Agenda 2001 Album     |     home
Random Pictures   |   Sun's Talk   |   Ballmer's Talk
Ballmer's Talk
Steve worked hard on his talk. When I walked into the main room way before we started (in order to get an aisle seat near the front to take pictures) he was huddling with his PR company head, Pam Edstrom, and others. When he saw me, he got up to come over and say hello, and then excused himself to go back and work on his speech. It was nice of him to take the time to say hello, and at this conference he was a very accessible person to all: he hung around for hours talking to almost anybody who came over about all sorts of things. (All the while the PR people walked around doing their thing -- it reminded me of the "spin doctors" after the presidential debates.)

Here's my paraphrasing of his talk. It's based on my notes and a recording I made. Some of the words are his, some my impressions. One of Steve's assistants encouraged me to do a writeup so I assume I have their permission, and I felt, given their position in the industry and the stuff he covered (some in response to questions), it was worth doing for others to read. Other than what Steve said, Microsoft had no input into what you see here -- I didn't discuss it with any of them.

It started with 20 minutes of him speaking, followed by over 30 minutes of him answering questions from Bob Metcalfe and the audience. Steve speaks with lots of energy and enthusiasm. Hopefully the pictures will convey some of that.

The speech
The excitement they have about what will happen with the Internet is as great as anything they've been involved in since he joined the company 20 years ago. The key to taking people using connected devices to using them in new ways, he says, is software, be it software embedded in a router, applications, operating systems, platform software, web sites (which he sees as essentially software). The key skill set to turn on this new environment is software. That's where they are and are good at. They provide platforms and applications to let people do new and interesting things.

10 years from now, what will the Internet look like? How will it be different than today? New devices like phones and TV sets and more? He says "Right on!". They don't see PC devices going away. They see PC markets continuing to grow. But there will be new devices that people will use to access the Internet. But what will be different from now, most people say, is only that there will be new devices and more bandwidth. Broadband wireless, broadband to the home, broadband, broadband, broadband. Both of those things are true.

But from the perspective of someone who cares about software, what will be different about the software experience, how will the software environment of the Internet look different than it does today? What's wrong with the Internet as an "operating system". (Metcalfe has asked a question about what will be the dominant operating system in the future and the answer of the attendees was "the Internet".) Will it be HTML pages 5, 10 years from now? He thinks there will be changes.

So what will be different?

First, he focuses on "integration". How do you integrate information from numerous web sites and applications? How does the user do it? How does a program do it? How do they exchange data? What are the standards? Integration is very important and there is no model yet on how to do it. Standards are important to this, with XML, SOAP (which he says they are very active in), and other open protocols developed along with other companies like IBM.

Another area is "extensibility". Many web sites today do not have the extensibility and customizability that today's PC applications have. He says some people say that's good because extensibility brings viruses and confusion. But extensibility is another form of letting people build on the work of others. So how do web sites provide extensibility to other web sites? How do you let 3rd parties add functionality like he heard a particular company was doing? This isn't there yet in the "operating system" of the Internet.

"Ease of writing" applications. The tools for the development of web sites have not improved the extent you would expect given the advances in other areas.

The "UI" 10 years from now. Will it look the same as today? He feels we'll have speech and natural language built in. We need a user-understandable integration ability: What will be the equivalent to the clipboard in PC apps? Today there are two user interface paradigms: PC and browser. How do we bring those two together? How will we make interfaces scaleable from a handheld to a TV set on up to a full-screen device?

What new device types will the user interfaces have to support? He sees many people in the audience with laptops. He thinks in the future there will be tablet devices which are like the laptop but with a nicer user interface with handwriting for notetaking, conferencing, etc.

Then he talked about devices more. He see that there will be a wide range of new devices. He thinks those devices are going to be smart. While Microsoft has a "dialog" with their competitors about "thin" clients vs. "thick clients", he thinks the rest of the world only thinks about thick clients. The television industry talks about "advanced set top boxes". What's an advanced set top box? One with more memory and processing power than the old ones. In the wireless industry they talk about more intelligence in devices. The telecommunications industry talks about caching systems. There is all this talk about moving more processing power out to the edge. Well, what's the ultimate edge of the network? It is, of course, the device that you're using. The trend will continue to have devices with more intelligence built in to them. As we talk about peer-to-peer we talk about intelligent devices with only some taking advantage of things at the center.

He doesn't see how in 10 years we will distinguish between software applications and web sites. That is, all software will be expected to have the good attributes of a web site: self managing, self installing, takes care of itself, updates itself. And all web sites will be expected to have the best characteristics of today's PC software: rich user control, the best possible interaction that can be done on the device.

So as Microsoft looks out over the next 10 years, they love the new devices: wireless devices, TV devices, gaming devices, some dedicated devices. But, with the PC in an important role. They think bandwidth is important. But if we don't look at where the software platform is going, you can't predict the future.

What is the ".NET platform" that they recently announced? It's a set of work that's aimed at the opportunities he mentioned. New user interfaces, integration, etc. It's not a set of products, though it might be dribbled out as them since in this day and age he doesn't think they need another 6 year development cycle. It is a piece of middleware that is targeted at addressing these kinds of opportunities and to give software developers who want to build Internet-OS applications a platform that lets them take advantage of and address some of these new software opportunities that will exist in the Internet. It's middleware that runs on devices: Windows devices and non-Windows devices. It's software that runs in servers and it's software that needs to run out in the Internet cloud, because if you want to provide a very rich application integration model some things do need to be centralized, other things may be very decentralized and done out at the edge of the network.

Pieces will come to the market, all aimed at bringing about a change in the software environment that lets people target the next generation of the Internet operating system.

To have the impact they want to have, they will have to reach out to partners as much as they ever have. They need partners to take a bet on doing things differently. They'll need to prove to people there's something there before many will join them. They'll have to build some web sites and services themselves. But they need to reach out and bring in ISV / web site partners and he's been spending a lot of time trying to find people who want to bet on them for the early stage. Talks now are not just about API, but more of the discussions center around how do they go to market? They know they have to be marketing partners as well as technology partners. They know they need a good story. Lots of focus on outreach.

He closed by saying that Microsoft's agenda long term is enriching the software platform that we think of as the Internet in the ways he discussed.

Time for Questions and Answers:

Q: What bet are you asking people to make? A: There need to be ways to do these things. Nobody has the chutzpah in this day and age to say "let's go proprietary". Things will be based around XML. That's got to happen. There'll be other protocols on top of it. SOAP is an opportunity to standardize on how you deliver XML payloads on the Internet. But there is a whole infrastructure about how things can get done only in the marketplace -- it could be ours it could be somebody else's, so there are varying levels. The word "open" is a conundrum. They've said that open means everything's documented, people can do what they want to do. To do integration means working together on standards.

Q: What's wrong with Sun's and Oracle's views? A: In his view, neither understands what users do with computers. They have a sense of web site development, of operations teams, but anybody who denies that you want to have intelligent devices, it's naive. What is everybody talking about, the only thing the conference has its own panel on? An application that requires a rich client: Napster. [Metcalfe says: "Thin doesn't mean stupid. Thin means thin." Ballmer replied:] Thin means stupid. Thick means smart. The only thing you can say is how well managed is that device. The "sin" [his word] that we have today that they're cleaning up, is that smart devices sometimes were hard to manage and take care of, so they gave others the opportunity to say that "smart" meant "unmanageable". "Sounds like a lot programmers I know", he joked.

Q: What about Linux? What is it doing wrong? A: Hard to say "it", since it's a bit amorphous. They compete with Linux. What advantages does Microsoft have? It's very hard to do innovative work in the development in the development model of Linux, but it's easy to move quickly on some kinds of small, specific areas especially for your own use. He sees there being lots of different versions for each company providing its business model.

Q: With Microsoft's investment in Corel will there be a Microsoft Linux? A: No comments other than those they made when they did the investment.

Q: How do customers deal with bugs when they don't have access to source? A: Some customers have access to source and some don't. They try do only their own distribution, but customers can propose changes. Hundreds of customers have source and work with them. Linux has had impact on them, such as encouraging them to compete by making Windows embeddable, in non-PC devices and servers.

Q: When will they move .NET to Linux? A: They will if there is a way for them to make it a business. But they want .NET support in Linux, and they know Corel has expertise in that area so they left an option to work with them in doing that in their deal if the business makes sense, but the Open Source world makes it more challenging to find ways to make intellectual property pay the bills.

Q: [Person from RealNames commented how astounded they were that Microsoft partnered with them and gave them access to IE, so it shows that you can partner with Microsoft.] What about the DNS, isn't it broken, shouldn't Microsoft do something about it? A: It's just a little bit of a technology problem and a little bit of a standards problem, but it's mostly a business problem because who's going to invest in something with a non-obvious return.

Q: [Elliot Noss of Tucows:] What do you say to ISVs who do products, especially those that compete against yours, such as media players and browsers? A: Sometimes we'll support them even though they compete, but he doesn't think anything he'd say will ever give them much comfort if they compete with Microsoft's core platform, but people have been competing with them for years, such as utilities, but you have to accept the inevitability that functionality of these core products that they or any other company builds will continue to expand. So you have to decide if you want to live in this space and continue to compete and expand. He'd like to offer more comfort, but he can't, but he doesn't want to scare them off. They should contact Marshall Goldberg at Microsoft and his job is to deal with people who feel like they might compete with Microsoft so they can treat them in exactly the right way with appropriate respect for concerns of that kind.

Q: What are you doing to make the base operating system more reliable? Windows 2000 still freezes up for me, what to do? A: If Microsoft fails to continue to improve the core operability [his carefully chosen term] of the PC device, the PC device will not be very relevant. That's mostly their job to get done, though some is working with device and systems vendors. The greatest shame that it took them so long to finish Windows/Windows 2000 is that in some senses they missed the cycle of improving on the reliability issue. They are mainly concentrating on Windows 2000 for this. They find that the Windows 2000 support line calls are way down in comparison to their other OS's. Next year is the year they will bring the Windows 2000 codebase to the masses on consumer machines. He encouraged reliability feedback, even reminding people of his email address.

Q: How different will the usability experience be on Xbox (a Microsoft hardware device)? A: Xbox is a completely different product concept. It is a single function device. It is not infinitely or arbitrarily expandable. One of the differences between "appliances" and "PCs" is the level of extensibility that is designed in. They have very much restricted the extensibility of Xbox. It will be only extended in ways known to us and controlled by us. The good news is that it should be an environment that is much, much more predictable. It just won't be open to third party innovation in an arbitrary way that the PC is. It's like comparing apples and oranges.

Q: In just 9 days a half million PS-2s (remember the term PS/2 [the old IBM PC]?) are coming. How's the Xbox coming? A: Well, they're targeted to launch next year. Feels like forever, but it's like anything else in the software world, unless you announce and give potential partners a lead time you can never launch the product you want to launch. They have a very strong lineup of games developers around the world. The performance looks good. Game consoles have a very different business model than PCs. With game consoles you come out with this incredible and powerful device, and then you freeze the system definition for five years. The hardware definition doesn't change every year or every six months in game consoles like it does with PCs. So, they'll have a machine that looks very powerful when it ships and then in a year or so after it ships the basic PC will actually be more powerful. But part of the reason you get the stability and the predictability you want is you say we're not going to change that device.

Q: To what extent are you stressing the Xbox as an Internet access device? A: Xbox will have built-in networking capability, it will be part of the definition of the thing -- for instant messaging, Internet gaming, etc. -- but do they target narrowband or broadband connections? Strong head of steam right now saying they should only target broadband, but people say that may limit the potential, but they look like they will bet on broadband the way they bet on 16-bit and 32-bit OSs -- ahead of the market. Q: [Bob Metcalfe] Will there be an Ethernet jack on Xbox? {Bob invented Ethernet...] A: Absolutely! [Comment from Bob Metcalfe:] I hereby endorse the Xbox!

Q: What about CE and how it relates to PDAs, cellphones, and Internet appliances? A: They have essentially two codebases that they are investing in for the future: the Windows 2000 codebase and the CE codebase. The CE codebase will increasingly be targeted at low power devices. Embedded NT will still continue to consume more power. Set top boxes will be CE. Whether other things powered out of the wall will migrate away from CE remains to be seen.

Q: Does Microsoft have a philosophy on architecting for privacy, when people usually just take the defaults? A: The questioner made a good point [Brad Templeton] that users will do very little to protect themselves from privacy violations. It is amazing. People often times will not take the patches they provide. They have an opportunity to help give greater transparency in the user interface to let people know what's happening. They tried a beta of IE with lots of info showing about privacy, and then tested it in usability testing with high expectations, thinking they got it right and all. First they learned: no user understood it. It was just too complicated. Even though they were giving them fine-grained, granular control, users didn't understand it. They are giving this much thought and UI development time.

Q: What is Microsoft's perspective on Intellectual Property? A: He completely believes that Intellectual Property rights should be honored. There should be technologies that allow that. The hard issues will not be technology issues but rather the conversion of business models. They talk about the migration of the software world from applications to services. They are going through a business model change with .NET. How do we say how much to charge per month? How do you unwind the pricing of today's albums to a new business model for the recording industry. Industry by industry they and competitors will provide technology to help you enforce digital rights, but the harder problems in each case are business model ones.

Q: Stocks are down, way down, how do you retrain, attract and motivate employees, when (as Steve told Metcalfe earlier) so many companies chose to "outsource their compensation plans to Wall Street"? A: (Credited the "outsource" line to Jeff Bezos.) The big problem is not where stock prices are, the big problem from an employee perspective is where stock prices were. If they were at the level they are today and didn't go through the up and the down, they wouldn't have the emotional issues they are having today. People were counting on what wasn't real. (He's said this before, he says.) Since the market is off pretty broadly, employees can't say there's one company that will take off. Some of the mentality that every startup will soon have a market cap of billions of dollars is gone. That's good, otherwise they'd have problems. Repricing options is not an advised approach for them (accounting treatment is a nightmare). What you can do is issue new options. Have some old options? Here! Have some more! They did that a few months ago in order to address this issue. You will see more a balancing out of more predictable cash compensation along with stock. There will be innovative schemes.

People have to be excited about what your company is doing. People who leave them, by and large, leave because they are going to something they are excited about doing. Very few leave for compensation purposes. You can always work the comp issue. He does feel a little bit like watching the stock price is now the job of the HR department as much as anybody else around the company.

Bob Metcalfe: Thank you very much! A: Thanks, Bob! Thanks a lot!

Steve Ballmer next to Rod Canion, talking with people in a crowd after the speech