the foremen can see what's going
on. In fact, and electrical foreman, the first time the
system went into production, slipped one of his projects from
4 days to 7 days and the whole schedule shifted and the power
went down and the megawatts went down and the overtime went
up and he said "My God, I had no idea it made so much
difference. You never told me it mattered. If we have to,
my guys can do it in 3 days." Big savings. Lots of
visibility. In hard times, the first budgets to go, as you
probably know, are the research budgets and the last thing to
go is anything that can save money in operations. We're not
really building many power plants any more and their number
one budget item today is maintenance. So their really
anxious to put systems like this into operations.
Its interesting to note that we couldn't possibly have
done this without the power of Genera. They showed me the
first user interface they started with and I looked at it.
Then they showed me the one they are actually using and they
visually looked the same to me. And I said, what's the
difference. Why did it take us 50 iterations to go from the
first to the current and they said "Well, we really didn't
know what we wanted." Then I said, "Well, whats the
difference. The new one doesn't LOOK much different than the
old one" and Alfred said "the difference is we can USE the
new one and we couldn't use the old one." Anytime when
you"re talking about a graphical user interface that can be
used by real operations people, the ability to rapid
prototype and use presentations and work closely with the
user to get it right is priceless. They wouldn't have paid a
nickel for the first generation of that system and they saved
4.2 million dollars using the final iteration. They now want
to put a Symbolics computer between them and every mainframe
they own so they can get some real work done.
The second application success story is American Express.
As you know, they've been doing a validation system for the
Green and Gold cards using Symbolics machines. They've got
48 3620's approving Green and Gold cards saving 47 million
dollars. That is, their computer room is full, they don't
have any more space and they need more machines and we're
trying to figure out a way to get some XL-1200's rack mounted
for them so they can double their stack. (Heckler from the
audience asked - "Couldn't you get them to remove just one of
those IBM memory boxes to make room?) This is IBM's computer
room, mind you. IBM insisted in making a run by doing the
validation software for the Optima card on the mainframe
using Aviance software and they tried very hard but currently
have not succeeded. Now there's talk of moving that card
over to the Symbolics as well. One of the side effects of
the Symbolics machines is that they're arranged in a fault
tolerant network and it doesn't matter how many of them fail,
the system will remain running and that's a big plus for the
system. IBM estimated it would take 6 3090's to get the same
level of fault tolerance.
The reason I'm talking today about these applications is
because I know you guys need help with your management to get
the OK to use this technology and I've found that examples
carry a lot of weight with managers. The next major success
story, which we're not going to be able to quote in the press
just yet is (editor-note: secret stuff left out). They are
saving millions of dollars.
And finally, on the lighter side, if you attended Apples
introduction of the FX Macintosh, the 90 second video playing
behind John Sculley, showing the animation of the machine
converting from the Macintosh CI to the FX was done entirely
using Symbolics Graphics and Paint software. They gave us
credit for the software in the presentation by saying
"Hardware by Apple, Software by Symbolics" and when Sculley
showed us the tape when it was finished, I didn't have the
heart to tell him that it was really only their power supply
that was contributing to the video. Apple is a fairly fat
company (hardware wise) at the moment and at one point, when
we needed to do the rendering and we wanted to use 12
Macivories to do the final rendering because they wanted to
be able to redo it if they wanted to change anything. We
said, "You get 12 Macintoshes and we'll get you 12 MacIvories
but we'll need a switch to be able to switch a monitor so we
can checkup on the computers while their doing the rendering
all night. They said, "well, we can't get you a switch,
would it be ok if we got you twelve 19" monitors instead!"
We said sure.
Well, those are 5 customers successful in a very big way
and they're a lot more successful than they would have been
attempting to run ordinary software on ordinary computers.
As soon as we can get releases on some of these other
customers, we'll get those out to you. I strongly urge you,
if you've got success stories, that somehow, get permission
to talk about your particular applications. From your PR
department or from ours. This is a battle of word of mouth
to senior management. These are big successes and every one
of these companies will be more competitive because they are
using this new technology. We've got to get the word out and
maybe the pendulum will swing back the other way.
In terms of (the size of) our company, I'm pretty
pleased with our list of accomplishments in the past year.
I'd have been even more pleased if we would have achieved 2
other goals. The first one was that we did not meet or sales
goals on the UX-400, for a variety of reasons. Last year, we
asked this group at SLUG how many of you wanted it and a lot
of hands went up. We asked how many of you could find the
money and a lot of hands went up. We asked you how many of
you could find a Sun in a hurry to plug it into and a lot of
hands went up. We didn't get delivery on all those hands.
One of the problems was we rushed the product out without a
Beta test because we felt it was comparable with the XL-400
and we had some glitches on installation. We required tech
support upon installation which made it difficult, especially
if you were installing it on a borrowed Sun. Sun
Microsystems, of course, picked that opportunity to screw up
their computer systems and were not able to ship as many new
systems. Also, you guys surprised me by wanting to buy the
whole Sun from Symbolics instead of buying the Sun from Sun
and the board from us. We figured a technical crowd like
this would love to put together their own machines and it was
a shock to us to learn that many of you preferred to buy the
complete system from us. Anyway, that was one of the
reasons, in addition to this wonderful economy we were in,
for our financial losses this quarter that set us back
further than planned. We expected the UX-400 was not only
going to carry us by making it a lot easier for you to deploy
Symbolics machines but we hoped it would get us into other
Unix users that were previously not using Symbolics machines.
Our impressions so far were that Unix users are almost as
addicted as we are to our technology and it didn't accomplish
what we wanted.
The last goal, which I think we've all failed at is that
we haven't attracted new businesses to this technology within
the last year. We've been scratching our brains, trying to
do everything we can think of to figure out how we can make
the product line more successful and the only thing we can
think of is to get more successful applications out there and
bring in more people that way. That brings up a very serious
effect hanging over the heads of us all. Sooner or later all
of you guys are going to die off and we have to have an
adequate supply of people coming in. You all are going to
use up the number of lines of code you've got in your body, I
don't think we'd be able to agree on a number but I think we
can all agree that once you get past that (of course, you can
probably do more Lisp lines than you can assembly or "C"
lines) but sooner or later, you're going to exhaust the
number of lines of code in your body and we have to have an
additional supply of people coming in. Any suggestions on
that score would be very helpful.
In terms of future direction of Symbolics, a year ago, I
said we were going to concentrate on making our machines very
much faster and improve the software for our current
customers and we were going to concentrate on applications.
A year later, in spite of our UX battles, and the stress of
bringing in the revenue every quarter, we are much further
along on that strategy. Basically, there are three or for
points to it.
The first point is that in spite of our TE's efforts, we
are not close enough to our customers and our future revision
of our product has been largely inspired by our perception of
the technology and we are going to continue to stay on top of
the technology and we are trying to bring in new technology
to the world, but at the same time, we're going to get much
closer to the customer and actually get involved in
applications and try to get inspired in future improvements
we make into the system. In order to do that, we've
established a list of valued customer we consider critical
and we're going to try to get a lot of people from Symbolics
to work closely with those people and work under the
assumption that if we make them happy, at least they'll buy
it. If we pick them right, at least the same stuff should
apply to the majority of our other customers. This is a big
cultural change for Symbolics. We've made a lot of progress
in 2 years and we still have about 2 years more worth of
progress to be made to get where we need to go.
The second thing we're going to concentrate very hard on
is applications. Three applications, in particular to
attract new customers. We're finding it very hard to go head
to head with "C" Customers and explain why this technology is
so much better. Instead, we're going to concentrate on some
applications because we can do them better with our
underlying technology. We're not talking about changing the
underlying technology, but just camouflaging the technology
so people can better understand the underlying design. The
applications we have chosen and are working on to support are
graphics, hypertext/concordia and interactive planning and
scheduling based on statice.
We picked these applications because we have customers
who have been very successful with those applications and
also because they use all of our underlying technology to its
fullest and we believe we can continue to evolve that
technology. I observe that Genera was not invented first as
a technology, then went out looking for applications. Genera
was invented because there was a customer with a real problem
at MIT in the AI lab and half the AI lab really wanted to do
AI and many of the other half became the Symbolics founders
and they wanted to build systems to help those people do AI
(so they wouldn't have to do AI themselves) and out of that
came hardware and software for doing AI applications at MIT
that turned out to be real useful in a lot of other
applications. It was then relatively straight forward to
build a company and generalize that approach to the world.
What we're going to do in our next generation is going to
concentrate on these three applications. Its been my
experience that whenever you put world class developers that
close to an application, they solve much more than the
application they are given and out of that comes all sorts of
technology that's useful.
In the Graphics area, the three markets that will be the
area of attack are first Video Film animation (the first one
where we've really had our foothold) then moving in from
there to CAD visulation and then from there into Print using
new High Resolution High Definition Photo Retouching. So
that should keep us busy for quite awhile.
On the Hypertext front, we're first starting out by
selling Concordia to other computer companies so they can
have as good as documentation on line as we have and we're
well along with 1 company in particular. What has helped
this is what I call a no excuses document examiner. This is
the document examiner that runs on a plane vanilla Mac. Its
in "c", has no server, no ivory board, just plain with all
the capabilities of the Document Examiner. Anyone that
builds documentation in Concordia can be sure they can
deliver that documentation anywhere. We'll be porting that
application to other machines and we're interested in hearing
from SLUG what those other machines should be. The real
market for Concordia though, since there is a limited number
of people that want to deliver on-line computer
documentation, is in operational decision support systems
that need a lot of pictures and a lot of data and they need
to retrieve it in a hurry. In which case, whatever it is,
they have to find it and get to it quick. Alcoa is doing
pioneering work there and we're really anxious to see if
other customers can find other applications as well. The
world is on information overload and Concordia is a great use
of the presentation system and is a great way to display what
the Symbolics system can do and we are planning to use this
as a beach head into companies to show them what the
technology can do and then spread out from there.
Interactive planning and scheduling uses everything
Genera is good at and we've had some really successful
applications with Houston Lighting and Power and some airline
companies. We're close to bringing in two or three other
large airline scheduling systems. Its interesting that the
actual scheduling part of that is only 5% of the application.
The rest of it is really Genera. Its object orientated
everything, presentations, Josuha and AI, and the networking.
So we think its a pretty consistent set of applications that
use all of our strengths and can complement our current
customer base and what we're doing for them.
In terms of you as our current customer base, we're going
to speed up our hardware. Not a single person has yet asked
us for slower hardware so we're going to continue to make the
hardware faster. In addition to that though, unlike our past
history where we have come up with every scheme in the book
to keep everyone using our proprietary hardware, the big
change with tradition is that we are not going to be
depending exclusively on proprietary hardware in the future.
I mentioned our software and delivery focus and you've seen
how much we've put into CLOS and CLOE and we're working
closely with the Unix Lisp vendors so that you can develop on
Symbolics and deliver ANYWHERE. We think its very important
to put NO obstacles in the path of delivery. In addition to
that, we've reached a fundamental milestone this year with
CLOS and CLIM. Up until now, whenever I've asked our
developers couldn't we every take the Genera development
environment and rebuild it so that it would run on standard
machines, the answer has always been, "It would be very very
expensive and very very hard." The story is changing with
CLOS and CLIM. It is now a whole lot more practical to make
these kinds of capabilities available on standard machines.
We are in the midst of a major negotiation with a large
computer company that I can't name, that is also very
interested in seeing a Genera like development environment be
their flagship product. This company has a tradition of only
settling for brand names and nothing else and is very anxious
in participating in the funding of this project. We got the
project down to much more manageable proportions because of
CLOS and CLIM is where they are. I hope sometime within 6
months you'll be able to hear the specifics of that so you
can be sure that you'll be able to have the symbolics
development capability on your favorite Unix box without
having to add extra hardware. I suppose I could tell you the
company has 3 letters in its name but that wouldn't rule out
many of the major competitors these days.
The last thing that Ken Tarpey will go into more detail
about, that is part of our strategy, is really a major change
in our software support by moving it from the west coast back
to the east cost. It should have never been on the west
coast to begin with. It was moved out there by a previous
management team which had a house in Malibu, and for some
strange reason that wasn't the slightest bit technical,
wanted to be closer to software support. We were reluctant
to move it back because we were afraid of changing so many
other things at once and we knew people were being concerned
about service and we were afraid of having too many
disruptions during the change. At this point, we feel we can
handle the change, we've moved it back to Cambridge (I mean
Burlington. We still affectionately referred to it as
Cambridge) and we've gotten some very excellent developers
involved in the support and we're rotating many of the rest
of them thru on a weekly basis. You should see quicker turn
around as soon as we get thru this transition period on some
of your questions because you won't have to wait for the
delay due to timezones across the coast. So I would strongly
urge you, those of you that are not under software support,
to take a closer look at it in a month or so, when we get
fully up to speed. It could be a very economical expenditure
on your part to get access to some of our best developers so
don't abuse the privilege and ask for the best developers
only when you have hard questions and let anyone help you
with the easy questions and we'll be able to give you really
first class software support. We will also learn a great
deal from you about what we should be doing to improve the
product at the same time.
I'd like to review the few points we've gone over and
I've made a list of what I really wanted to get across today
and let me sum it up as quickly as I can. First of all, we
are here this week to listen to you, to find out whats on
your mind and to find out what we can do. If we want Lisp to
succeed, one of the ways is for Symbolics to go out and one
of the ways we're going to do that is to listen to you.
We're not here to tell you what we know better than you. We
want to hear from you. The second thing is, I want to make
sure that everyone understands that the XL-1200 is the
opportunity of a lifetime to upgrade your machine. Any of
you that don't think you can get the funding for the XL-1200,
even though we've tried to make it as easy as possible,
(we've priced it less than a Sun 4/370 and we've allowed you
to keep your monitor and mouse and just send the box back and
we'll send you a new XL-1200.) You can practically justify
it on your maintenance expenditures but any of you that have
very hostile management that can't get approval for this
machine, I strongly urge you to NOT go upstairs and demo and
see this machine run. You guys will make yourselves
miserable. You guys spend your lives in front of these
machines, and this machine is 5 times faster than the machine
you're probably sitting in front of. You don't want to know
first hand what its like. Its a little like skiing out west
in the Rockies. Once you come out here to ski, you never
want to ski in the east again. A couple of words of advice
from the management people I've been talking to with
customers, one of which was Jerry Fishman, the number two guy
at Analog Devices who happens to be on our board. I said,
Jerry, we've got this machine that's five times faster than
what you've got and are you going to buy any of them. You've
got really old machines and you're buying refurbs. Are you
going to buy any XL-1200's and he said "Yup, I'm not going to
approve any of them, but we're going to buy them." And I
said what do you mean and he said "Five times faster? My
people would kill me if we took their Symbolics machines
away. I keep telling them I'd like to see them go to
something cheaper on standard machines, but I can't convince
them. Their not going to ask me, they'll find a way.
They'll put it thru the system somehow. They'll lease it or
they'll do something else and I'll never see it, but come
back in 6 months and they'll have XL-1200's, you see." And
sure enough, the orders are coming in from Analog Devices and
Jerry hasn't signed any of them yet. When I told Jerry how
much he could save on maintenance, even he got interested. So
take a good look at that machine. Very reliable, very low
maintenance, VME Bus based, Ivory processor, plugs in
standard peripherals. So if you're ever going to get cut off
by your management, you might as well be cutoff with an XL-
1200 instead of a 3600.
The third thing I want you to understand is that we want
you to deliver anywhere you want, any way you want, whether
we make money in it or not. We don't want you to develop
applications and not be able to deliver them. If you don't
deliver on MacIvories and you don't deliver on XL-1200's and
you don't run CLOE, I don't care how you do it, DO IT. Just
Deliver. And if you need help from us, we'll help you get
these applications out there delivered. Get some success to
your bottom line so you can get your budgets up and we can
all get the credit and make all of our lives a whole lot
easier. That's a change from 2 or 3 years ago but we are
very serious about wanting you to deliver.
The next point I want to make sure I get across is that
We're very excited about a development environment for
standard hardware machines. We have made a lot of money in
the hardware business over the years but we've also got
ourselves pretty frustrated. We tell people how wonderful
our stuff is and we've always heard this But..But..But. and
if we can get away with decent performance on the standard
machines we'll do it. In the long term, we can then get this
technology into quite a few places in the world without
having to say But. We are optimistic that if we can provide
that solution even if its a couple of years out, that you
folks will continue to support us and buy machines in the
interim. When Computer Vision announced that they were going
to switch from proprietary hardware to standard hardware,
everyone said, "Ok, I've always been able to justify the
machines in the short term, it was always the long term that
was killing my management. Now that I know there is a long
term out there, its ok, so that's fine." That's what Jerry
Fischer said when I asked him if he could be excited about
having a standard Unix workstation that could be running the
Genera product he said that great. He said he could still
justify the machines in the short term based on their current
benefits.
Another thing I want to make sure I get across is that
nothing succeeds like delivering results. When I go out and
visit customers, I can see the customers that are getting new
money and new applications and new people and their the ones
that have delivered something. You go into a place like
Intel and they have their whole blackboard covered with their
plan to upgrade their whole machine. They've got new people
coming into the group, their looking to grow 50% this year in
staffing and its because they've got some real successful
applications out there. That's the key to success.
Finally we, like you really want this technology to
succeed. We're at least as addicted to it as you are.
Everyone of our applications is programmed in Genera.
Everyone of our programmers are addicted, none of us wants to
go back to working on ordinary machines. We are not going to
give up. We are going to get this technology out there so we
can use it, so you can use it. No matter what, its not just
a matter of money and our 24,000 share holders, its a matter
of us and those of us who stuck with it and those of you
whose stuck with it that want this stuff to succeed.
Thank-you.