HOST:
Stuart Feldman, Director, IBM Institute for Advanced Commerce

TRANSCRIPT FORMAT:
Stuart Feldman's Remarks
Covers what the Institute for Advanced is, its goals, activities and progress to-date, along with its impact on the research community, IBM, and customers.

Analyst Questions & Answers
Questions on the impediments to e-commerce, auctions and IBM's auction software, virtual communities and electronic marketplaces, intelligent agents, distributed object frameworks, customer service, and attending IAC conferences.

START OF TRANSCRIPT

Stuart Feldman: Good morning and thank you all for joining us today. I'll take the next several minutes to cover what the Institute of Advanced Commerce is, what our goals are, some of our activities and progress to-date, and our impact on the research community, on IBM, and on IBM's customers.

The Institute for Advanced Commerce began on January 1 of this year. So we are a new organization.

We cross a number of organizational lines in IBM, because our goal is to help accelerate the creation, experimentation, deployment and serious business use of electronic commerce technologies.

We are interested in working with customers, with other researchers -- Especially people in universities -- to move forward the boundaries both of theoretical and practical knowledge in the entire area of e-commerce.

We of course have focused on a certain set of activities and a limited number of research themes. But they are quite a rich set. And we have had I believe considerable impact already.

You can see some of the work we've been doing by looking at the Web site www.ibm.com/iac. There will be more papers and more conference proceedings posted in the relatively near future. You are welcome to look and then come back periodically.

Also we have a mailing list. If you fill in a subscription on the Web site we will send you an e-mail every couple months or so telling you about new activities that we have underway and pointing out areas to look on the Web site.

The idea is to send you just enough information, not to overwhelm you with a 10 page e-mail or to send you spam every few days. So feel free to subscribe.

We've been attacking a number of aspects of recognition. One major activity we've had underway is sponsorship of conferences. This year we co-sponsored a conference on the electronic marketplace and economics in February in Texas.

We ran a conference on privacy in May for senior executives from large corporations. And if you look on the Web site you'll see some transcripts and audio clips from that.

In October we co-sponsored a meeting on information and computational economies held in Charleston.

In Zurich last week we held a conference on trends in electronic commerce for European professors.

We have planned at the end of January a conference on the impact of globalization, once again inviting very high level executives from customers mostly in Europe.

We have a conference on negotiation technology coming in March. We're co-sponsoring a conference on use of electronic commerce and Internet in April. We have several others upcoming.

The goal of our sponsorship and our involvement in these conferences has been to increase the interactions among players of a number of disciplines -- not just computer scientists, not just economists, not just lawyers but people from all of these areas discussing what are the key issues and how do we move forward.

I think that we have been remarkably successful in promoting some of this interaction. And from IBM's point of view we've had the opportunity to learn a great deal for our own technology experiments in this area.

The newsletter I mentioned earlier will announce conferences that we are involved with. So this is another reason perhaps for you to sign up.

In addition to such conferences, we're also working on specific hard technologies of our own within IBM. Much of the work is within IBM Research. Some of it is with customers at a very early stage.

The most advanced work that we have is on auction technologies. We recognized quite early that this was going to be very important in a business context.

We are creating software that will fit into the overall electronic Commerce business chain, not simply setting up a freestanding auction site, although it can be used as an auction driver.

We believe that there will be large scale transformations of the way many businesses act by getting into the world of competitive bidding rather than that of fixed price offers.

We appear to be right about the interest. We have a number of large companies who are using our software in an early beta stage.

We have versions of this available for beta download on the IBM Web site at this moment.

Again this software is distinguished by a very careful principled view of what auction technology needs to be in a business environment, not simply in an experimental environment -- one in which it is possible to have security, one in which it is possible to have proper scale, the ability to run many auctions with many people, the possibility of having widely different types of auctions, Yankee auctions, Dutch auctions, Vickrey auctions, 50 or 100 varieties or more depending how you count, and making it easy for people to use.

We are sufficiently comfortable with the work that we've done here that we are now branching out under the Institute into more advanced negotiation issues -- how do you do many-on-many negotiations, complicated negotiations involving bundling and so forth.

We believe this is going to be part of a fundamental change in the way people do business. And we would like to have a very strong research contribution here.

One reason we're sponsoring a specialist conference on negotiation technology is both to share our solutions and to learn how other people are attacking some of these.

We also have available technology for promotional and couponing offers. Once again our approach is to go after a general architectural solution, one that can be embedded and used as part of real business, not simply as a one shot experiment that has to be rewritten each time the retailer changes direction.

We have more future-looking technology work underway for people to control and understand and analyze customer access to Web sites for commerce, in order to understand customer behavior at a level that is useful for a merchandiser, not for a webmaster, and to present the right sort of statistical data to provide very real time information and also complex analysis of the results.

We believe this is a major direction, having looked at the difficulties people have guessing what customers are really doing when they visit interesting Web sites.

As a very long range research project we've been working on understanding what happens when you have thousands and eventually billions of interacting agents in an information economy.

As we move forward over the next couple decades we expect to see literally a billion or two billion people involved in the digital economy around the world. We expect that they will each have hundreds, probably thousands of software agents, software representatives.

And we are therefore very concerned and excited about how this future will play out. What happens when you personally have 10 to the power 12 (a trillion) software agents around the globe? What kind of wonderful economic opportunities does it bring? And more likely in the short run, what sort of technological catastrophes do we need to avert?

We have therefore been studying and publishing a lot in this area. It is an area of intense interest in a number of research institutions. Our eyes are on the 10 to 20 year time focus although we're trying to understand what the shorter term implications are.

So those are some examples of projects we have underway. Some of these have already been used by IBM customers.

I want to be careful to dispel any idea that IBM research and advanced development work is pure ivory tower. We are working very hard to move our ideas as rapidly as possible into the commercial arena.

We are very interested in solution work with leading customers. The Institute for Advanced Commerce is a way to help talk to customers and to find the right homes for some of these experiments.

Plus we are working to do not just flashy little demos but the grade of research driven technology that will realistically be usable by large scale enterprises that are going to be building their companies over the next decade on electronic means of doing business.

We talk to customers all the time. I've made presentations to over 40 so far this year at various individual meetings.

We talk to universities. I and my colleagues have visited more than 15 universities so far this year, in many cases talking to business school people as well as computer and other technology people, because we want to both bring these communities together and to understand the insight each brings.

So in sum, the Institute has been in operation at this point for a little less than a year. We've had the opportunity to create some new technologies, to try them in the real world, to move them into IBM product planning.

We have in addition had the opportunity to learn from and I believe influence the research agendas of people on several continents.

We're doing very leading edge and active work in electronic commerce. And we expect to be continuing this even more actively in the coming year. So I think at this point I will be happy to take a few questions.

Analyst Questions & Answers

1. Impediments to e-commerce
2. Auctions & IBM auction software
3. Virtual communities & electronic marketplaces
4. Intelligent agents
5. Distributed object frameworks
6. Customer service
7. Attending IAC conferences

1. IMPEDIMENTS TO E-COMMERCE

[ BACK TO ANALYST QUESTIONS & ANSWERS ]

Analyst: Question on your perception of the impediments to e-commerce -- where are they going to come from? And how are we going to deal with them?

Stuart: I believe that the main impediment to the growth of e-commerce is going to be creation and growth of customer confidence. And customer in this case means both business and consumer.

It will take innovative technology to make this trust work well - this involves privacy management. It involves security.

It'll also take time and experience for people simply to become comfortable with doing remote electronic business. It took a long time for 800 number catalog sales to become standard. It took a long time for people to get comfortable with walking up to an ATM machine.

So in part it's technology on the security and privacy side. Partly it's going to be experience. Partly it's going to be social pressures and economic advantage. That's probably the main driver.

Other issues such as the world wide wait and the delays in problems with the network - those problems will simply go away as we build up more network capacity, as people put in more servers.

Issues of internationalization are going to be very tough. But English will eventually not be the only major language on the web. People will be buying and selling in multiple currencies and multiple modes. That also is an impediment at the moment to worldwide growth, although obviously no problem for North America.

But I think the fundamental real issue is going to be building and deserving confidence in our overall structure.

Analyst: When you talk about e-commerce and people becoming more comfortable as we had to with 800 numbers and then ATMs what do you mean by a long time, since time gets compressed these days?

Stuart: Well it is not obvious to me that time truly gets that compressed. What you're going to get is a very rapid take-up. The people who are perfectly comfortable clicking and using computers, especially the under 30s plus your techie types: That population is just a given.

My mother never got up to the point of using an ATM, sorry to say. So there are certain parts of the population which are unlikely to become very active.

And my guess is that we're going to see, on the consumer side, a rapidly rising but a continuous penetration growth over at least a decade.

While on the consumer side there's going to be this long penetration, on the business side, large scale business already does as much as it possibly can using classic EDI, for example.

And EDI is simply going to be evolving very rapidly. It's another Research area that I haven't discussed. But the whole XML trend is going to move EDI rapidly into an Internet environment. We expect to see that trend moving not only the most committed and very large enterprises, but also moving down several tiers.

And so the time scale for business is measured in a few years for very significant effect.

We're starting to see some of this at the very high end where some people we're dealing with already are moving in this direction for '99. And there's going to be a lot of motion over the next few years as you move down into at least the medium enterprise.

2. AUCTIONS & IBM AUCTION SOFTWARE

[ BACK TO ANALYST QUESTIONS & ANSWERS ]

Analyst: I'm interested in focusing in on auctions. Can you name some companies that you're in a beta stage with?

Stuart: I'm sorry that I cannot name the individual companies. The people who are working most directly and early with us are looking for specific advantage.

So let me talk about them generically rather than by name. I apologize, but I really am not at liberty to name them until they go public.

There are at least four with which we are in some stage of deployment or experimentation at the moment.

Let me mention areas in which they are - process industry is one. People who have excess capacity and excess production and are trying to sell by auction rather than by fire sale. Some of these people are also immediately moving on to the much more interesting interactions that come when you want to sell your mainline products and reschedule appropriately all your process mills.

This is a very fruitful area. We're working with one company. We've had discussions with several others in this general area.

We are of course working with some financial companies that are interested specifically in exploiting auctions of financial type instruments. And they of course have a somewhat different set of criteria and time scales.

So those are two very realistic areas in which we have specific work. And these people are interested in making sure that the auction capability fits in with their overall story rather than simply a little auction on their Web site for the fun of it.

Another obvious customer that we're working with is IBM itself, which is of course a large retailer and business to business supplier. Again, I can't commit names until we actually go live on something.

Analyst: Can you be more specific on process industry?

Stuart: Like oil refineries, steel mills, people who are producing in many cases complicated and multiple products from a very expensive large-scale production capability.

And what's interesting here is that they can schedule their runs in a variety of messy ways and there are tear-down and set-up costs as they shift from one product stream to another. They tend to be commodity or specialist producers and also producers of very large quantities and value.

They are almost always B to B. And when you start looking at the details it turns out that they have a much wider variety of products than an amateur would expect. There are many more grades of hydrocarbon that come out of a refinery than you would expect. There are many more kinds of copper or steel that come out of mills.

Analyst: Do you find that the fact that they're doing multiple products is a key issue or just the fact that they're dealing in large quantities of sort of starting with commodity raw materials?

Stuart: The large quantity means that this is deadly serious business. And so they have a very different attitude toward what they're doing.

Also they're always dealing with business customers -- ones with which they are going to have a significant financial involvement, ones in which there are going to be logistics involved.

The fact that they have multiple products means that there are very tough issues of rescheduling of a mill. They don't just tear down a production line for producing one small item. But if it's high enough value, you might consider rescheduling. It turns out that's an incredibly complicated mathematical problem.

Analyst: So auctions are a way to deploy excess?

Stuart: Auctions are a way to get a value, to get a real sense of what the market values are.

It is also a way, if you can measure what your costs are, to maximize the value of your very large investments, like mills that cost 100 million or a billion dollars. So you're not lightheartedly rescheduling. And every few percent of efficiency is a huge amount of money.

Analyst: If you had to choose between the segments of the market that you think will be adopting this first, are you seeing it in more distribution companies or more in manufacturing companies?

Stuart: I guess it's hard to give a real answer because we are dealing right now with a few process companies.

Retail and distribution are obviously going to be very heavily playing the auction game because it's a natural way to attract customers in a better way than today's methods.

But those are going to have a very different character. The retail customer approach to doing commerce and the steel mill way have very different embeddings in their overall business models.

And so I would make the larger statement that I'm going to expect to see negotiation appearing all over the place, virtually anywhere people are buying or selling or bartering. And that it's going to be almost uniform change of business practice.

Analyst: You mentioned that the beta auction software is available for download on the Web site. Is it available to any company or only existing customers?

Stuart The beta is available to anybody who wants to give it a try. It runs on top of IBM's Net.Commerce software. The beta has of course some limitations.

I should mention that these are betas that are serious, not just Internet betas.

Analyst: That was my other question. It's going to be integrated with Net.Commerce?

Stuart: We will include it in the overall IBM commerce server world.

The downloads are for System/390 and NT at the moment. We have UNIX versions that will be coming. So these are available for experimentation at this point.

I can't commit product, of course. But we are encouraging customers to give these a try. We are working directly with a small number of customers to make sure that they are successful.

Analyst: On the auction technology, the financial firms you're working with, are you working with investment bank, brokers, dealers about auctioning their services, say underwriting capability, as opposed to just instruments?

Stuart: At the moment we're only discussing instrument type activities rather than selling their next contract or their next business deal.

Analyst: You've mentioned some ideas of how you feel that business is going to use auction open market negotiation more and more.

That however implies that there will be very little friction in the back channel, the logistics back channel that is the basis of many long term production agreements between one business and another business, which is that they have an information back channel which makes their businesses run together, the whole concept of an extranet.

The auction model makes it very difficult to have that kind of linkage and therefore would tend to drive costs up I would think.

Stuart: Let me be very careful. I did not want to suggest that auctions and competitive bidding were going to displace all other forms of business interaction.

If you were one of these new auto plants in Brazil where five suppliers all have pieces of the assembly line, obviously you're not going to go out and auction for pieces to somebody who isn't on the assembly line.

If you're Proctor & Gamble and Wal-Mart, then you have a deal that makes a lot of sense.

However what the new electronic realities permit is that if there are cases where you can afford to count on the open market providing what you need, because it's not one of the 15 goods that you absolutely need for your production line but just the goods that keep your operation running, then you may very well be as happy going out for bids.

But of course the bid is not just financial. There are also bids on terms and conditions. You know, "you can have this for $100 next week or $200 today" is a perfectly valid bid.

So I don't want to suggest that there's no reason for having long term, very stable arrangements for mainline business activities in many cases, where you only go out for a competitive bid once a decade.

But I believe what we're going to see is a rapid rise in the use of bidding on all the secondary activities where price or exact terms are far more important than comfort in your relationship.

3. VIRTUAL COMMUNITIES AND ELECTRONIC MARKETPLACES

[ BACK TO ANALYST QUESTIONS & ANSWERS ]

Analyst: However you could project that kinds of communities of need would arise where let's say all the windshield suppliers in the world get together and define a set of de facto standards by which they deal with the people who purchase their goods.

And this is in fact the kind of community - the financial institutions with open financial exchange are essentially doing the same thing. They're building a way for communities of "I have this service". What kind of services do you need to build up?

Stuart: If we take the extreme example of the auto assembly plants with multibillion dollar investments, you really don't want anything that disrupts that production line. And therefore you may make a bargain with the one of four companies in the world who can provide you brake assemblies. And so that's one extreme.

And then there's the other extreme of what we agree are now fungible securitized goods, where you know exactly what is the current interest rate that you're going to sell me at for something with this level of risk. And at that level there is very little reason to do anything other than competitive bidding.

And so this is a fundamental trade off between how much you can risk disruption versus how much the lowest possible bid on terms, conditions and price makes sense.

Analyst: I have been researching extensively in the area of virtual communities and the way in which they might affect electronic commerce and business to business relationships particularly.

Do you think that another major opportunity that you haven't really talked about yet is in what I'd describe as the white space between the formal organizational forms and between organizations rather than just in a straightforward transaction support, which is really what you've been talking about until now?

Stuart: A wonderful question, thank you. We are starting to investigate a number of directions of the world of electronic marketplaces.

And an electronic marketplace can be associated with an individual producer or an individual consumer of course. But it is much more likely that there is going to be some form of intermediary between them in an area which right now is likely to be quite literally a white space in the social and economic landscape.

The winners in this area in many cases will be creating perceived value either as neutral parties or as active participants in economic activity.

And we expect that, in many cases, they will not simply be a place where you go to make a purchase and leave, but a world which helps organize a significant chunk of your business activity.

And what is unclear of course is what are going to be the models that are most successful in various parts of industry. In some it's already clear that the producers will probably dominate as they do now. But in many others there are going to be whole new families of places where consumers and intermediaries talk among each other.

At a technical level we are looking at meeting places where agents talk to each other and what the standards are for inter-agent communication.

You may not have intended the question to open up to the possibility of non-human community. But we're expecting to see human-to-human, human-to-automaton and automaton-to-automaton types of communities.

Analyst: We're pushing the boundary on exchanging things. You can't always tie it down to a buy/sell transaction. What sort of work have you been doing in that area?

Stuart: We're starting to look very much at multi-party negotiation where among other things you may very well have more than two parties to a transaction. They need to share information in order to figure out what sort of business they will have together and then how they'll divide up the spoils and the work.

And at the mechanical level we're starting to look at how you manage the detailed interactions.

There's clearly a large scale social layer that we're also investigating in my own laboratory on how various people reasonably negotiate, discuss, get close to each other.

And it isn't just the final stage where a legal transaction gets closed. It's the early stages where you are exchanging pieces of information that are very important.

One of the research themes that we have been focusing on is the issue of the future of business interactions -- much more dynamic groupings of players, as in the Hollywood film model, in addition to some others where a number of previously uninvolved parties get together for a particular project, share plenty of information and then depart.

So this could be a very different social model and a very different business model than the long term supplier consumer model.

Analyst: And much more difficult to bring about and create when the parties have never met.

Stuart: Absolutely. One of the glories of the Internet technology, the new digital world is that we can have reasonable ideas of authenticity via certificates.

We can take advantage of possible enormous increments of bandwidth coming so that people can have long term video connections or sharing of large amounts of simulator data or business data among parties who have not actually dealt with each other.

We are interested in and have some work going in a number of these aspects. And I believe this is another one of the major changing areas. It's one of our seven research themes.

4. INTELLIGENT AGENTS

[ BACK TO ANALYST QUESTIONS & ANSWERS ]

Analyst: I'm wondering in your research in the negotiation area, are you doing much on the aspects of the legal issues of agent technology?

Stuart: We have not touched very much on the legal issues. We have been using agent technology in some work in the Tokyo Research lab, which created the Aglet - agile and mobile agent technology - and have now applied that to a currently running industrial mall for buying travel services. This is a very explicit use of an agent type technology.

We also have some work going on at our Zurich Research lab aimed at insurance sales negotiation, which is very much an agent type technology without actually being called an agent.

But in both cases we are using these in very restricted environments where there's no specific legal problem. And we're just using agents as a way of improving the interaction and improving the quality of what we deliver.

Analyst: So the agent isn't making a purchase commitment for instance?

Stuart: No in both cases. The agent on the customer side is simply acting as an intelligent filter. So that when a lot of offers come back you can get them ordered and presented in a sensible fashion.

Your computer can talk to the other computer for an extra round to refine the query. At no point are we, at present, offering empowered agents.

We have the longer term research effort -- the information economy stuff that I was talking about -- where we are simulating a world in which there are buyers and sellers and intermediate agents which are interacting in a variety of complicated ways, including use of machine learning technologies.

But this is a simulation environment, not a real-world negligence environment.

5. DISTRIBUTED OBJECT FRAMEWORKS

[ BACK TO ANALYST QUESTIONS & ANSWERS ]

Analyst: I am wondering about an enabling technology that's not yet been discussed -- distributed object frameworks.

Stuart: Okay, let me address that a couple ways. First of all, we take an object viewpoint as so much a given of the story that we don't even mention it. The future of documents is going to be active material that you move from place to place.

We accept as a given that object interchange, XML, is driven in part by being a nice flat way to communicate complicated types of documents which are themselves really objects.

This distributed access, distributed management is also a necessity in this world as we start scaling from a few thousand users to a few million to hundreds of millions.

Analyst: Could I follow up on that issue? As I delved into the object world I realized that objects were just the tip of the iceberg.

You need management services, an architecture, methodologies, people versed in those methodologies, modeling tools, so that a whole infrastructure of supporting technologies needs to be built up.

How do you see that occurring or how that would affect the impediments or the enablers for that supporting infrastructure for this object world that you take for granted?

Stuart: First of all, one of the glories and the impediments is the so-called legacy world. And of course we're creating new legacy systems every year. And that's not a joke, it's a real problem. There are now legacy CGI-bins and so forth.

But most of the interesting business type information is held in relational type databases. Most of the interesting documents are carried either in drawers in people's offices or in flat file systems. and there's a need to wrap all of these in an object management world in order to best manipulate them in the future.

IBM has some very large activities in these areas which I don't want to talk about at length in this conference call. But we are very much worried about being able to interconnect objects and to wrap previous data and previous prophecies in a modern object that you can manipulate.

And the point about having to manage these, to create them, to control them -- absolutely correct. There's nothing simple here. You have to model, design, analyze your bottlenecks, Especially when you move into the distributed world, there's nothing simple here. It's just that we are assuming that that is the direction in which people will be moving as they move whole enterprises into a full e-commerce and then e-business sense.

And yes there's going to be a lot of pain and a lot of learning. But this is a clear and essential direction.

6. CUSTOMER SERVICE

[ BACK TO ANALYST QUESTIONS & ANSWERS ]

Analyst: Could you update us on customer service related activities? You talked about buy and sell side stuff. What's going on in e-mail, auto-response, natural language query, parsing, a better search and configuration technologies?

Stuart: Let me address that only briefly. We have not been focusing on those issues at this moment under the Institute umbrella.

IBM Research, of course, has a number of efforts on mail classification, mail response, and a great deal on knowledge management in general. And IBM is creating a whole new organization to do customer relationship management.

I would rather just point at those as major corporate directions in general. Because those are already very large scale activities within the company.

They are obviously determinants of long term success because if customer satisfaction is not high enough people will not want to go to electronic means. So it's a clear necessary condition for widespread customer acceptance.

7. ATTENDING IAC CONFERENCES

[ BACK TO ANALYST QUESTIONS & ANSWERS ]

Analyst: When you have conferences and meetings do you allow analysts to attend as participants or observers?

Stuart: We have several kinds of conferences. One kind are just straight scientific conferences which anybody is welcome to attend. Sometimes you're asked to submit a few paragraphs explaining why you want to be there. But we'd be delighted to have you there.

There are some other conferences which are explicitly invitation only - I mentioned the privacy and impact of globalization conferences - where our target audience is senior CIO types or vice presidents of marketing.

But we invited a few analysts to the privacy conference. And we will probably be inviting a few to the European conference.

The main issue will be not overwhelming the room with too many people.But we would be delighted to have a few analysts attending.

###