One weekend in 1995, Steve Perlman tested his theory
that the Web could look as good on a TV screen
as it did on a computer monitor. In 3 days of roundthe-clock
effort, he built a thin client for surfing the
Web, using a television as a display. He invited his
friend Bruce Leak over to see what he'd built, and
they knew right away it was a big enough idea for a
startup.
It was a natural project for Perlman, by then one
of the leading experts on display technology. At
Apple, he helped bring color to the Mac. Later, at his
first startup, Catapult Entertainment, he built one of the first systems for network
games. Now he wanted to bring the Web into people's living rooms.
A little over a year after that first prototype, Sony and Philips sold the first
WebTV set-top boxes to the public. In 1997, WebTV (now called MSNTV) was
acquired by Microsoft for over $500 million.
Livingston: Take me back to the weekend in '95 when you built the WebTV
prototype. How did you get the idea? Why did you decide to do this?
Perlman: For many years, I've been interested in making television interactive.
What I mean by "interactive" is something beyond just changing channels up
and down, to get it where people can have access to content that's more interesting--to
be able to find what they want and then to be able to view it on
demand. For example, what we now consider to be DVR, or what you do with
your TiVo. At the time, it was considered something you'd only do in an editing
suite. If you were a network professional, you might have a disk-based digital
editing system.
I wanted to do all those things, and I even did a lot of the work at Apple. In
fact, just a month ago on the History Channel they showed some of the early
stuff I did at Apple. It was 1989. I was showing a system where we had video on
173
Steve Perlman
Cofounder,WebTV
13
CHAPTER
the screen, images moving around, and animation, and several video sources.
You could pause, rewind, and manipulate the things. That was a big prototype
system, but we could never get it out the door because there wasn't enough
content to drive a system like that. You could theoretically bring in live video,
but in 1990 there wasn't a hard disk big enough to hold live video. Theoretically,
you could try to create all sorts of content for it, but who would ever create all
the content if there are no devices to receive it? So we had a chicken-and-egg
problem. Nobody would buy the devices because there was no content, and
there was no content because the devices weren't out there.
But there were lots of offshoots from that work; QuickTime came out of
that work. We took the video decompression technology, developed it, reduced
it to just a software algorithm, and that was turned into a product by Bruce
Leak and his team. A whole bunch of other things grew out of it--some of the
video products from Apple and so forth.
Then, at General Magic, I went to work on a PDA--but I worked half-time
at General Magic and half-time I was still working on how to make inexpensive
delivery systems on a television for interactive TV, and work with video and
games and things like that.
Livingston: You worked on your own projects on your own time?
Perlman: On my own time. I relinquished half of my stock options. I worked
out a deal with them where 2 and 1/2 days a week I worked on my own stuff,
2 and 1/2 days a week I worked on General Magic stuff. And then what happened
is General Magic, in my last year there, said, "Hey, we want to do video
stuff too." MagicTV is what they called it. So I worked full-time then to try to
create an interactive system for them. But they ran into financial difficulties
and other problems getting the product out, and shut down the MagicTV
effort.
I said, "OK, it's time for me to move on." That's when I first started--and
cofounded with three other people--Catapult Entertainment, which made a
modem for Sega and Nintendo video games that would modify the execution of
the games, so people could play existing titles with each other over the phone
line. That involved building out the network infrastructure to connect people
together--remember, people didn't have the Web in their homes back then--
designing the hardware, and also reverse-engineering the games. So I learned a
lot about the consumer market and about getting stuff out into stores. From the
founding of the company to the point where the product was on store shelves at
Toys"R"Us and the network was up and running, was 6 months--including custom
silicon that we did, as well as shooting the plastic molds for it, boxing it, and
getting it through distribution.
Livingston: And you did it in 6 months?
Perlman: Six months. We reverse-engineered four video games: NBA Jam,
Mortal Kombat, a hockey game, and some other one. We were just working
around the clock, literally. What I would typically do is not sleep for 2 nights;
then I would get 4 hours of sleep and go back to work for another 2 days in a
row, and then get 4 hours, and so on.
174 Founders at Work
It was the hardest I've ever worked in my life. Sometimes I'd take
10-minute cat naps by just laying my head down on my shoulders--just so I'd
get some REMs. As soon as the dreams come, it resets your brain a little bit and
you're able to work again. We were sleeping at our desks. People would bring in
pizza. My wife would sometimes cook some turkey meatballs and spaghetti in a
big pot and then bring it over, and everyone would just chow down.
Livingston: Surely your wife was nervous about you sleeping only 4 hours every
2 days?
Perlman: She was. She got one of those fold-out futons that would fold under
my desk. She didn't like me sleeping on the floor.
My admin, who came with me from General Magic, tells stories about coming
in in the morning and trying to clean up. She'd pick up a folded pizza box
and get scared because she'd find a guy sleeping underneath it--it was covering
his face. It was really bad. My dog, when my wife would bring him over, he
would find burritos, because the place was just a pigsty.
But we had the product out in 6 months because we knew we had to meet
that Christmas. It was out by September.
Livingston: So you had a deadline?
Perlman: We had a hard deadline. But, it was a great learning experience for
me. The guys that we hired to get our network software working, they just did
not deliver. They couldn't work on that kind of schedule. So we pulled it in and
did it all ourselves. It was a matter of just cranking it out.
We used a programmable gate array that we could then freeze into a permanent
gate array to make it cost-effective. That was the only way we could get
the hardware working that quickly. Then it was just a matter of hard work on
the games and everything. We partnered with THQ, which is a video game
company who had a distribution channel to all the video game retail outlets, so
we could get the product out quickly.
I also learned about working with people, because one of the guys I
cofounded it with, it just didn't work out between us. He had his perspective of
where he wanted to take the company; I had mine. I realized that these things
are like a marriage. When you cofound something, you've got to have people
that have a similar kind of perspective on where you're going to take the thing.
Otherwise you're just locking horns all the time.
Livingston: Had you worked with him before?
Perlman: I knew him before. General Magic was developing products for Sony,
and Sony was particularly interested in MagicTV. I'd known him at Apple
because he'd done some industrial design there. He went and got his MBA and
then went to work at Sony. And so I was seeing him at Sony. We weren't friends,
and I didn't know him very well outside of work, but, when I left, he said,
"They're shutting down MagicTV. So what are you going to do?" I said, "I don't
know. I had an idea for this thing I wanted to do with video games."
Steve Perlman 175
I had figured out a way to make existing video games work online--you
know, a two-player game like NBA Jam--we hacked it so the software, instead
of looking to the second controller, actually would set up a link through the
dialup connection to another box, and the two kids were able to play each other.
And of course they didn't have to buy new software because we were working
with game software that's already written. Great way to bootstrap an online
game thing.
Of course, we were way early for the online game market, and we were at
the tail end of the cartridge market. There were a million things I learned from
that, because it ultimately did not succeed as a business. Financially it was OK,
but as a business, it was not successful.
But the biggest lesson I learned was: I wasn't getting along with this guy and
it was time to move on. So I stayed there for about a year. We started that in the
spring of 1994; I left in the spring of 1995. And then I was very tired. I was
physically, bodily tired, as you can imagine after such a hard effort.
I was determined to just go and tinker for a while and explore things. I saw
Netscape 1.0 and thought, "The World Wide Web is kind of cool." I'd been on
the Internet since college--then it was the ARPANET. Back then, the
ARPANET only connected up a few institutions, but through the years I continued
to use it as a software engineer might use it.
Livingston: Were you an engineering major?
Perlman: No, I have a liberal arts background. My engineering background is
as a hobbyist. I built a computer when I was 16 and then designed a graphics
display to go with it and things like that. I'd read Kilobaud magazine and Byte
magazine, and I'd go and print up some company letterhead, which I'd send to
the chip companies--that are now people I work with officially--and I'd say,
"Hey, we have great plans for new products. You should send me some samples."
So I'd get all these chips for free. The ones I could get for free, I'd design
circuits around their capabilities. They weren't the ideal chips, you know. But
what are you going to do--you're a kid in high school; you had no money.
I was in Connecticut and everyone else was in California, so I was 3 hours
off. I ended up shifting my schedule and actually was getting up around noon
because that's when stores would open: Jameco Electronics would open at 9:00
a.m. in California, which is noon in Connecticut. So what's the point of getting
up before noon, right?
I've always been a hobbyist, and it's one of the reasons I kind of seamlessly
go between software, hardware, networking, and material science. I don't
care--it's whatever it takes to make the damn thing work. I don't have much
formal education in these things, but you learn. You build enough stuff; after a
while, you see it. And if you reverse-engineer enough things, you learn what
other people have done.
I designed a software-based modem when I was in college and I got an F for
it because the professor said it would never work. But I got it working at my
first company. The professor was quite nice about it. I sent him an email later
on and said, "This email is being sent to you on the modem that I designed at
176 Founders at Work
Columbia." And he said, "We try to make the right judgments and we don't
always. I'm glad that I did not dissuade you from continuing on with its development."
I thought that was a very nice thing to say.
Livingston: So you leave Catapult and say, "I'm just going to tinker around and
see what happens?"
Perlman: Netscape 1.0 comes out. I get it working, and I said, "Wow, this is
really great," because people are putting up websites that anybody can go to. I
went to campbellsoup.com, and there was a Campbell's soup can and recipes. It
was the early days of the Web, so there wasn't too much, but I thought, "The
kind of people that would be interested in these recipes probably aren't using
computers and connecting to the Web."
Remember, this is before a lot of people got computers in order to get email
and be on the Web. And then I thought, "This could be the thing I need to
break that chicken-and-egg problem." Because if I can get these pages that
were really designed for PC screens to work on a television screen, then... It's
not ideal content; a lot of it is stuff really suited for someone on a PC. But some
of it, like this Campbell's soup site--and there were many other sites, music
sites and all that--is suited for the casual television entertainment experience.
That might be enough to bootstrap us so we could do what I really want to do,
which is these richer--what we now call broadband--interactive experiences.
Things like DVR and so forth.
Before Apple, I was at Atari and Coleco. I designed video game systems
there, and I knew an awful lot about how to create a very high-resolution image
on a television screen by doing special image processing. If you try to put a
high-resolution image on a TV screen, it's interlaced. Interlaced means it draws
all the odd lines in 1/60th of an second, and then it draws all the even lines. If
you have a continuous-tone image--the kind of image you see in the real
world--and you capture it with a video camera, your eye, even though the
whole screen is only refreshed 30 times a second, will look at each of these individual
fields, all the odd lines and all the even lines refreshed at 1/60th of a second,
and think it's flashing 60 times a second. At 60 times a second, if you stand
back in the room, it's your foveal vision; it seems like a non-flickering image. So
you look at a TV, and it doesn't seem to flicker.
But, if you now put content in one of those fields and then very different
content in the other fields--for example, take black-and-white horizontal lines
as you might see at the top of an old Macintosh window, and you put that on a
TV screen, it flashes like crazy. In fact, it can put an epileptic into a seizure; it's
that bad. So what they would do before is only have the TV draw half the lines
vertically. All the video games back then, instead of having 480 lines, they would
only draw 240 lines. I had figured out techniques where I could do image processing
on images that would be intended for a computer where they would be
smoothed out in such a way that you would not see them flicker. They would
look extremely sharp on the TV, but they would not flash, so you could now do
a high-resolution image on a TV. The technology was in some of the
Macintoshes, but not many people were hooking up Macs to TVs.
Steve Perlman 177
The other thing--and this is an interesting point--back in the 1980s, when
I developed this technology for Apple, software patents were not things that
people filed. It was mainly hardware patents. Later on, people started filing
software patents. The reason is software was considered an algorithm, and an
algorithm is not patentable. A Fourier transform is not patentable. It's considered
a mathematical function. This technique for stabilizing the image--the
basic underlying principles of it--were things that patent attorneys said we
couldn't file patents on, so it was open for anyone to use.
But still, the way I did it at Apple wasn't enough for what we needed to do
with the Web. We had other things to accomplish, and so what I did was take
those basic ideas and added on a whole bunch of other stuff and filed some
basic patents around it. I knew that it was possible to take an image intended
for a computer screen and get it to work on a television. So I went to Fry's and
got about $3,000 worth of parts and built something over 3 days and 2 nights.
(Much like I was working at Catapult. Back then, that's the way we worked.) I
then got this image up of these web pages on a TV, and it looked perfect. It
looked just like the image looked on the computer screen. I grant you, back
then, computer screens were largely 640?480 and web pages were a little bit
smaller and so on, so it did happen to work for the time and place we were in.
I called my friend Bruce Leak, who I mentioned before is the guy I worked
with at Apple. He had taken a lot of the technology that we had developed in
the Advanced Technology Group, like QuickTime and also the color
QuickDraw stuff, and then developed these technologies into products. We had
a good partnership working together. He was at another startup at the time,
Rocket Science Games. It was the middle of the night--it was midnight or
something--I called him up on his cell and said, "Bruce, get your ass over
here." He said, "Why?" And I said, "I've got something to show you. I'm about
to pass out."
So he comes over and looks at it and says, "Well, so what? What did you do
to the TV set?" And I said, "I didn't do anything to the TV set. It's what I did to
the signal going into the TV." And he's like, "No way!" And I said, "Yeah!"
I remember he said, "Man, we've got to form a company." And I said, "Ah,
yeah." I think that was the first moment I even thought about it. Then I was
thinking we should get a good name for the company, and immediately we
knew it was going to be called WebTV.
After that, one thing kind of led to another. We were able to attract Phil
Goldman to come, another top-notch developer. He created MultiFinder for
Mac, and he wrote a lot of the OS for the General Magic device.
Then we went to Marvin Davis, a wealthy financier in Hollywood. He had
made a lot of money because he invested early in Catapult. As I said, Catapult
was financially successful although it was not successful as a product. He told
me that, whatever I did next, he wanted to put money into it--because he had
turned around his Catapult shares and sold them to Viacom and made some
outrageous profit in about half a year. So I went down to Hollywood with Bruce
to meet with Marvin Davis, and we demonstrated WebTV to him--the prototype
I had--on a TV set in his office. I'm not sure he immediately saw what the
178 Founders at Work
value of it was, but he nonetheless committed to put some seed funding in. We
ended up raising $1.5 million from Marvin, and that's what we started the company
with.
That was in July of 1995. I think I got the thing working first in April 1995,
so from April to July, I kind of pulled together the business plan and at least the
first couple of guys that were going to help me, spent a lot of time calling different
people who we might be able to work with, went looking for office space,
and so on. We were working out of my dining room in my house.
After we got the money from Marvin, we went and found an old BMW dealership
that was vacant. It was mostly garage, but they had a little bit of office
space. There was no connectivity there; I think there were three phone lines
going into it. But, it was about 90 cents a square foot per month, so I thought,
"OK, perfect." It was right near downtown Palo Alto, and so we moved in there.
Literally, we had three phone lines. There was always one of them with a dialup
connection, because we were doing experiments and everything. I was trying to
do business calls on the other one, and there were modems always interrupting.
We finally were able to convince Pac Bell, the phone company at the time,
to bring a T1 line in there. I remember talking to the guy and saying, "We want
a T1 line here. We've got this big business we're growing. We're eventually
going to need very high bandwidth connections and optical fiber, and all this
kind of stuff." I hear some paper flipping in the background, and he says, "Is
this some kind of a joke? It says here on the manifest that this is a car dealership."
And I said, "No, no. We're running a big online service business. It's
going to affect people all over the United States. It's going to be really huge."
The guy says, "OK. Who put you up to this?" It was like a joke trying to get connectivity
there. We literally had to go to several levels up in Pac Bell until they
finally believed that we were a startup using an old car dealership to set up an
online service.
Livingston: How many cofounders did you have when you started?
Perlman: There were three total: Bruce Leak, Phil Goldman, and me. Phil
passed away 2 years ago of a sudden heart attack, sadly. That was a real tragedy.
So then we started hiring people and getting things going. I'm doing something
that I wasn't that familiar doing, which was business development. That
was all new for me. As I said, I may not have an engineering degree, but that's
what I've always done for my vocation.
I called Sony and said, "Hey, we should go and do this cool thing." Sony was
interested--I networked through some of the contacts I had made at General
Magic, but they were slow getting through the company. We also began to
speak with Philips. Sony finally said they wanted to go forward with WebTV, but
they'd have to be exclusive for a year. They'd brand WebTV with a Sony logo,
and they'd distribute it through their stores, and so on. But we could begin to
have other licensees for the technology after a year. So we told Philips that they
would have to wait a year, even though they were all hot to trot. At the time--
and probably still today--Sony was the stronger brand in the United States.
Steve Perlman 179
Then we went to raise more money. The Davises had committed to $3 million
that was going to be in tranches. We had $1.5 million, but the last $1.5 million was
contingent on us closing a deal for a consumer electronics partner...
Livingston:... who would manufacture it?
Perlman: If we could get the deal through Sony, they would manufacture
WebTV. Sony's a big company. It takes a lot to get through the system there, and
we just could not get the deal through the system. As hard as we tried--they
were almost ready to go--we couldn't get a commitment. So we went back to
the Davises, and they got very nervous. They don't know about technology, and
they said, "Well, we're only going to put in a million and a half."
Well, now we had hired all these people--I think we had over 30 people
then. Though we were quite frugal, it still was a high cash burn. We were just
about out of money. So I mortgaged my house, liquidated all my assets, and
brought in all the cash I could to help it. (Although I did make some good
money from General Magic and Catapult, it wasn't until after that point. Both
companies did their IPOs after that. There was a holding period for General
Magic, and so on.)
We didn't tell the employees that we were running low, because we didn't
want people to be in a panic. We were going to tell them if we were really hitting
a wall, but I could keep the company going a little bit longer.
Then we started going and talking to other investors and VCs, much sooner
than we thought we were going to have to.
Livingston: Because you were expecting the second tranche?
Perlman: Yes. These days I look at it, and I think, "Jeez, even $3 million is a
fairly modest amount for the scope of thing that we were trying to do."
I remember we spoke to one semiconductor company that we got very far
along the road with that made a processor. When it came down to literally days
before we signed the investment document, they added in a section that said
we would be obliged to use only them as the provider for all of our silicon. In
other words, they set it up so that our backs were against the wall, and they
were getting us locked in. We knew that, if we were locked into one provider
for silicon, we would have no way to negotiate prices. That would drive up the
cost of the unit, so we couldn't do that. We even tried to explain to them that,
"You guys are investing in a company. You don't want that to happen." But they
felt very clever about this strategy and taking these wet-behind-the-ears entrepreneurs.
So there was another 2 months wasted. We were watching the bank account
dwindle. Then we started speaking to VCs, and we talked to a whole bunch of
different ones. We spoke to Paul Allen at Vulcan and a couple of other companies.
We talked to Sony and Philips about possibly investing, but they weren't in
a position to invest. We found that nobody was willing to make that first step. In
fact, I think a lot of them were sort of like vultures waiting for us to fail, and
then pick up the pieces--because they saw the value of what we were doing--
for a bargain.
180 Founders at Work
Livingston: Can you describe the investors' initial responses? Did they say,
"What the heck is this?"
Perlman: The biggest issue they had was the concern that people did not want
to interact with their TV. I mean, we showed working prototypes, but that
wasn't enough. By then we had a browser working that we had written from
scratch. In less than a year, we had a browser working. To give an example,
when Microsoft did Internet Explorer, they started with Mosaic. We couldn't fit
Mosaic into our system. We only had 2 MB of RAM, and we had a 112 MHz
MIPS CPU, and we had 2 MB of ROM and 1 MB of flash memory.
None of these existing browsers could fit into a memory footprint so small.
So we had to go write the thing up from scratch. Of course, we had to deal with
the reality that a TV screen was very narrow. We had a different user interface
for the remote control. We had a custom chip, and we had a programmable gate
array doing the video. We were doing the image processing that I mentioned to
eliminate the interlace flicker and to sharpen the image. We were building the
whole network side, which was all the servers and the network that would handle
and proxy the information. For example, if a large JPEG came in that we
knew that a TV could not display, we would resize it in servers and send that
down to the box to make a faster experience.
Then we had to go set up a whole dial-up network. We had to make relationships
with dial-up providers all around the country, so that they would automatically
find a local phone number to dial. So there was huge range of things
we were doing to make this thing work. I don't know, but, if I were advising a
VC, I'd see a bunch of guys with all these pieces of the puzzle and they were
executing and working with so little capital--I'd say, "Wow. I don't care what
they're going to do; something's going to come out of it." But that's not the way
that most investors look at it.
Now, looking back, I think some of the investors saw us as potentially carrion
ready for the taking, if we ran out of money. None of them said that, and at
the time I wasn't thinking that, but now I've seen it happen. I think other
investors were just nervous. Because all the other Internet plays are happening--this
was 1996, and huge deals are being done with the Internet. But, they
are all purely web-based: software running on servers somewhere. There were
no actual capital costs. We were talking about building a box that was going to
be deployed to people's homes. It has to be manufactured; there are inventory
risks, all those kinds of things. It was just not something they were used to
doing.
But the biggest thing people would say is they didn't think people would
want to interact with their TV. They could imagine them changing channels
with their remote control or playing a video game, but, as far as doing something
more advanced with the TV--like surfing the Web or doing email, or the
future things we were doing, where you had video content on the TV along with
the program guide (believe it or not, back then there weren't program guides
on TV) or having video eventually recorded on a disk with pause/rewind--
people thought that was crazy. I know it sounds so obvious now, but back then
they thought it was crazy.
Steve Perlman 181
Then we found one venture capital firm, Brentwood Venture Capital. Jeff
Brody, a VC there, saw it and he thought it was great. He said, "We want to
invest." And they were prepared to put in $4.5 million.
We were just about to sign all the paperwork. It was great, since we were
plumb out of money. I would have lost everything: my house; I would have
been deep in debt; the company would have folded; it would have been a bad
scene. Then we get a certified letter from Sony, and it said, "After due consideration,
we've decided not to proceed with you in deploying this product."
Remember, they had told us they had to have a 1-year exclusive. So we weren't
very far along with anybody else. We'd begun discussions with Philips, and we
told them it was a year out.
You have to disclose this to an investor, so we went and told Jeff Brody. It
was a real seminal moment for the company. He could have said, "OK, then,
I'm not going to invest if you don't have anyone to deploy your product." But he
said, "I believe in you guys, and I think this is going to make it. We'll still go forward
on the same terms." As soon as he moved forward, Paul Allen wanted to
get in. So he put in the other $4.5 million, and we ended up raising $9 million
that round.
After that, everything began to change. First of all, Philips came back, and
they immediately said, "We want to do a deal with you." Because they had been
sitting on the sidelines. We said, "We think we might be able to do a deal sooner
than 1 year." They said "Great."
Meanwhile, we had hired a consultant, Spencer Tall of Asia Pacific Ventures,
who had done a lot of deals with Japanese companies. He spoke Japanese
fluently and, in particular, he had a personal relationship with Idei-san, who was
the CEO of Sony at the time. We told him, "Look, we got this letter from Sony.
They said they're not going to do the deal with us." He says, "Well, let me find
out why this thing got bottlenecked, how it actually got shut down."
He went and called Idei-san while he was in the United States--this must
be April of 1996, maybe May. It turns out that he was at a business meeting in
New York and he had his chief technology officer with him. We're busily working,
and, when you're building stuff, you're always doing different builds. They
always have bugs in them. None of them were really working, because we were
in the development stage. And I got a call from Spencer, who said, "I just got off
the phone with Idei-san. He said that his guys didn't think the thing really
worked and they're skeptical this would ever be successful as a product. I told
him he really should reconsider. So he dispatched his chief technology officer
on a private jet to your offices to get a demo. This is your big chance to show it
to them."
I said, "Great. When's he sending him?" He said, "No. He dispatched him.
He's in the air now. He's going to be there in 2 and 1/2 hours."
I said, "Spencer, we're in the middle of development here! We need more
warning than 2 and 1/2 hours!" We had been adding a bunch of code, so it was
really crashing all the time at that stage. So I went back and talked to Bruce and
Phil and said, "Look, we have one last chance with Sony. Their CTO is coming
here in 2 and 1/2 hours."
182 Founders at Work
Phil said, "Well, we do have a build that's compiling now"--it took a long
time to compile all the source and then we had to release and test it--"and it's
going to be done in about 2 and 1/2 hours." I said, "How do we know it's going
to work?" He said, "Well, it probably won't." So I said, "What do you mean?"
He said, "All the recent builds we've done had major bugs and serious crashes.
We did do a lot of fixes here, though." And so I'm thinking, "Holy cow. This is
our big chance, and we're in a really bad stage of development." But we had no
choice, so I said, "OK, let's roll the new build out when the CTO comes and see
what happens."
The guy arrived about 15 minutes before the compile was done, and so we
kind of wined and dined him. We brought in a vegetable tray and had some
drinks there and were talking with him and tried to be polite. He said, "I really
don't have a lot of time. I need to see your WebTV prototype now." So then I
look over to the prototype area, and Phil had just walked in with a WebTV prototype,
"Here it is. The new build is loaded into this box."
So for better or worse, it was ready to go, and we sit Sony's CTO down on
the couch. I remember saying to Phil and Bruce, "What happened when you
tested it?" And they said, "What do you mean? This is the test." So I thought,
"Great. We're doomed."
We turned the thing on, and I don't know how, but it was perfect. It ran perfectly.
It just happened to be a good build. It was pure chance, but it went
through all the paces. We could go to websites and we typed in URLs and went
to all the different things, and there it was: WebTV did what it was supposed to
do. You could see the Web on TV.
We talked about the image processing and flicker elimination and showed
him the hardware and everything, and he looked very impressed. In fact,
shortly thereafter, we got a call to come to Tokyo to present to Idei-san himself
and his staff. In the end, he brought in engineering teams from all over the
company simply to see the image processing we were doing to make such a
sharp image on a TV, because they had never seen that before at Sony, even that
one element of technology.
The one website that the CTO went to that didn't work when he was in Palo
Alto was a Japanese website, because we didn't support the Japanese characters.
We had one engineer, Mark Krueger, who we had worked with at Apple,
working from Japan. He married a Japanese woman, so he lived there, and he
had an ISDN line to a house in the middle of a rice paddy, literally. He was
picking up a little bit of Japanese. When we went to Japan, we got to the Tokyo
Hyatt, and I remember Bruce had a development system there--we had
hauled these big computers with us. Mark had a development system at his
place in the rice paddy. And the night before our demo with Sony, they went
and did another build. They didn't tell me about this, but Bruce stayed up all
night working with Mark, and he integrated Japanese language support into the
code. So, literally, we arrived in Japan with an English-only browser, but by the
next morning we had it running English and Japanese.
Livingston: You didn't know they did this?
Steve Perlman 183
Perlman: I didn't know. Bruce told me on the cab ride over. I said, "What about
the stability? Bruce, Japanese is nice but..." And he said, "Don't worry. It
won't crash. It will be fine, it will be fine."
We gave a great demo in Palo Alto, and now we're going to give this demo
to the president of Sony Corporation, and we're going to fall flat on our face!
Well, it didn't crash. It worked beautifully. The CTO was there, and we said,
"By the way, there was one web page that you went to in Palo Alto that didn't
work. Well, it does work now." We typed it in, and, sure enough, it showed
beautiful Kanji, and we won him over.
Then they said, "We want to go back to the original contract we negotiated
with a 1-year exclusive." And we said, "We would love to do that, but now we
have a deal with Philips, and so we can no longer offer you an exclusive." They
were very unhappy about that, but, in the end, they felt it was worth doing. So
there it was. We had a deal with both Sony and Philips--at the time, the two
powerhouses in consumer electronics.
Now that we had these deals in place, we raised Series C. I think we raised
about $35 million.
Livingston: Did you get funding from the same people?
Perlman: Well, Brentwood re-upped, and I think Vulcan re-upped, and the
Davises did what they did at Catapult--they flipped. They sold their shares to
other investors. They're not technology people, so they saw it purely as an
investment. And they were happy as clams. I think they got seven times their
money in less than a year.
And then Microsoft came in, interestingly enough, and Citicorp and
St. Paul Venture Capital. Some individual investors came in, and also Seagate, I
think, put some money in. And Washington Post Group. A lot of people were
interested in the subject area. We expanded the board then, so the board was
now the three cofounders; we had Randy Komisar as an outside director, and
we had representatives from Brentwood, Vulcan, and I think that was it. Maybe
we had one other guy.
Then we cranked. We introduced the product in July of 1996--one year
almost to the day after I got that first check from Marvin Davis to fund the company.
It had custom hardware, a browser from the ground up, proxy servers,
and so forth. The whole network was supported, and I was true to my word
when I called the guy at Pacific Bell and told him that we were going to be running
a nationwide online service.
Livingston: How did the idea for WebTV evolve? Was it to make the Web available
to people who might not have computers?
Perlman: Yes. I should go back even further. My mission, even before then, was
to connect average people together doing non-engineering things, the things
that interest them--to foster better communication, sharing of ideas, and for
pure entertainment. I love storytelling; my favorite college class remains "The
Novel."
I wanted to figure out how to do communication. I wanted to figure out
consumer electronics. I wanted to figure out ease of use, you know, interfaces.
184 Founders at Work
I wanted to work with televisions, audio systems. That's what I've been interested
in, and it has driven all the things I've done.
When I joined Apple and interviewed with them, they weren't even interested
in doing color, and we brought them over to doing color. We created the
whole color model as well as the rest of multimedia for the Mac--music and
sound and everything. We made the Macintosh from a little black and white
computer into a multimedia powerhouse. And it was driven really by what my
ultimate desire was: as a delivery vehicle for multimedia and a means by which
you can interact. Video games are one kind of interaction. That's great, but
there is more than just that.
I think that, in the end, if you have enough people communicating with one
another, it's going to be really hard to go and blow each other up. They may
send nasty messages on blogs, and they may argue and maybe somebody will
write something unpleasant in Wikipedia about you, but that's a lot better than
blowing someone else up.
Livingston: Was it hard designing something for non-technical users?
Perlman: It's extremely hard, because you have to design for someone who's
not you. After a while, as you develop interfaces and have experience with
them, you begin to think with the intuition of a person who does not understand
the inner workings of the system. And you also have to do a lot of testing.
You have to be good at testing. You have to know what questions to ask people
and what problems to present to them.
The following is not something from my personal experience--it's a story
told to me by the Mac team--but they said that, when they first did the dialog
boxes for the Lisa, instead of saying "OK," it said, "Do It." They found that
people were reluctant to click on that, and they couldn't figure out why. Then,
once they had a test subject there who just wouldn't click on it, they said, "Why
didn't you click on that little button there?" He said, "I'm not a dolt. Why would
I click on that?" People were reading it as "dolt," not "do it," because it was an
unusual combination of words. So they changed dialog boxes to say "OK." That
little change greased the skids for people to click on dialog boxes.
It's very small stuff like that, very often--that somebody sees something and
has the wrong impression. The only way to learn that is by doing a lot of testing.
In fact, that's one of the reasons why the iPod was such a phenomenal success
where the MP3 players before were not. The iPod had the design sensibility of
an average person just trying to listen to music, whereas the previous MP3 players
were kind of technical exercises in understanding how music files are
stored, and perhaps required very delicate balancing of your fingers to hit the
buttons the right way, and so on.
Livingston: Were you inspired by the Apple II's use of TV as a monitor?
Perlman: Well, Apple IIs did work on TV screens, and I was inspired by the
fact that it was a friendly-looking computer and that it had color. But it was not
an easy-to-use computer. That's one of the reasons I didn't join Apple earlier. I
didn't see where that was going to go.
Steve Perlman 185
I was very impressed with the engineering. When I looked at the floppy disk
drive for the Apple II and I saw that it was a Shugart SA-400, but it was missing
most of the chips that every other computer had in it, and realized that Woz had
hacked the thing and was doing a lot of it with a combination of software and
hardware, I was deeply impressed with the engineering. But it was not something
that I could see an average person using. I could see, probably, more
likely someone using that than a CP/M machine. Remember, this is before even
the IBM PC.
But I was ready to leave the world of computers. I was working at Coleco in
1983. At the beginning of 1984, I was calling up Lucasfilm and other people in
the film industry because I thought, "Well, the IBM PC"--which was introduced
in 1982--"is taking over the world. Its graphics display is so poorly architected.
It doesn't even have square pixels. It has a palette of just eight primary
colors--actually two grays and six colors. Clearly, it's being driven by business
applications and so on. My dream of where computers would go turns out to be
the wrong one. It's not going more toward the average person; it's going more
toward businesses. And that's OK." So I figured, "If I can't do it in people's
homes, at least I want to be involved in creating exciting experiences on the
silver screen and on television."
Then I saw an ad in 1984 for a Macintosh, and it changed everything. I saw
that it had all this cool graphics capability. They clearly were interested in the
user experience: it was designed for an average person in simplicity; it was very
graphically oriented, albeit in black and white. I decided that there was hope.
I kept calling Apple again and again, trying to find somebody who would
talk to me, to get an interview there. Finally, Alan Kay, who I had worked under
at Atari and now was running a research group at Apple, came to visit
Connecticut to give a talk. I told him what I thought about the Macintosh and
said, "I want to make a color Mac and make it low cost." He said, "OK, OK. I'll
see if I can talk to somebody and get you an interview with the Mac team." That
led to three interviews, and I ended up not working on the Mac team, but for
the team secretly making the color Macintosh.
Livingston: Then you moved out to California?
Perlman: Yes. Actually, it was the second time. I moved out here before to
work for Atari. Then that went bust, so I came back to Connecticut and worked
for Coleco.
Livingston: Do you think there are major differences for a new startup in
Silicon Valley versus the East Coast?
Perlman: Oh yeah, phenomenal differences. I can't speak for every kind of
startup, but for something involving technology--and even a lot of things
involving content--it's just so much easier to do it here. You have resources
here and people who understand technology. There's a high concentration of
talent that you can draw on. You don't have to relocate people to get them
there.
186 Founders at Work
Then there's Sand Hill Road with all the VCs and other potential investors,
who are all clustered together. You literally might do two or three presentations
to different VCs all in the cluster of buildings on Sand Hill Road.
The other thing is that there's kind of an attitude here that people should try
things, and, if they fail, if they understand why they failed, they may actually be
a better investment in the next round than somebody who quickly succeeded
just by sheer luck.
Livingston: Were there any powerful interests who did not like what you were
doing and tried to stop WebTV? Like maybe Microsoft saw it as a threat to
Windows?
Perlman: With WebTV... Microsoft, I found out over time that they probably
. . . I mean, they're a very cautious company, and they proactively worry about
any potential threats. I don't know if they saw WebTV immediately as a threat,
but they saw it as a potential threat.
We didn't sell that many units the first Christmas. We were too high-priced.
We were $329 when it came out, and the lesson learned there is that you don't
charge both a high price and a subscription fee. Just one or the other, right?
When we repriced WebTV at $99, then we sold a lot of them.
So when Microsoft came to acquire us, we only had 56,000 subscribers,
which was a fairly modest number. But they still were very interested in us and
that convinced me--perhaps wrongly, but nonetheless convinced me--that
their real objective was to capitalize on this market, to grow from what we were
doing. Also their desire to create the campus here in Silicon Valley was the
other thing. So I kind of thought--and maybe even this was their objective at
the time--that they really were going to develop this area of advanced television
systems. But, as time went on, it became apparent that they simply wanted
to make sure that nobody else successfully deployed a product in this area.
I think they saw WebTV as the only viable player out there.
Who knows? Maybe it was a compound decision for them. Maybe they
thought, "Well, maybe there's a market here, and maybe we can protect our
flank to make sure nobody else does it." I don't know.
But there were some things that I was not allowed to do, which made it
impossible for me to stay. They reneged on their commitment to support
RealNetworks and Java, and I didn't know how we were going to build a good
web-surfing experience without Real and Java compatibility. Then, as we went
through the budgeting process and everything else there, and I began to get to
know the other top executives at Microsoft, they were talking about negotiating
this and that funding, and cutting back products to the point where they no
longer made sense. I said, "Look, can't we all agree on what is the right objective
for the whole company and then fund that? I don't care if it's in your group
or my group or whatever, but we should do the right thing." It didn't work that
way and, of course, any big company is like this. People have certain things they
control. That's why there's politics in large companies.
I can't operate in that environment. I'm just far too focused on the end
result. And so for all those reasons--the fact that they were very resistant to
Steve Perlman 187
adopt other technologies that they felt might be competitive, and the fact it was
just a large organization, no worse than any other large organization, I'm sure,
but, nonetheless, it was a large organization with a lot of politics--it became
unbearable.
But I think in the end they recognized that WebTV was a profitable thing
for them, because they ended up investing in it more, and now it's become
MSNTV. And it actually is a significant profit sector. WebTV was marginally
profitable in its 18th month of operation and has been profitable every month
since then, to this day. In 2005, 10 years after founding, WebTV (now called
MSNTV) grossed about $150 million for Microsoft with 65 percent gross margins.
Over its 8 years in the market, WebTV has grossed over $1.3 billion. We
never expected people would still be using a dial-up connection and browsing
on their TV in 2005, but there's still a significant market there for a device like
WebTV, primarily for older people who want to be connected to the Web and
email, but just will never buy a computer.
The other things we did with it that I was very excited about--that I was
hoping to really capitalize on, which was moving to satellite with DVR and
adding more interactivity and eventually moving to broadband cable--they
have not been successful at pursuing. I think with the satellite stuff they actually
introduced a couple of reasonably good products, but they got tripped up in
the business negotiations with that, and the cable operators are very resistant to
working with Microsoft.
Now their new thrust, which I think there's some significant opportunity
with, is with IPTV. And that sort of is what the legacy of WebTV moved into.
Peter Barrett is heading up that effort. He was the person who created the first
browser on WebTV. And here he is still at Microsoft--over 10 years later, if you
can believe it--and he's now trying to get television to work through the
Internet, while what we started doing was trying to get the Internet to work
through a television.
Livingston: What has the potential to go most wrong in the first year when a
startup is such a fragile organization?
Perlman: The worst thing that can happen to a startup is if the founding
team--or the people who are leading the thing--do not get along. And it's
deadly when they don't get along in front of the troops. I've come to realize
over the years that companies are just the people that make them up. We like to
think of them as business enterprises and having this value and that value. Well,
if you are going to distill it down to just patents and you are just going to go after
people for infringing your patents, I agree. Those companies are simply made
up of intellectual property assets. But any organization that actually has a product
they are trying to ship and/or service they are trying to provide, it mainly
comes down to the people. And the attitude of the company distills from
the top.
In an organization that is very large and has existing businesses that have
been running for a long time, you could have some not-so-great things happen
in the top of the company, and it doesn't have as much impact. Maybe the
188 Founders at Work
employees aren't so happy, maybe they don't think so highly about their jobs,
maybe they don't work as efficiently, but the company can keep plowing along.
But, in the early days, you've got nothing. All you've got are problems--
problems that need to be solved, obstacles that need to be overcome. You need
to have an incredibly strong bond and an incredibly synchronized view of the
world amongst the key players if you are going to succeed.
A synchronized view of the world doesn't mean you don't argue about
things, that you don't have disagreements. You must agree on the philosophy,
though, and on the vision. There are many ways to get there, but if you can't
agree on the vision, then obviously you're never going to agree on how to execute.
And you've got to respect each other. You've got to have cordial relationships.
You've got to be decent people. I can't tell you how many times I've seen
companies where people are fighting or arguing or nasty things are happening--fists
punched through walls.
There's one example where a pair of pliers was tossed by the CEO at the
controller. They sailed over her shoulder and lodged into the wall behind her.
She dropped her papers and said, "I quit," and stormed out the door. I've seen
this stuff happen at different places, and these companies--no surprise--have
not prospered.
But, then again, I've seen companies that really have a lot of execution
problems. General Magic is a good example of that. They really did not execute
well in what they were doing. The product was very expensive; it was over
$1,000. It was heavy. The battery didn't last long. The screen was not very
bright. And it was loaded with all sorts of features that were not really needed
by a mobile professional, which is what a PDA--this was 1990--was targeted
for. Palm, on the other hand, made something small, light, the battery lasted a
long time. It was inexpensive and focused on things like a calendar and address
book. I saw General Magic working on getting bunny rabbit animations working
to make it cute. It had infrared output, and they were making it so it could
tune channels on a TV. They were doing things that were just taking sideways
turns from the core product, which were interesting things to work on in a playground
kind of environment as engineers, but were not focused on the product
execution.
But the thing about it is, the three founders were very, very closely bound
together. They worked together well, for better or for worse. They projected a
common vision. They exuded stability, which made everyone else feel stable in
the company. And it made the company strong. They were able to survive. The
company ended up lasting over a decade. It did an IPO, though it finally fizzled
out. And the product was never successful in the market. It only sold a few
thousand units. But it shows an example of where you have so many things
working against you--the product was not one that was marketable, and you
are facing all these problems--but because they had a very strong core, they
were able to survive as an organization. To me, that was the most important
thing.
The key thing about Phil and Bruce is that they had hearts of gold. They
were nice people. They were not in it to get rich. I mean, money certainly is
Steve Perlman 189
freedom. But they both had a vision of creating something great that people
would love. That attitude from the three of us permeated the rest of the organization.
And the organization functioned well. On top of it, we had a strong
business model and good execution on the technology. We were actually profitable
18 months after we launched.
We could have had all the technical talent and engineering know-how and
business knowledge, but, if we were acting like Chinese fighting fish in a tank
together, the whole company would have failed. And through those really difficult
things that I told you about, where we were running out of money, when
we got the certified letter of rejection from Sony--all those different things I
think would have easily knocked over a weaker triad. We persevered through all
of it, and we never stopped. None of us ever doubted that we were going to succeed.
And none of us ever stopped to question whether or not we trusted the
other. We never had to look at our back. And that is what allowed WebTV to
persevere.