Speaking With Apple’s Sandy Benett

As I write this, Apple Computer is at the center of a hurricane of speculation. With MacWorld in San Francisco having just ended, the company is expected to soon announce the extent of its restructuring which is likely to result in significant layoffs, product strategy changes, and possibly even a management shakeup. Caught in the middle of this controversy is the Personal Interactive Electronics (PIE) division, home to the most famous PDA—the Newton.

Late in December, and only slightly before the current storm, I had a chance to speak with Sandy Benett, acting vice president of the PIE division at Apple’s Cupertino, California headquarters. With the successful launch of the Newton 2.0 operating system behind him, Benett was eager to share his guarded optimism of the world’s best-selling pen-based personal digital assistant.

A Bit of Background

Benett received his undergraduate degree in Mathematics and his graduate degree in Computer Science, both at the University of Maryland. After teaching computer science for six years at the college level, he set off to do some consulting in artificial intelligence, a specialty from his graduate school days. Reflecting on this, Benett noted: “I have a background in teaching which helps me with some of the management challenges. However, I didn’t teach kindergarten so I didn’t get enough of the experience that I could occasionally use.”

“I have a background in teaching which helps me with some of the management challenges. However, I didn’t teach kindergarten so I didn’t get enough of the experience that I could occasionally use.”

Within the field of artificial intelligence, Benett worked mostly with reason-based and picture processing systems. He recalled: “Actually, I did some work on aerial photograph identification. I also did some natural language processing work for the National Library of Medicine at the National Institute of Health (NIH). Then I decided I wanted to come to California since I’d been coming out here every summer when I taught.”

“In 1982, I joined a start-up company to do optical character recognition. Though I didn’t have a character recognition background, but a pattern recognition background, they recruited me to work on that. Having been in Silicon Valley for a year and a half and learning all about start-ups, especially the good stories and none of the bad, I went off to work at the start-up and did that for about 5 years.”

Benett described a hand-held optical character recognition appliance on which they worked. It could scan text and recognize it using a PC as if you had typed it. Benett remembered: “The company stayed around for about 5 years, living from hand to mouth, and we actually got the product out, but it was too expensive. We were bought by Everex because they wanted to take the technology and marry it with their scanner product. But Everex was not a very interesting company to work for and, within 3 months, we were all gone from Everex.”

Xerox a Go-Go

Leveraging his interest in character recognition, Benett’s next step was with Xerox Imaging Systems. However this was short lived as a new and exciting opportunity presented itself—another young and hungry start-up: GO Corporation. Joining GO with an employee number in the low-forties. Benett found himself again working on the problem of pattern recognition, only this time it was handwriting.

Recounting his early days at one of the most high-profile mobile computing companies, Benett described: “I was hired to come in and manage the team, and the guy who was already doing this was supposed to move on to another job. But he didn’t go for about six months, so I was actually working on the handwriting recognition system, which was fine. It was an interesting problem and, after the six months, took over managing the team. I subsequently wound up running the software division. There were two parts to it: the operating system (OS) part, and what we called the substrate. I wound up running the substrate part, you know, the layer above the low level OS stuff.”

The PenPoint Dream

The early pen computing days certainly were exciting. Many felt that they were at the vanguard of technology and indeed they were. Benett concurred: “It was really cool and it was interesting because when I got first recruited, they wouldn’t even tell me what they were doing. It was very, very secretive in those days and, in fact, for the first two years I was there, nobody knew what we were doing and the press was all over the map because we just didn’t talk about it very much.”

“When I got there they had the first prototype, but just got it running the day before. It was an 80286-based slate that had a little bit of handwriting recognition and a little bit of the OS. And it was really cool! I thought it could be revolutionary and it was interesting to participate in something revolutionary with a really good group of people.”

Benett stayed with GO Corp. until almost the end. However, with Apple expressing interest by late 1993, Benett was torn. “It was really a difficult decision for me whether to come here [Apple Computer] or try to stick out the year after investing five years into GO and really believing in the approach that we took. But the way that they put the two companies together [EO, Inc. and GO Corp.], I just didn’t feel that it would work. There was just too much of a culture clash.”

Reflecting on some of the reasons why GO ultimately failed, Benett noted: “It was a really good group of people, very smart, with a very good idea. But what went wrong was that they decided to get out of the hardware business completely. The issue was not so much that they split up with [what became] EO. Starting a separate hardware company was not that bad of an idea at the time. In fact, at the time, it looked like the right thing to do because it freed us from competing with our licensees, which on the surface looks like a really good idea.”

“The bad idea about it was that you lose all control of the operating system because you don’t have a hardware reference platform. And so we were being pulled in many different directions at the same time and never succeeded in any. Each of our hardware licensees—and we had some substantial companies as licensees—had a different vision of what the product should be, what the hardware incarnation of the software product should be.”

Pulled in All Directions

“For example, AT&T’s vision was a smart phone and they never were really a licensee—EO was their licensee. Toshiba’s vision was more of a slate-type device with a large screen: a portable desktop companion-type device. Fujitsu, on the other hand, had more of a vertical focus. IBM had their vertical focus but different than Fujitsu’s—IBM was looking at heavier and larger products, and more ruggedized. And NEC was very interested, but they of course wanted it to be compatible with their own 9800 technology rather than PC compatible technology.”

“So we [GO Corp.] were getting pulled in a lot of different directions and then, of course, AT&T came in and wanted us to port everything to the Hobbit [RISC processor] because they wanted to sell Hobbit chips.”

“So we were getting pulled in a lot of different directions and then, of course, AT&T came in and wanted us to port everything to the Hobbit [RISC processor] because they wanted to sell Hobbit chips. So here we were, well down the road on a 386 implementation and now, all of a sudden, we have to switch gears and try to do both because we had all these licensees who wanted to do 386 and AT&T who wanted to do Hobbit. Finally we realized that we couldn’t do that, and switched to a Hobbit-only system.”

Commenting on whether GO tackled too ambitious a problem, Benett noted: “In some ways, I think we were too unfocused. At the same time. we may have been too ambitious in trying to support all those licensees rather than trying to do one or two products and get them out. If we were doing our own hardware, we would have gotten the product out and then the other licensees would have seen what it was and could have reacted to it by maybe making their own designs a bit more realistic.”

Shifting to Newton from PenPoint

When Apple first approached Benett with an opportunity which they clearly felt was equally ground breaking, Benett showed little interest. He recalled: “Shane Robinson recruited me to come to Apple. He called me out of the blue one Sunday night at ten in the evening and I had breakfast with him a couple of days later. At the time, I really had no intention of leaving GO, but I talked to him anyway. I was very impressed with him and I talked to the team of people that I would be managing if I came here.”

“By that time, I believed that EO and GO were not going to be successful the way they were put together. For example, I thought they made the wrong choice of which management team to keep and which to let go. Keeping Alain Rossmann on the EO team and letting Bill Campbell in the GO team leave was a big mistake. I personally didn’t believe that Alain could be successful in the way that he envisioned the company going forward. And so, I came to Apple and thought that we could be more successful.”

Questioned a bit more about Rossmann’s vision, Benett explained: “He was enamored with [AT&T] Bell Labs and he was enamored with voice technology. Rossmann thought people would want their written email read to them as they drove their cars which may be possible, but I don’t think that should be the focus. He thought that PenPoint could become VoicePoint. Some of his ideas were interesting, but I knew we wouldn’t be funded long enough to make any of those happen.”

“Rossmann thought that PenPoint could become VoicePoint. Some of his ideas were interesting, but I knew we wouldn’t be funded long enough to make any of those happen.”

Everyone, including us at Pen-Based Computing, initially felt a sense of relief when AT&T became owner of both EO and GO Corp.’s technology, thinking that the telephone giant’s vast resources would be employed to sustain the growth pains in this nascent industry. Benett concurred: “We all thought initially that we now have deep pockets and we’ve got technology that we can draw from. For instance, when the Bell Lab guys came down, we talked about all the different technologies that were possible. But to take those technologies from the lab and place them into products was going to take quite a while, and it just didn’t seem that AT&T had the passion.”

According to Benett, this led to a deadly spiral from which the company simply couldn’t recover: “One of the problems we had was retaining people, because when you change direction and culture so radically, a lot of people decide that it’s not something that they want to continue to do. And while we were able to maintain the core, we just weren’t able to execute. We sort of got caught up in a limbo for a long time and that was really demoralizing.”

Moving to Apple represented something different for Benett. Not only in terms of technology, Benett was attracted to the organizational structure of the Newton group. He confided: “One of the appeals of the Newton was that it was a separate entity within Apple. Somewhat but not isolated, it was separate from the rest of the company by which we could concentrate on doing this business rather than doing the Apple business which is primarily Macintosh.”

But Benett entered at a rather strange time. Introduced at Boston MacWorld in August, 1993, the Newton was already receiving its share of negative press by early 1994. Benett recalled: “By the time I got here in 1994, the press was awful. But, you know, the morale here was OK. Some of the key [version] 1.0 people were already gone by the time I came to interview so I came with my eyes open. I knew that there were a lot of problems with Newton.” But Benett had spotted an opportunity. In his words: “I was an outsider from an Apple perspective.”

Identifying Newton’s Weaknesses

When asked about some of the criticism and the problems surrounding the Newton, Benett responded summarily: “What was wrong with Newton was that it didn’t work. The handwriting recognition did not meet user expectations and the problem was that all of the hype highlighted it. Though it was a really good technology and a good product concept, the parts of the product that were highlighted in the hype were the parts that didn’t work.”

“When we were at GO, we bought a Newton the first day that they were available and I sat and played with it for four hours. And my impression was, we’re in trouble—meaning the industry was in trouble, not that GO was in trouble.”

He continued: “When we were at GO, we bought a Newton the first day that they were available and I sat and played with it for four hours. And my impression was, we’re in trouble—meaning the industry was in trouble, not that GO was in trouble. I thought that we were in trouble because of the expectations that had been set for this machine—we were not going to meet them and we were going to have to have a re-education process with the customer. The customer expectations were going to be so high that the whole platform was going to be deemed a failure.”

“In other words, the uphill battle that we were already fighting just became that much steeper. And so, when I came to Apple, I knew we had a lot of work to do in order to be successful. But I believed in the people who were here and in the technologies that were being developed. And I think that we’ve done that with 2.0 [the Newton 2.0 operating system].”

Of course, handwriting recognition formed the focal point for both public and media criticism. Benett was aware of the issue at hand. Commenting, he noted: “I think ParaGraph [developer of the Newton’s cursive handwriting recognition system] did an excellent job. But I think that the platform put a lot of limitations on how good their technology could be. Given the limitations that they worked under, they did an excellent job. The problem was that the expectations were set so high, I mean, people equated it to writing on paper. It’s not, and it never will be.”

“As good as the handwriting [recognition] is today, it’s still is not like writing on paper, and I think we did ourselves a disservice by setting the expectations at a level at which it was almost impossible for anybody to succeed.”

Of course, the end result of all the hype was not entirely bad. In fact, it created a level of awareness for the product and its category that was almost mythical. Benett agreed observing that: “There was a lot of awareness. I mean, everyone knew what a Newton was. If you were holding anything larger than this and as long as this [gesturing with his hands], it was a Newton even though it could have been anything else. To this day, there’s still a little bit of that. There’s a saying that there’s no bad press or something like that. That awareness of the product category was very important.”

Managing Relationships

Putting his skills of managing a classroom to good use, Benett started in 1994 by managing the internal handwriting recognition group, along with the relationship with ParaGraph International. Benett noted: “My first goal was to work with ParaGraph and the internal ATG [Advanced Technology Group] to improve the handwriting recognition.” It was around this time that Benett discovered an interesting recognition technology brewing within Apple’s renowned research structure.

He recalled: “When I first came here, I heard that there was a project going on over in ATG that was being developed for the PenMac. Apparently there were three guys over there working on it and, having come from a handwriting recognition background, I was curious to see what they had. I had heard that it was print only, and that it wasn’t very good. So I went over to meet with Dick Lyon and when I saw what they had done, I was pretty impressed. But it was running on a Mac and it was huge. They didn’t really have a product to put it on.”

Ah, the PenMac. The PenMac was a much speculated potential product that never saw the light of day. Benett explained: “It was one of those projects that kind of got put on the shelf and never really died, but never really went anywhere. It might still be over there in someone’s road map for all I know. So I talked to Dick and said: ‘hey, there’s a product that this thing is well-suited for, and it’s called the Newton. If you guys could get it ported to Newton, I’ll help you champion it.’ And they did a great job.”

“They brought in some people and they got integrated with our handwriting recognition team here to learn the ins and outs of the Newton operating system. Then they got it down to a size that works and, in about a year and a half worth of work, took it from a prototype to a product. They’ve done a great job doing that.”

Opening the Handwriting API

Recognizing the importance of having a platform that supports a range of handwriting tools, Benett stressed the new and distinctly open nature of Apple’s handwriting technology. Benett quipped: “I don’t suffer from the non-invented-here syndrome. We’ve opened up all our handwriting APIs to any developer who wants to come in and write a handwriting engine for the Newton, or provide some other method of getting input into the system.

Of course, Apple hasn’t always been so far-sighted. Earlier versions of the Newton forced developers to use the keyboard API for handwritten input, preventing a developer from doing clever things with spatial and other kind of information that may make sense to a user. Benett agreed: “You’re right, the API wasn’t available. We were very close when I first got here and one of the things that Shane [Robison] started and I’ve continued is to open the system up in many ways. Not just the handwriting APIs, but lots of other ways because people are going to want to provide interesting technologies that further the platform.”

“Accurate handwriting recognition is the most important thing. I think [Palm Computing’s] Graffiti taught us a lot.”

Benett explained: “What happened with [version] 1.0 was that people wanted to write certain applications but they couldn’t because the APIs weren’t available. So we showed them back doors to get the applications out there. The problem with that is that as you evolve the technology, some of those back doors close. So you’re much better off opening up the APIs and making it an open system rather than a closed system.”

“And that was the major philosophical change for the Newton group. But that was my philosophy and Shane’s philosophy. We did a lot of that at GO Corp., where we opened up a lot of the APIs and we got a lot of developers involved. I thought that was successful for us so we implemented it here. And I think it’s going to be a real plus for us going forward.”

With Newton 2.0 sporting two distinct recognizers—one print and the other cursive—which is more important to the user? Benett was clear with his answer: “Accurate handwriting recognition is the most important thing. I think [Palm Computing’s] Graffiti taught us a lot. The Newton group did a bunch of market and customer surveys which said early on that people would not change their handwriting styles. And so you had to have what is called ‘natural handwriting.'”

Benett explained: “Nobody really writes pure cursive—a few people perhaps and they’re probably teachers. Everybody writes a mix of printing and cursive. And so the Newton group right away went looking for somebody who could do natural handwriting recognition which is a much more difficult problem than printed. And they found something that was state-of-the-art in the ParaGraph stuff, which they licensed.”

“But in the end, how important is natural handwriting recognition in the future? That’s going to be answered by the customers and we’re going to learn which is best since there are several different handwriting technologies available. But what is important to customers is accurate handwriting recognition. And what Graffiti taught us was that if you can be accurate, people will change their handwriting. I mean, the Graffiti alphabet is a departure for everybody. Nobody writes like that, but yet, they’re a very successful application on our platform and on other people’s platforms.”

Interestingly, Apple’s exclusive hold on ParaGraph’s innovative cursive technology expired in June 1995. When asked about the implications, Benett was direct: “It means that ParaGraph is still a valued partner and they are no longer bound by exclusivity. We plan to continue to work with ParaGraph, and to continue to incorporate their technology as long as it’s appropriate. But it also gives them the capability to market this technology to other manufacturers which makes their business more viable, and helps us in the long run.”

Does this mean that Apple may be interested in advancing their own cursive technology? Benett observed: “We’re always looking at moving technologies ahead, whether we develop it or other people develop it. We’ll always look at technologies to move this platform forward whether it’s handwriting recognition, speech recognition, or other ways of inputting information along with better connectivity to the desktop. All of those are very important to us but we don’t have to do them.”

Horizontal vs. Vertical Markets

Following the initial backlash against the Newton, many now feel more confident with Apple’s higher-profile embrace of vertical market applications and solutions. Benett, however, countered this perception: “We never really went at the horizontal [market]. Instead we realized that we believed our own hype a little too much that this was a consumer electronics product. It’s not a consumer electronics product in the true sense of a transistor radio, for example.”

“We’ve learned a lot by being in the marketplace for two years. We learned that what we did right was to develop an environment in which people could develop really interesting applications along with custom applications.”

Benett continued: “We still believe that there is a large horizontal market for the platform and the Apple-labeled devices. It’s just that it’s a different market than what we thought it was originally. We’ve learned a lot by being in the marketplace for two years. We learned that what we did right was to develop an environment in which people could develop really interesting applications along with custom applications. It’s easy to develop on the platform and that opens it up to a number of vertical applications. These people generally want a custom application, or two or three custom applications, to make their business easier for filling out forms, reading meters, whatever.”

“They need a small, light, portable device that will run some custom software. So we learned that people have found that they can do this and it’s easy—it takes three to six months to develop an application on the Newton. From beginning to end, this makes it a perfect device for verticals, so we should go after that market as well. That doesn’t mean that we get out of the horizontal market.”

Benett pointed out that they continue to actively work on making the development process even easier. He revealed: “We’ve opened up a lot more APIs in the process, and we’ve added a lot more facility to the operating system. And since it’s an object-oriented operating system, the application developers can inherit that behavior from the operating system. They don’t have to write all of that code. So a lot of the facility that’s in there, they just inherit, enabling them to write even more sophisticated applications with less effort.”

The Ups and the Downs

Following the fanfare of the MacWorld introduction, it seemed that the emotional and rational reaction to the original Newton had little choice but to recede. Apple regrouped, introducing a successor in April of the following year, showing both its commitment to the market and its understanding of customer feedback. Looking back, Benett sees this as a significant milestone. He commented: “I think things started turning around on the upswing with the introduction of the [MessagePad] 110 and then the 120.”

“We made (digital) ink a first class citizen allowing you to use ink anywhere. So even if you still have trouble with handwriting recognition, you can use ink in places that you couldn’t before. That was really important!”

“We came out with better hardware products and minor revisions to the operating system that fixed some of the more obvious problems. And we started to get a loyal following of people who began to see the utility of the product because it performed more reliably.” Benett noted that the new devices also enabled Apple to marry the technology a bit closer to a new range of wireless options, especially in Europe. He added: “When we introduced it in Europe and provided the capability to hook up GSM phones to start using the technology in a more professional way—enabling people to use it as part of their work environment—that helped us to start gaining some momentum.”

Benett was also quick to acknowledge the shortcomings of the earlier Newton, problems that he feels have been solved in Newton 2.0. He began: “There are a lot of things that didn’t work very well. Our mail system wasn’t very good; we didn’t do fax receive which customers thought was very important; we didn’t have an open architecture where you could add the transports and the different ways of getting information; and our connection strategy was lacking some things.”

“For example, we didn’t do synchronization—we just did export, making it hard to get information from your desktop to a Newton and keep things in sync. And we didn’t do anything with third party software as far as communicating between desktop and the Newton. Everything had to go through the [Newton] Connection Kit. We didn’t have a strategy to do that.”

According to Benett, Newton 2.0 fixes all these problems and more. Information input, for example, received a big boost in the new system. Benett explained: “We made [digital] ink a first class citizen allowing you to use ink anywhere. So even if you still have trouble with handwriting recognition, you can use ink in places that you couldn’t before. That was really important! We also have more communications solutions from third parties, so that these things [Newtons] really can become communication devices. There are cellular modems out there now, RF radio cards, paging cards, and a lot of things that we can now do with the new architecture.”

The Future of Newton

Though we spoke with Benett before the expected Apple announcement, the question of the future of Newton as a product, and as a project within Apple, came up in the conversation. Benett was firm: “I think the support is strong in Apple and I think it’s getting stronger with the introduction of 2.0. The fact that we were so successful at [Fall] COMDEX also helps us. There was a lot of visibility around Newton 2.0, and a lot of the Apple executives were there and saw the excitement.

“They also saw the commitment from the third parties and the other developers for the platform and that helped us as far as our relations within the company. Also, the fact that we are doing a better job of integrating with Macintosh and Windows machines makes us more viable to the rest of the company.”

Responding to the often-heard rumors directly, Benett stated: “Newton is not going to be cancelled. But Michael Spindler has made it clear that Apple has been looking for investment partners to share in the Newton success and to help fund us going forward. There are a lot of scenarios of how that could work. I don’t think anybody knows what that means right now, but it could mean that we have someone come in and invest in it.”

“It could also mean that they spin it out as a separate company with Apple and some other parties investing in it. The commitment to Newton is very high on Apple’s side. But Apple is going through some reductions in their margins and they’re looking for ways to continue to move the technology forward without it hitting the balance sheet as hard as it has in other times.”

Benett confidently summarized: “We’ll be successful in the next couple of years. There’s a lot of interest in making sure that Newton has the resources to be successful, and there’s a lot of investigation going on as to what that means.”

Gauging the Competition

Though the mainstream media was rarely able to focus past the Newton’s innovative and often fanciful handwriting recognition system, the company does indeed face rather stiff competition from an assortment of vendors and system developers. Benett is more than aware of this and noted: “We’re competing against a lot of people. Geoworks is our competitor. They’ve got an interesting system but they compete for mindshare more than anything else.”

Comparing the platforms, Benett offered this opinion: “We’re sort of a bit higher end than maybe Geoworks is. They’re looking at low cost devices, though somewhat less functional than ours. They’re doing a good job at getting more functionality into their device, but where they compete with us a lot is in mindshare”

Continuing, he noted: “They’ve come out with a $349 device with Hewlett-Packard and people right away say that it’s competitive with Newton. If you compare what the two machines do, they do some things better than we do, and we do some things better. But are we really in the same space?”

Asked about General Magic, Benett’s answer is similar: “Again, they’re in a slightly different space than we are. They’ve put all of their efforts into the communications end, especially with PersonaLink, the service from AT&T. They really have two different businesses. They’re an operating system and they’re also Telescript. I don’t know which one is going to be the majority business going forward. I think they’re trying to sort that out right now. We find Telescript interesting and we’ll keep our eyes on it—we have a license to it. But we think we still have the better development environment and better support from third party developers though they’re working hard to catch up in that area.”

“Everybody thinks that Microsoft is working on something, including us. Nobody knows that they really are, or what they’ll deliver. But it’s not going to have Windows on this type of device— we all know that.”

As for Microsoft, Benett appears as confused as everyone. He speculated: “Everybody thinks that Microsoft is working on something, including us. Nobody knows that they really are, or what they’ll deliver. But it’s not going to have Windows on this type of device— we all know that. I’m interested in what they’re going to do and we keep hearing this code name Pegasus. But we don’t know if it’s real.”

“I think it’s to their benefit to just keep the word around so that people wonder what they’re doing. But from my point of view, we just have to execute. If we come out with good products, get our licensee products out there, and develop solutions to problems, people are going to flock to the platform. And when Microsoft comes out with something, they’re going to have to catch us.”

“The biggest question that we all have is when is this going to capture the imagination? We’re not targeting the Everyman right now—we don’t think that this technology or market is ready for the Everyman. We’ll be perfectly happy capturing the imagination of the mobile business professional and selected vertical applications. That market is big enough to make us successful.”

Sandy Benett
Apple Computer
Acting Vice President
PIE Division
One Infinite Loop.
Cupertino, CA 95014
(408) 974-0656
(408) 974-8949(fax)
[email protected]