Gates gambles on Longhorn

Even after all his triumphs, Bill Gates is still a gambling man.

In some of the most difficult economic periods in high technology's history, the chairman and chief software architect of Microsoft appears undeterred in doing what he has done best: selling Windows.

This week, Microsoft launched a developer version of Longhorn, its next operating system. Gates says Longhorn, which is expected to be released in 2006, will spawn an array of new applications not possible with previous generations of Windows while making entire infrastructures work better.

However, Longhorn also offers a radically new programming and storage model along with the promise of extensive training for developers and managers. Will that message appeal to big companies that are struggling with security, tight budgets and minimization of risk? And will its benefits outweigh the cost to changes in systems at a time when companies remain leery about spending more money on technology?

To address those questions and other difficult issues, CNET News.com's Mike Ricciuti spoke to Gates on the first day of his company's software developers conference.

Q: Several years ago, many analysts believed that we had seen the last large monolithic operating system releases from Microsoft. Given Longhorn, did they just get it wrong, or has there been a change in thinking within Microsoft?
A: Well, this isn't monolithic. I mean, believe me, there is an architecture here that is extremely componentized and broken down into pieces.

But Longhorn is still a single, large operating system release, as were Windows 2000 and Windows XP?
That's right. It's integrated innovation. There's this constant trend that, as you are seeing things that many applications are needing to do and getting standards for how those things are done--those things move into the platform.

Over the last 10 years, there have been a lot of middle-layer-type things--queuing software, transaction software, etc.--that was fairly expensive, had a different model and different way to debug.

Now that we are moving to this Web services world--a loosely coupled, message-based breakthrough that computer scientists have dreamed of for decades--all of the things that let that be possible need to be in the $50 operating system. And so here we have Indigo, which will be in Windows and let you do transactions and queuing.

The person getting the benefit of those won't know that that is going on. But you also get to use the great Avalon graphics and the ability to navigate information with WinFS. Application developers don't have to duplicate those things, and yet there's no cost to having those things be in the platform and one way of doing debugging and performance.

That's the miracle of software, in terms of how we can get better and better things. So here we have something that was done through middleware coming into the system. We've seen that with media playback capabilities, with the browser, and that will continue...There is sort of that crowd that thinks that when the valuations broke that somehow technology advances wouldn't come. There's a general attitude now to not see that we will be delivering more software advances and productivity in this decade than we did in the last.

Given the state of IT spending and the current economic climate, some analysts say that Longhorn is a bold bet for Microsoft. Do you agree? What will drive businesses to adopt it?
Understand that for Microsoft it is not that IT spending will have to go up to take advantage of the new innovation. What Microsoft can do, because of our high-volume, low-price model, is put innovations in the software that not only simplify software development but reduce management costs, security add-on costs, directory management costs that an IT department has.

So our paradigm is very much that through software innovation, we simplify things that they are doing today; we surprise them by being able to simplify things, and that is where you get the opportunity for them to move forward with new hardware and new applications, using Web services, wireless, Tablet, things of that nature.

Longhorn is not built with the expectation of IT budgets going back where they were in the 1990s. The price of Longhorn is the same as the price of Windows today. The $6.8 billion in R&D (research and development) we spend every year--we provide that, and you just get it. That's what we have always done with Windows. So we are able to bring IT advances by having people save, and they will do a new generation of applications.

Some chief information officers I have spoken to in recent weeks say they want simpler desktop systems--not more complex systems. How do you approach those customers when discussing Longhorn? What's in it for them?

Longhorn, in terms of unifying the way you do things, is about letting people get at more of the power and having more simplicity.

Longhorn, in terms of unifying the way you do things, is about letting people get at more of the power and having more simplicity. If you are referring to network computers, those were tried and were the worst of all worlds, because they had a browser in them, and the browser didn't have enough memory, and it would get out of date, and you have to put in these ROMs and things.

Those things were thrown out because they weren't responsive or rich, nor did they stay up-to-date. If you want a thin client, we have that with Windows terminal server, and Longhorn will have terminal server capabilities so that you can project out the Longhorn richness onto thin clients.

In the same vein, plenty of CIOs continue to invest in Linux, particularly on the server. Will Longhorn change their thinking? Is that one area where Microsoft will sell Longhorn's management capabilities?
Certainly, that is a huge area we need to advance, and we think all of the operating systems need to do a lot better to build in the management. We have made great progress with Windows. People will be pretty stunned by how far we have come with recent releases. In terms of Linux on the server, it is certainly taking share in the Unix space. There are only two operating systems growing on the server, Windows and Linux.

Departmental level Windows is the overwhelming thing that people are using, particularly now with Windows 2003 Server. At the central level, there has always been competition between Windows and Unix. As we have gotten our scale richer, we are gaining in those areas. Linux is the Unix to compete with now, so every new release of Windows has to be effective for the problems that people want to solve.

Security is still a major issue for your customers. There were reports last week that some companies were reluctant to enter long-term contracts with Microsoft due to security concerns. What do you tell them about Longhorn that could change their minds?
First of all, I think you are mischaracterizing the reports. We have been very focused on helping customers with their security things, and we weren't even out there really touting these enterprise agreements like we normally do, because we have been very focused on helping people with their firewalls and updating. The two big things that stop propagation are software updating and firewall protection. In terms of keeping the update very small and flowing automatically, there is a lot that we need to do.

On the firewall side, we have articulated exactly how we are going to make that easy to audit and do super well. Our challenge is that because of the density of our systems, we have to block propagation, and the two ways to do that are to make sure people are up-to-date or turn on firewalls. So that's where we have a clear path to go through those things with customers in the coming months, to make sure that they are not having these security problems.

Developers are getting, in some sense, their first look at Longhorn today. How long of a ramp-up process will be necessary before IT developers are comfortable with it and begin to integrate it into their daily work--and more importantly, plan on Longhorn as part of future projects?
By knowing what Longhorn will look like, it will influence quite a bit of what people are doing today. For instance, the trend toward XML (Extensible Markup Language) Web services, as people use Indigo, built into the platform.

		Get Up to Speed on... Web services Get the latest headlines and company-specific news in our expanded GUTS section.

That will reinforce the move toward Web services as people see all of the schematizing, the common information types for contacts, appointments, documents, annotation.

They will actually see that, OK, they can adopt those schema standards for the types they are using. I think it's extremely positive for XML Web services. It's the commercial ISVs (independent software vendors), I would expect, who will start to allocate resources to do unique Longhorn exploitive applications.

We're announcing here that our Office group will do a version of Office that requires Longhorn and that we are putting most of the Office resources into that effort. So that's a pretty direct thing by the most popular application that runs on Windows--the one that really shaped the way people thought about Windows user interfaces and Windows data exchange--that kind of thing.

What about legacy application support? Is there a straightforward path for moving older applications into the Longhorn era?
We demonstrated VisiCalc running on Longhorn today, to show that 20 years of compatibility is a serious thing--people who want to run Electric Pencil and dBase and early versions of 1-2-3 if they want. We have built a compatible operating system. That's one of the things that Microsoft does. We have put a lot of resources into it. Even if we are providing new capabilities, we run those existing applications and make sure there are benefits with those applications. The greatest benefit comes with a new application written for Longhorn.

But the way that we let you search for information, replicate information between devices--those things are benefits, some of which you get even of you use existing applications. So we are going to show for a little bit of effort what you can do in Longhorn, and then if you are willing to do a lot of effort, what the additional things are. The whole dialog is just kicking off here.

Will WinFX, Longhorn's new programming interface, look sufficiently familiar to experienced Windows programmers?
Well, anyone who has worked with managed code will see that WinFX is taking APIs (application programming interfaces) and doing them in a very consistent, rich way for managed access. Anyone who has seen the .Net framework will see this as an evolution of that. There has been a switch in the last four years toward using managed code. I'd say maybe half of the stuff that gets done now is in managed code. So that just pushes this a little bit further.

On my last slide (at Monday's keynote) I had five action items: fundamentals, which are all of the things around security and updating and managed code; use managed code; use Web services; take advantage of the rich client; and, finally, get involved in the Windows communities that will really shape what we end up doing in Longhorn.

Can you talk a little bit about how some other Microsoft systems and initiatives, such as Xbox, embedded systems, Tablet PC, etc., fit into the Longhorn scenario?
Tablet came out a year ago. We have over 400,000 of those machines out now. A new update of the Tablet Windows software will ship sometime next year. Then, there will be a major update of Tablet software with Longhorn. And so we are really thinking about how ink and voice fit into the shell, and even in the demos, we showed how to do little ink annotations on the files--like a sticky note. The whole recognition and infrastructure for dealing with ink is dramatically advanced in Longhorn.

We get some of that in this intermediate release, but the really big release is Longhorn.

So every group at Microsoft is talking about the Longhorn wave.

So every group at Microsoft is talking about the Longhorn wave. We may have a few groups have releases between now and Longhorn, like Tablet and Media Center. But most have done their pre-Longhorn release and are now focused on the Longhorn wave.

What about the Longhorn server? Can developers begin to build systems without the server version of Longhorn?
The server and the client have not been on the same schedule; as we get further into Longhorn, we will further clarify some of the schedule things. Today was not about schedule. We talked about having the beta (of the Longhorn client) next year, and we gave people the developer preview bits. But we are very clear that this is a technology-driven release that will take quite some time to get in full shape.

All of those schedule things will become clearer--the notion of what intermediate releases there are for the server. And there are a few things we need to improve between now and then. We have talked about one, which is a security-focused release of the server, and we have even talked about some of the capabilities. We will get all of those intermediate milestones, then explain how the Longhorn server is the culmination of all of those things.

And developers can begin to build programs without the server version of Longhorn?
That's right. There are benefits when you get WinFS on the server or Indigo on the server. But we're full speed ahead on developing Longhorn apps, and some of the people at this conference will begin developing apps right away.

We've been hearing about the WinFS unified storage idea since back in the Cairo days. Why now for WinFS? What has changed?
I'm the biggest believer in this. It took a while for the database technologies to mature to be able to deal with heterogeneous data. That's really the XML revolution into the database. There is no coincidence that, as we are developing this Yukon version of SQL Server, we thought, "Wow, we can take some of that and have it be in the file system." It's probably due to the power of the machines, the amount of information people have. The world of your address book is different from the world of your songs, which is different from the world of your files, which is different from the world of your e-mail.

The number of commands that you are learning--the number of interfaces--is very large. People notice that if we put features into one, like the ability to be notified of something new, then they want that capability. They want to know about new music or new family photos or new address changes. You end up, for every new capability, building in the separate silos. People want notification, and they want control, and they want security and replication and rich search. They want a lot in these individual silos.

Now we're saying, "OK, you can get it in one place." Partly, it's the hardware evolution; partly, it's the database evolution; and it's partly that the benefits to users to pulling these things together is critical, because we've really got as many concepts as people can learn, yet they are not really adept at each of those silos because there isn't the sharing that WinFS will bring.

It sounds as though Longhorn helps to revive some of the earlier .Net My Services ideas about personal information management and seamless computing. Is that true?
From the idea that we are schematizing personal information, yes. But this is under the control of the user, stored on your client. That's the big difference. With HailStorm (the code name for the scuttled .Net My Services project), we didn't really have the rich query capability to navigate these things. We didn't have content indexing and all of that.

We had the idea of replication, but because we didn't have a (concurrent) OS change, it was very server-based replication. Of all the initiatives we have taken in the last decade, I would say that HailStorm was the one that hit a dead end in terms of how people responded to the information being on the server and the fact that we didn't have rich search capabilities.

Some of the concepts about how people manage their information are revived, correct?
Sure. I think the idea that your address book should be usable by other applications, that your calendar should be accessible by other applications--that is a big part of Longhorn, because we move those rich user schemas down into the platform so that applications can all get at presence and phone numbers and annotate the address book, instead of things like each application having its own address book.

Some reporters and analysts have pitted Microsoft against IBM over the future of Internet computing. The two companies have cooperated on Web services and other areas. Are your visions that dissimilar?
There is enough in common that I could talk for a long time. And then, depending on who at Microsoft or IBM you talk to, they might choose to emphasize the parts that are different. We're a software company, and we believe that the magic of software leads to reduced complexity and new development platforms and helping security issues. They do software, of course. Most large accounts have .Net and WebSphere in them.

I think that IBM has been our primary competitor in key areas for a long, long time. What's important for the industry is that we have a common vision around Web services. (IBM software executive) Steve Mills and I demonstrated that about a month ago in New York. That is the infrastructure that allows e-commerce to happen and allows software anywhere on the Internet to talk to software independent of computer language, operating system or hardware.

IBM and Microsoft put their best people on this, and it's a real contribution to the industry. It's something people dreamed about for decades. We have been very clear about the road map and getting those standards to workshops. So that is common work. IBM still has its profit stream from mainframes and large systems.

That's very different from us. We believe that software can reduce hardware costs and development costs. And because of that, we focus on what we can contribute to software. So you will see lots of things for which we will compete, and that is very healthy for customers. But we have this common vision of Web services behind it.