Taking the Next Step to Office Productivity

Computing Japan: Joining NeXT was a natural progression after Apple, right?

James Higa: Yes. By the time I left Apple, it was getting pretty big and bureaucratic. NeXT was more of a start up. A lot of innovation went down the road from Cupertino [Apple's HQ] to Redwood City [NeXT's HQ].

CJ: In Japan, when I say "I use a NeXT," people respond: "Oh, NeXT Haven't they gone bankrupt? " How has the change from being a maker of boxes to a seller of software affected NeXT Japan?

Higa: There's a hit of a time lag between the US and Japan. Even in the US, people are saying "Didn't you go broke'" The US is changing, though, with Sun's endorsement, and OpenStep, and so on. We'll probably see something similar here, and it's alr eady starting to happen.

It used to be really hard to get into companies before -- we'd make a lot of cold sales calls, and most people would say, "We haven't heard of NEXTSTEP. We don't understand what it is." It's the opposite now the calls are coming to us. People are sayin g to us, "Sun is going with this, and we're a Solaris house, so we're going to have to learn something about this. Could you come over?"

CJ: These are Japanese software houses, right?

Higa: Yes, right. Japanese software houses, Japanese customers. In the not-quite-a-year that we ve been a software-only house, we've picked up a lot of steam. I think a combination of a couple of things has helped us. One, object-oriented technology is coming into its own, and two, the whole trend towards the break from mainframe to client-server. Clearly, people see it as a three-horse race: Cairo, Taligent, and NEXTSTEP. Objects are the trend of the future.

CJ: How have Canozz Hanbni, the distributors of NEXTSTEP in Japan, taken the more to a software company

Higa: Their strategy, with the success of the Macintosh, has always been a solutions-oriented approach: "we've got Macs, we've got IBMs, we've got Canon brand -- whatever you need, we'll provide the solution." It's probably more reflective of customer attitudes in Japan than of Canon themselves. "

CJ: Do you think that the move to total solutions in Japan, rather than shrink-wrap user-installed solutions, is customer-driven?

Higa: I think the customers and the market in Japan are where the US was about four or five years ago. I know everyone says this, but in this case think it's true. Seven or eight years ago, when I was encouraging software shops to go mail-order, I was told, "lt'll never happen in Japan. Japanese customers need to touch it, see it, pick it up, go to the store." In reality, things are clearly going that way.

A few years ago, was saying that a lot of the computer shops were doomed. Manufacturers were starting to go direct, mail-order was picking up steam; in the US, the Computerlands and RusinessLands just don't exist today, because the customers have more knowledge.

We're starting to see the beginnings of the same movements in Japan, as Dell's recent success shows; Japan is following the US's lead. The trends toward client-server computing, the trends toward object-oriented programming, the trend toward more custo m applications, the trend toward enterprise-wide software architectures -- all these are on the horizon.

CJ: NEXTSTEP is pretty low profile; you don't see it in every Akihabara store. If someone reads this interview and says, "Hey! That sounds like what my company should be looking at, " what should they do?

Higa: They should contact us or Canon Software. We're out doing demos every single day. For these past two or three years, we've stuck to one marketing strategy, having tried a number previously.

We're focused on custom applications; we're not a PC for the masses, or a Macintosh. The technologies we're using will filter down and get more widespread, but at the moment we're focusing on users who need to write distributed, enterprise-wide applica tions. We can't compete in the low-end game. There are almost no margins there any more; it's brutal, both in hardware and software at that PC level.

One of my motivations for coming to NeXT was that I'd always wanted to promote computing. Companies have put PCs on desks as personal productivity tools, given everyone PageMaker and wordprocessors and spreadsheets. Maybe that's done some good for each individual, but it's had a negligible effect on company-wide productivity and efficiency.

CJ: There aren't a lot of NEXTSTEP commercial applications. What is NeXT doing for independent software houses?

Higa: Well, it's obviously not our main focus. Ironically, there are more Japanese applications shipping for NEXTSTEP than there are in the US. There are actually four Japanese word-processors for example. There's a lot of consolidation going on in app lications -- most companies run on a handful of productivity applications -- and there's a lot going on in emulation. A lot of people are going for multiple personalities: launch OpenStep and go into Solaris Windows, Or Microsoft Windows, or Mac. Legacy a pplications should run on any platform in the future.

CJ: How do you integrate these with your custom applications?

Higa: I think we're going to see more and more network-aware applications -- a new class of application. Some of the more innovative applications, such as MindShare, Notebook, PencilMeIn, Diagram!, and Concurrence, that we've seen already on NEXTSTEP a re showing the way. They're a new breed in their messaging capabilities.

Actuallp, what constitutes an application under NEXTSTEP is a difficult question to answer, because of the Services function and the interaction between applications and so on. Applications in the PC world have got to learn something about this. All ap plications are going to need APIs in the future. We are going to see a change in our perceptions of an application -- it may be core code with an API that can be subclassed or whatever. The totality of the thing will appear very different to the user than the conventional application does now.

CJ: Who tuill he providing this kind of thing?

Higa: Probably the big players, because they're the Only ones with enough momentum and market share to impose these APls. We'll probably see the emergence of a new class of players, halfway hetween shrinkwrap providers and integrators. I think we're in for a change in the nature of applications. It's a very tough time for developers the world over -- apart from Microsoft.

CJ: So, where does NEXTSTEP give the compelling advantage?

Higa: We're embarking on the hcfiinnings of a more networked society. I hate using "information highway" it's too hyped a metaphor -- but it's clear that information is crucial to ompanies who need a competitive edge. In many ways, information is bus iness these days. Competitive advantage lies now in how efficiently and fast you can organize and dissemnate the information for your business

Most competitive advantages we seee now have a database behind them. MCI's "Friends and Family" campaign is a good example. It's a 100% information-driven customer service. It's an intelligent billing system, really. What's behind it was a database and the custom applications to make it happen; it took AT&T forever to catch up while they ceded market share.

CJ: So that's the kind of thing: maintaining a large database, getting the infermation? on the right desks at the right time?

Higa: Yes. Packaging infermation in new ways, being able to analyze information -- that's your competitive advantage, that's your productivity. Federal Express couldn't have happened fifty years ago, because the informatjon technology to track all that wasn't around.

If you're really worried about increasing your company's productivity, you need to increase your enterprise-wide productivity, not individual productivity. Inventory management is crucial, and we're entering the period where, if you can't ship in 24 ho urs, you're not really competitive.

In order to do this kind of thing... well, you could write 1-2-3 macros forever, but you're never going to get the right inventory management system for your business. You can't go to a computer shop and buy the "MyCompany Inventc.,ry System" off the s helf. Most companies are running this kind of thing on very old COBOL programs, which are not flexible. If you have a new sales campaign or you put a new product out, you have to wait till you can churn out the mainframe code. So the sales force comes to you and says, "We want to do a sales campaign on those customers who've bought version 2.0 in the past three months for a special price on the West Coast." When you go to the MIS department, they tell you, "Right. We'll have the code written in eight mont hs."

CJ: To change the subiect a little, NEXTSTEP has got its reputation in two or three sectors: one is the findncial services sectors, banks and so on, and another is US government dgencies. Have they given any ideas to the NeXT development teams for f uture productivity?

Higa: Well, it's fair to say that government services are the most intensive users of custom applications. They're very good developers, and they do very large enterprise-wide applications. We've got a lot of feedback from them, which has made its way into products. The same with all of our customers, actually. We're now starting to see new markets developing: telecommunications, for example.

Telecommunications is an information-driven industry, and becoming more and more so. Take the complexity of mobile phone rates as an example. Custom applications can help the sales force better explain all these options to customers. McCaw Cellular is one of our biggest customers. It used to take them two or three weeks to hook up a new user to the service. It now takes about 5 minutes, and the person in the store can do it.

CJ: So two or three weeks pre NEXTSTEP, five minutes postNEXTSTEP?

Higa: Right. Try doing that with 1-2-3 macros!

CJ: I've spoken with various people in the financial services market who have commented that their older black NeXTstations have been retired because they were too slow. Do you Know of any cases where people have said "NEXTSTEP is a great system, bu t it just can't cut it on the available hardware"?

Higa: When you get hold of something as powerful as NEXTSTEP, as those people did who bought the first NeXT cubes and NeXT stations, all of a sudden your expectations balloon. Especially in the financial market, there's been a dire need for adequate so lutions in a very computer-intensive environment. I don't think anyone could have gone in there at that time and satisfied all the needs of that community.

Obviously, the NeXTstations made a dent, but for a lot of the applications that these people are writing, even today's machines aren't going to be fast enough. It's got a lot better, with the Pentiums, and PA-RISC, SPARC, and DEC have just signed on, s o there's a lot of hardware out there for people to start running their applications.

CJ: Just how many processors that will run NEXTSTEP are there at the moment?

Higa: Intel, PA-RISC, Alpha, SPARC -- and we still support the Motorola.

CJ: And PowerPC?

Higa: When we see enough of them ()Llt there, we'll port. What it comes down to is: as soon as we see a market out there, we'll get it out. Right now, it's just too early.

Coming back to your earlier point: customer comments and feedback led us on the path to Distributed Obiects, Portable Distributed Obiects, Enterprise Obiects -- because making the client faster is n(,t necessarily the best way to increase performance. A lot of the applications used by the financial community need the speed on the servers, which is where the original idea for PDOs started. If you have really computerintensive procedures, 80% of the hardware out there isn't going to cut it, but it would be great if you could do it on a really fast server.

CJ: So you put the Cray running PDOs in the basement, and run NEXTSTEP on your desk?

Higa: Sure. That is a deeper user need than was immediately apparent. What appeared on the surface was, "we need faster clients." When we started diving into it, we found that people wanted to be able to run applications on hardware that was suited to them. It could be a server in London, and an even faster server in Tokyo or wherever. The whole notion of obiect repositories, and portable distributed obiects, is something that we learned as we started working with our financial customers. It's hard to make the mindshift from "client and desktop." Hardware vendors are still pursuing the hardware race on the desktop, but when you start talking about enterprise-wide systems, the problem's a little bit bigger than that.

CJ: So people who are constantly lacking up the clock speed and so on are digging themselves into a ditch?

Higa: No, not at all. We love fast clients, we want faster clients, and we hope the race keeps going forever. But for many customers, the equation is more complicated than that. They need the freedom to deploy applications, or parts of applications, an ywhere that they want, wherever it best suits them. It's been difficult up to now, because there haven't been tools to let you do that. Hopefully, now we're giving people the freedom and the tools to implement these huge applications the way they want to.

CJ: Could you explain about NeXT's new Enterprise Objects Framework? Higa: There was a very deep problem in that more and more custom applications were database-driven. The big problem with trying to write these applications was that the data and th e user interface and the logic -- the business logic, for example, how you set your credit limit -- are all hardwired into this one little application typically made with a 4GL or SVL. So your credit rating calculation, which is something you use all over your enterprise, is replicated in about 20 different locations throughout your organization. What happens when your credit rating procedure changes? You've got to take 20 applications apart, fix it -- all with different tools for developing different lay ers. You never know whether you got it all or not.

That was the big problem. In hindsight it seems obvious, but what struck US was, "Gee. That sounds a lot like an object. Why don't we lust take the layers apart and have an object that codifies some of your business logic, and have it independent of th e data structure?" So for the credit rating, you can use that object multiple times, enterprise-wide.

CJ: It's a distributable object?

Higa: Yes, it's a distributable obiect. Basically, we've separated the display, methods, and data into different layers. Your credit approval object may need to access different databases, for instance. Previously, you'd have had to recode everything. Now, it doesn't matter; it's transparent to the application developer. Also, if you wanted to view the data in many types of window, this necessitated everything being hardwired, so all the logic is stuck round all the different windows. 1 Which makes no sense; it's a maintenance nightmare. What we've done is to separate out the logic. It also lets you compose teams that take better advantage of people's skills. The great thing for the database people is that nothing changes. They're still in charge of Oracle or Sybase, just as they always have been. They're still the experts, and can still develop these databases. I There's a new class of people who can concentrate on writing these business objects. And then there's a third class of developers who write applications using previous objects. CJ: So for a multinational with a Japanese office, you could have the database developed and located in New York, manipulated by Enterprise Obiects developed in London, which are then shipped to Tokyo where they're used in locally-developed applications Higa: Yes. And you don't have to rewrite the business logic dozens of times every time it changes because it comes from one source. Just plug in the new object, used by many different applications. Enterprise Objects brings the obiect-oriented paradigm to the relational database, and combines the two. It's a unique product. CJ: What is NeXT's product line-up Higa: NEXTSTEP for Intel and Motorola now, HP RISC from later this year. SPARC native NEXTSTEP from some time later than that. That's all the announced future native NEXTSTEPs; I can't talk about others. PDOs on HP and SPARC and DEC. That's 68% of the mar ket right there, and there will probably be more in the future. OpenStep has been announced for Solaris about the same time as the SPARC release. Of course, there's always the DEC port of OpenStep for Alphas -- DEC is working very hard on this. OpenStep is being based on NEXTSTEP 4.0 APIs and will be the first OpenSte p-compliant release of NEXTSTEP, incorporating Mecca technology. CJ: What does Mecca include Higa: Unicode, which is great news for the Japanese market. New memory management frameworks. Enterprise Objects and foundation classes are also part of Mecca technology, and some of this will make its first appearance in 3.3. We can expect to see 4.0 sometime next summer, and 4.0 will contain some, but by no means all, of Mecca. Any applications written under 3.2 will, of course, run with absolutely no changes under 4.0; that's one of the things we think we do pretty well. You know about the name "Mecca"? CJ: No Higa: It came from Steve Jobs' line at the Expo, when he said, "Why stop at Cairo when you can go to Mecca?" all of this CJ: What about Japanese versions of Higa: We reckon about three to six months after the release of the English versions. We're pretty good about this sort of thing. It's actually more of a documentation issue than anything else. Version 4.0 will actually be one universal release, thanks to Unicode. We think it's going to go pretty fast. Unicode came out of a bunch of us at many different companies working on multilingual programming: Xerox, Apple, etc. We all got together and asked ourselves, "If we had to do it again from a clean slate, how would we go about it, throwing our company a llegiances aside?" For a lot of manufacturers, Unicode is the logical choice to use internally. It's the simplest and cleanest solution for internal code. We will probably see it become a standard implementation. For a few years or so, we'll probably see a lot of filters and automatic conversions to keep compatibility with existing data. But eventually, I think we'll start seeing native Unicode. There really is a need for this. Do you realize that twothirds of the world's population can't sit down at a keyboard and type t heir own name into a computer? The whole thing comes down to how a culturally-dispersed enterprise can develop worldwide applications. Unicode is part of the answer; so are PDOs. and so are Enterprise Objects. What is all this going to do to file structures and data structures ten ye ars from now? We're not going to sit still and wait for Taligent or Cairo to get going -- if they ever do. We'll keep moving forward. Actually, I'd like to have the Taligent QA manager's job -- that's security for life. [laughs] To be fair, we've been developing an object-based operating system for about eight or nine years now. We know it's not easy. CJ: That's an interesting point. How has NeXT heen able to produce a robust object operating system, which is now in 3.2, with relatively few resources, while a giant like Microsoft and a consortium backed by some of the biggest names in the industry are still at the vaporware stage.

Higa: More doesn't necessarily equal better. In fact, we started off with two advantages: we started off with a small team of very good people, and we had the advantage of being able to start from scratch. We had no baggage.

CJ: Where do you see NeXT a year from now?

Higa: Our goal is to become the leader in providing object technology and to be a legitimate alternative to Microsoft. I think it is clear we are already the technology leader. We're building up momentum; we are so much further along than we were eight een months ago. Object technology, client-server, and mission-critical application technology are all becoming important, and we're picking up a lot more partners.

I think by this time next year, we will be a legitimate alternative, with the largest market share of the emerging technology. Business in the 90s and the next century will dictate the direction. It's not a lone dream everyone else is trying to get the re. But we're on the leading edge, and we intend to stay there with the help of our partners.