Ballmer in charge

Last time we talked was two years ago, at the beginning of the .Net era. So where are we now in the .Net evolution?

Let me start by talking about where I think we are in the XML revolution first, and then [I will] talk about where we are with .Net. I believe that the IT industry collectively decided over the last three or four years that the world was going to be replatformed with XML as the key foundation. It’s not like anyone ever got together to decide this, but it’s like magnetic north, where it’s been clearly indicated that everyone is rallying around the concept. But I think the degree to which [XML] is pervasive might still be underestimated. A lot of partners, even at this conference, see it as b-to-b for big companies. But it is really about the way software in general is designed. It is all about the way security should work. Management — everything is changed.

So to answer your direct question, two years ago I might have said the jury is still out about whether [XML] was going to happen. We were questioning if it was going to be a legitimate competitor to Java. But now the world has decided to embrace [XML] as the basis for the next generation of technology. And there are a lot of benefits that come with it. Could it have been something other than XML? Sure it could have. It wasn’t the specific technical implementation; it is merely the fact that it comes out of Internet standards. It has deeper semantic abilities. It is all about interoperability and connection.

So in that context, where are we with .Net? I’ll say a couple things. No. 1, I would say we have some momentum in the market. We have customers using the .Net product line, like Merrill [Lynch], Credit Suisse, Marks & Spencer, CitiBank, and a variety of others. People who are doing some things for real. I feel like the degree we are out front is in some ways smaller and in some ways larger.

How so?

Smaller in the sense that everyone is talking the XML talk, even Sun; bigger in the execution sense: We’ve delivered. We delivered the first batch of products that really were built XML to the core — Visual Studio .Net and BizTalk Server — and we are getting traction. So I feel good about where our execution is relative to everyone else. That is not to say that other people aren’t doing some reasonable bolt-on XML support for the stuff they have.

You have some heavy lifting to do to convince your customers and partners to follow along. What do you have to do to set the standard for adoption and upgrading to this next generation?

We have to present the picture that we are behaving consistently within the framework of focusing on customer issues and customer satisfaction and trustworthy computing, which we have articulated. Now there are some things I would have done differently, in 20-20 hindsight. I would have increased the focus on not just security but deployment, because a lot of the security problems are actually deployment problems. We have the fix, but sometimes we can’t get the tools in their hands to deploy those fixes fast enough. Licensing — I know I would have given more time, and there are better ways to plan around the kinds of changes we made. We are still going to have to get smart about the full ramifications.

Do you expect any changes to be made between now and July 31 to Licensing 6.0? There is a lot of push-back on this.

No, there is nothing that will happen between now and July 31. And we will continue to learn over time what we need to do to earn our customers’ trust and earn their business. I think many of the people who are pushing back really don’t even know the new terms.

Isn’t that your fault for not educating them?

Yes, it is our fault. But you meet with plenty of customers whose companies already have licenses, and they are saying, ‘I think there’s a problem.’ I don’t deny there are some problems. I am sure there are some issues we will have to work on over time.

After July 31 we are going to see what the real issues are. I was talking to a customer at the show who was saying he thought there would be a problem, and I told him that we just completed a deal with his company and that he is now universally licensed and that total costs to his company are now lower than they were before. And he said, ‘Oh really?’

So we’ll go through July 31. We have a lot of customers who have signed licenses, and I am sure we are going to find some issues. My bigger concerns today are probably not in the larger accounts, but there may be some issues we need to address among the smaller-size companies, and we will.

How does the .Net initiative change the way in-house client and server developers work together?

What we are trying to do for the health of .Net and the health of the company and the sanity of the people who work there and the quality of the products we produce is to have a more orchestrated, ¸ber-road map of where we are going.

Now there are two things a road map can be. One is just a piece of paper that tells what everyone is doing and records a bunch of bottom-up plans. Or it can be something where we impose more broadly our vision from the top. We are not going to try to get everything shared. … That’s a path for getting everything stopped up. But at the same time we are trying to take the top four or five things we are trying to do in the Longhorn wave, Yukon wave, or the .Net wave and really have enough time to have some cross-cutting discussions about how we make that happen.

Bill [Gates] is dedicated full-time to being chief software architect. He also happens to be chairman of the company, but he spends most of his time being chief software architect. We have a body now that meets every month — the SLTT [Senior Leadership Technical Team], basically the top people involved in the product-creating strategy — and we present and discuss the issues. Bill is driving a process for each of these waves. For instance, [he will ask] what are the cross-cutting scenarios that I want, and then [he will] get all the right product groups together and bring these things in front of the SLTT.

People talk about what it takes to be a good manager, and people talk about being a good coach and getting people to like you. In a large organization those things are important, but at a certain level, … the most important thing you can do for your people to make them better is to give them a framework where they can work in harmony with the other people in the company. And from Bill’s perspective, [this framework] is the technical road map and the SLTT process; it is the scenarios. From my perspective, it is the P&L [profit and loss] structure and the multiple businesses because, given how we work, giving people a framework that gets them to make the whole bigger than the sum of the parts will mean we have better offerings.

It seems the scenarios construct is really working for you. Is this an evolution of the solutions approach?

I would say it is a pretty direct extrapolation from the solutions stuff, only I think we are just getting a little more experience. I think we use the word scenarios more often and solutions less, because solution sounds more like a fixed thing. At the end of the day, most of the things we do are customizable and programmable, and so people make them into the solution they want. But they are still scenarios. What will software deployment look like? Well, that is a scenario because it involves Windows, Office, and other applications.

We are working hard to make sure the whole is bigger than the sum of the parts. That’s not to say people are always comfortable with this. I can’t even say we will do it perfectly. All we have to do is do it a lot better than we had been doing it. I think that would be a big step forward for the customers.

Will this approach cut down on some of the technology civil wars Microsoft has had in the past involving who is going to gain control of a project? This seems to be a very democratic approach to development.

I would not use the word democracy to describe it. People have to come together and ultimately make some decisions. This is not a voting process here, the agenda of what we think is important. We get a lot of input. [But] once it is decided, people don’t get to vote with their feet.

We try to make a quantum leap here in terms of our ability to improve pulling together all the pieces. Our customers want us to do that. Simplifying concepts by unifying them. The customers don’t say they want us to do that, but when we do do that, they go, ‘Ahhh, I get it. That makes sense.’

I don’t want to go to six more meetings where people say, ‘Is your workflow strategy BizTalk?’ … Or ‘What’s your Office workflow strategy?’ Or ‘Why is synchronization in some ways better for e-mail and Outlook than it is for file folders and Web pages?’

To get that sort of synchronization sometimes — to get that high-level road map to work — you have to build something we don’t have today. So we have been talking about unifying storage. And so the best way to get all this stuff to work is to get them all to work just one way. Then you just keep tuning one set of parameters instead of having a thousand flowers blooming.

In your Fusion keynote today, you talked about who your competitors are, and it was the usual suspects. But one you didn’t mention is yourself. You are competing against the last generation of your technology.

Always. Given that our stuff doesn’t wear out or break down — you can question whether it breaks or not [Ballmer grins] — but it doesn’t break down like machinery does. To get anyone to do something new, we have to have something interesting and innovative.

People think, OK, should we just do more upgrades or try to sell more new product? We decided that users like us to do a level of systems integration so they don’t have to get involved in that. So take Sharepoint Team Services, which we put in Office. We could have sold Team Services as a separate product, and we could have introduced a new SKU [stock keeping unit]. We could have decided if it should work down level with the old Office or just the new version of Office, and we could do that with real-time communications. But we tend to like to put things together, integrate them, and then release batches?1202?– which are then ‘upgrades’ — as a form of introducing our new innovation. But no matter how you do it, because our stuff does not wear out, we have to convince people that some idea we have is better or more powerful than the older idea we had.

And there are two aspects of doing that. One aspect is, on a given product or upgrade release, does that make sense? Or do we convince you that over an nyear period of time we will have enough innovation that you will want to be licensed for that innovation. And this is part of this whole licensing discussion we are having.

We moved to the new licensing for two reasons, I would say. No. 1, we wanted to simplify our licensing. When we look back a year from now, I guarantee you that where we are now is simpler than where we were. Nothing is simpler during a transition because they are learning the new and are more familiar with the old, but I think people will eventually realize this is simpler. Our licensing got pretty complicated over the years, and some people came to not understand it all.

The other issue is [the timing of delivering] our innovation. I don’t want to have a lot of rough questions [from customers]. So when we do have new ideas, it’s like, OK, should we put it in this upgrade or that one or just have a stream of stuff that only our best customers have access to under the terms of their contract? I think in the long run we have a better chance of satisfying them with the latter than with hitting them with six different upgrades. I am a fan of [letting] the customer decide, OK, I want that piece or this piece. But they are licensed for all of it. So those are the two things we sought to do.

That gets back to the trust issue, where if you front-load it, that’s the big, lumpy upgrade; and if you back-load it with the vision, the .Net 2003 servers, you have a long gap in the middle where you have to sell into an untrusting audience.

The truth is, we need to be better at both. There are some things you can only do big and lumpy, like a new file system. It’s not like some sort of small user feature; if you change the file system, then the storage, the backup, probaby those things are going to change. Probably the shell ought to change.The applications … a lot of things are going to change. Can you call that lumpy? Or you can put out four new features in PowerPoint, sure — that would not look lumpy to users.

I think we have to build trust in the model, and we are going to have to deliver more over time. It will not come overnight. People will have to see that we do a few things, … that not all of our innovation is lumpy, that more of it comes evenly and steadily. People develop a confidence and faith; they like both aspects of the model. Over time, our support needs to become more intimately involved in that story.

We know that, but we don’t know exactly what that means yet. And this whole notion of software as a service — this is not a licensing or business discussion. It’s how does software get rearchitected in the world of the Internet? I mean, how many software products are fundamentally different because of the Internet? Not many. Probably the anti-virus guys keep up the best. They are always feeding new virus support in. I asked the audience [at Fusion] how many use Windows Update. Now that’s a small piece of software, but it is big. That’s the consumer market; we’ve got to get that to really purr in the corporate market.

The lack of broadband build-out and adoption seems to be holding up the ability for users to take advantage of some new services you’re talking about. When Bill [Gates] invested in Comcast cable, it triggered the Baby Bells to get on board with DSL. What are you doing now to deal with that issue?

The No. 1 thing is, we have to make software that is so compelling that people will start saying, ‘Well, of course I am going to get broadband.’

And 802.11 is one of those compelling features?

Yes, it is. I was in a hotel in Sun Valley, [Utah], this week that was not wired. So I turned on my PC, and XP tells me there is a wireless network available. So I connect to something called Mountaineer. Well, I don’t know what that is. But I have a VPN into Microsoft; it worked. I don’t know whose broadband I used. I didn’t see it in Bill’s room. I called him up and said, ‘Hey, come over to my room.’ So soon everyone is there and connecting to the Internet through my room.

We have to support all the modalities, whether it is GPRS [General Packet Radio Service] or 802.11. We have to make it easy to provision from a software perspective for things that go on with DSL [and] cable modems. … And the software has to want a broadband link. There are not many things you do where you say I really care about the performance. If I am downloading PowerPoint files, then I want broadband. So when you can do things with speed that you otherwise can’t do without speed, that is going to put a lot of pressure on the broadband community.

That’s the weird answer. The more normal answer is, yeah, isn’t it awful: There is not enough broadband, and the prices they charge … . I agree with all that. If you look at what happened in Japan recently, there was a big drop — it is now $25 a month for DSL. Vroom. They have gone from almost no DSL connections a year ago to 1.5 [million] or 1.8 million [subscribers].

So how do we make that happen in this country?

I think that is beyond us. If there is enough interest where the consumers are looking intensely for broadband, whether it is at today’s prices or tomorrow’s prices, that is going to shake out the marketplace. And I think the marketplace will do a good job. We just have to make it something obvious that consumers want. Now I happen to live in a neigborhood where there is no broadband network. No DSL or cable modems. Well, I have a neighbor, [Teledesic chairman and co-CEO] Craig McCaw, who has a wireless network that I think I can piggyback off of [Ballmer laughs]. So, yes, broadband is a bottleneck — but I also think we have to look at oursleves as a software industry and say, ‘Aren’t we, too, some of that bottleneck?’ We have not built software that makes broadband a must-have.