Peder Saether Symposium (March 9-10, 2000)
Professor Lawrence Rowe Presentation
Prof. Lawrence Rowe, Electrical Engineering and Computer Science Division (EECS); Director, Berkeley Multimedia Research Center (BMRC), UC Berkeley
I have to wake up the audience. "I Will Survive" Animation (NB: say "yes" when RealPlayer asks if it should be played even with the mpeg error.)
Now, actually, there was a serious point to that. And the serious point is, that’s a piece of media that went floating around the Internet. Somebody e-mailed it to me about a month-and-a-half ago, and many of you in this audience have probably already seen it. So I grabbed that media and threw it up on our server so we could stream and show it here. But now think if you were teaching a class about animation, and you wanted to teach people about character animation. That’s a fabulous piece of character animation, there was a sense of reality to that character, and all it is is bits. So if I was teaching a class about animation and story-telling, that would be a great piece of media that I could use in that class. By the way, I have no idea who that guy is. Actually, I went and searched for him on the Web. He lives in San Francisco. He did that because he wanted to teach himself animation. But just think, there are lots of content objects out there available for us if we just go out and find them and select them and use them in our courses.
Now, what I want to do, rather than get into a bunch of technical stuff, is fly at 35,000 feet and tell you what those of us in the research community think we already understand, and what things we don’t understand. And then at the end, I’ll say something about where I think we should be spending our time, because the technology is changing so rapidly that what we work on will be impacted by what’s going to be out there in three to five years, more than what’s out there today.
So what’s known? The Internet improves educational delivery efficiency. We know this. Lecture capture, streaming media has value. Distributed collaboration, as was discussed by several of the previous speakers, has promise, but it actually has a lot of problems. Interactive multimedia content, such as shown by Professor Scheer, is fabulous, it’s the holy grail we would all like to have, but there are problems, and we need a lot more experiments on effectiveness.
So what do I mean by the Internet improves efficiency? We know about Web publishing of assignments, syllabus, and whatever, we know about communication. But what some people may not be realizing is that we also have to re-engineer our information systems on the campus to integrate those legacy applications into the systems we use for delivering education. As a faculty member, I can’t wait for the day that I can take a spreadsheet that has my student grades in it, send it off someplace, and it would get posted into the university system, rather than having to fill out sheets of little bubbles and that sort of stuff.
Here is a website--I just grabbed this from one of the classes in our department. It’s typical of what we see a lot of people starting to put up in the way of websites.
Now, lecture capture and streaming. What I showed you was streaming media. And, in fact, we’ve been using that kind of technology to capture lectures and to stream them. As we speak, we are webcasting 10 classes on the Berkeley campus. And we webcast them live, and then record them, so that students can then connect later and watch those lectures again or off-line. So we’re doing that to a number of very large undergraduate classes; the classes have an enrollment of roughly 3,000 people. And at any given time of the day or night when we look at the monitor, we see on the order of 10 students watching lectures. And when it comes time for mid-terms and comes time for finals, the demand goes up dramatically. One interesting fact is that students use the lecture webcast not to watch the lectures remotely, live, but what they do is they use them to review. And the question we don’t know yet is, are they using it to skip those boring lectures, and then watch them at fast speed later? Okay, my point is improved education for local students. It also provides access for remote students. And there are many universities getting into the business of doing this. This is a relatively easy thing to do, and we should see a lot more of that sort of thing happening.
The production cost varies depending on the technology and the content quality, but we’re talking on the order of $50,000 to $100,000 to get in this business, not on the order of $1 million. What do we need? Well, we need better interfaces.
So here is a lecture browser that we have been experimenting with here at Berkeley. And what you see is, on the left, the video of Professor Alex Pines giving a Chem 1A lecture. In the lower right-hand corner you see the board that he’s drawn up on the blackboard in the large lecture hall on campus, and that’s showing the slides that he presents during his lecture. We think it’s very, very important that when you do streaming you should do two streams, one on the instructor and one on the content, and the two should be synchronized. So we’re thinking of a different kind of world than just television with a single stream. On the far left, you see a slide index. While the lecturer is talking, you can step forward and look through the slides, you can go to any slide that you want. You can find the slide, ask the video to synch to where the slide is, or you can say synch the slide back to the particular video. And a very important thing that we’ve started to experiment with, but which the students tell us they really want, is that search box in the upper right-hand corner. Now, the search box allows them to come back to the lecture material at the end of the semester and say, ‘I have a question. Where did you talk about sodium phosphate?’ They want to know where in the lectures you talked about that. Now, we can build those search indexes, but the problem is getting the content. So we’ve been getting the words both from taking the text off the slides, if people use Powerpoint slides, but we’ve also been experimenting with some folks at IBM using some of the speech-to-text tools to try and automatically generate those indexes.
Distributed collaboration. Remote access to people with instruments. Some conference models work; one-to-one conferences work pretty well. Probably the cheapest way to do it is to get Net Meeting from Microsoft and a couple of PC’s. Small groups can be made to work, but it costs more money and it’s more complex. Large one-way broadcasts where you have a single speaker talking to huge numbers of audience in television works really well too. But the design and operation of the facilities and the programs is, in fact, somewhat expensive. And you heard earlier people talking about the fact that audio is a problem. It’s more than just audio. I mean, audio is a very challenging problem, but there’s more than just audio to this stuff. Some important issues--you need high quality audio and video. Forget it, you’re never going to get TV quality at a 100 kilobits per second. The good news is bandwidth is on our side, as I’ll talk about in a bit.
Second really important well-understood principle is the sense of presence. One of the other speakers talked about the fact that you want to be able as a lecturer to see people at the remote end. We know that. The problem is what happens if you have a class of 4,000 people spread at a hundred different sites. How are we going to present that material, what the people are doing, to the instructor? What sort of interface would you present to them, so they could see the remote audience? So it’s a really challenging problem. And, lastly, somebody else mentioned it, and I’ll also say it, a seamless interaction, so that you can interact with people remotely without having to have five technicians and a bunch of special technology. This is really crucial.
Interactive media content. I think the earlier speakers have covered this stuff and said the right things. But remember, we have to compete with space wars, sex, and game shows because people see this on TV, and these boring talking heads are not going to cut it.
Okay. What do we not know? I don’t think we know the best design for the teaching and learning spaces. I mean, we know what a large lecture hall should look like and what it should do. We’ve had several hundred years to refine the model of what that space should look like. I think one of the pictures that I saw earlier today was very interesting, which showed the tables looking down upon the speaker, with projected screens on both sides plus a board in the middle. That’s a very interesting possibility, a lot of people are experimenting with that. But one thing researchers should be thinking about is more radical ways to think about what those spaces would look like, and they should be rapidly changing them, evaluating them, and moving on to the next choices.
The other thing is that we must come up with lower-cost techniques to produce content. The movie production model that everybody talks about with teams and specialists and all that, that works just fine and that’s definitely a way to work. But, unfortunately, they may be prohibitively expensive for all but the very large classes which have a large economic model to support that development.
And the other thing we would like to have, and I’ve seen some people playing with it, but we don’t really know how to do it yet, is that different people come to the content with different backgrounds and experiences, and you would like the content to tailor itself to what they need. A manager doesn’t need a detailed low level description, but somebody who is the lead engineer working on a particular topic does need a detailed lower level description. You would like the same content to be able to adapt to both of those situations. It’s a very hard, challenging problem.
Here’s my picture of our studio classroom. I threw it up here just to give you an idea of what we’re up to. The most important thing is that we have multiple cameras. In fact, we have three cameras in the room right now, and I’d like to put three more because my view is we should be producing and publishing multiple streams, going in all directions, with control by the end user.
We need models for seamless interaction with large scale distributed conferences. I think I said that before.
And if anybody knows how to solve this next problem, I will take you out to dinner, I’ll even pay to fly you back if you want to come back to California when it’s sunny. It’s very tough to hire and keep this IT staff. We’re in the middle of an Internet dot-com gold rush. It will end eventually, you know, we’ll have a 20, 30, 40 percent crash in the stock market, but until then, it’s nearly impossible to keep these people in the university, working on the kind of problems that we’re working on. And sadly, even though I think that the administrations have moved aggressively to try and scale up the production of these kinds of students, we’re not able to do that very rapidly. And the production of Ph.D.’s to be professors in this field is also very hard, because those are the very students who are quitting with a Masters and going out into the Internet companies, rather than sticking around to get their Ph.D.’s. So that’s a challenging problem. Anybody who has got good ideas on that, please grab me.
So here’s a quick assessment of the future. Several people have said this, but I think elite universities, certainly, and many other universities will exist for the foreseeable future. These new technologies we’re talking about are very well suited to on-demand training and continuous learning, and that’s where they’re gaining the most success. But I’m not so sure that they are as effective in formative and graduate education. I was interested in the talk this morning about the group trying to do Ph.D. education at a distance. I think that’s a challenging idea, and it would be interesting to see how well that works out. And I think we all recognize that education, particularly formative education, actually encompasses a lot more than what happens in the classroom. And that’s the important part of the process of maturation and accreditation that was mentioned in the Degoot, Brown, and Collins paper that was cited by the previous speaker.
Now, to the real meat. Communication bandwidth will be relatively free in five to ten years. One of the things that we have to do in computer science is project where the world is going to be in five to ten years, and then start thinking about how to build systems for that new environment. Back in the later 70’s, we said CPU cycles, which at the time were very expensive, would be relatively free. Today, they are so free, they’re not even worth a penny, because we set five computers down and use them for an hour or two a day, and the rest of the time the CPU cycles are sitting there doing nothing. Okay, CPU cycles are free. Next thing that came along was large color displays, and graphics became free. And what we’re not predicting is that in five to ten years, communication bandwidth will be free. Frankly I was a little surprised at the Swedish-Stanford project that’s doing stuff at T1 speeds. T1 is a very fast circuit, but, in fact, in Internet-2 we’re running links that are on the order of several gigabits per second, and we’re trying to figure out how to use that capacity. So as researchers, we want to be thinking about what to do with that very high speed bandwidth. One thing to realize is that we’re in the midst of a transition from a very few broadcast television channels to thousands of Internet webcasts. It started when we had few TV broadcasters back in the 50’s and 60’s. Cable systems introduced new channels, and we saw a proliferation of new channels. We’re about to see Internet distribution of television, which is going to explode the number of channels you will get, and eventually we’ll have this running over the Internet to all homes. And then, basically, everybody who wants to be will be a TV producer. So I think researchers should focus on effective use of bandwidth--how many video streams and what sort of simultaneous interactive experiences you can do with those.
So to that end, here’s my pitch for the world of the future, which is Internet webcasting: multiple video streams, varying picture size and quality, interactively between subsets of viewers, and a very strong belief in the use of intelligent computer software to improve the quality and reduce the cost of production. Here are some examples of the sort of things we’re talking about.
This is the way the world will look in our opinion, where you have lots of people sitting at--I’m sorry I didn’t have a TV image, all I had was this stupid little computer image--but sitting at TV-like interfaces with remote controls that might, in fact, be keyboards. End-way communication, multiple streams, and discussion threads. User control of that content, you decide which of the ten streams that are coming out of that classroom you want to watch. There will probably be a moderated feed produced by somebody who knows what they’re doing. But let’s take an example from sports. Suppose you’re looking at a soccer match. You know in a major match they may have 20 cameras, and the TV producers will produce a moderated feed of good information for you to watch the program. But suppose you want to watch the goalie for a particular team, there’s probably a camera on that goalie in case something happens. There’s no reason that you should not be able to have on your screen both the moderated feed and a picture of the goalie that you could watch and concentrate on. This will be possible with the amount of bandwidth that’s going to be available out there.
So my concluding remarks. Virtual universities will exist when the interaction with a colleague in Sweden or Norway is the same as the interaction with a colleague next door. People said it earlier, it’s absolutely true. We decided to come here and meet. There’s a lot more to this meeting than the time that we spend in this room, and the quality of the interaction is very different than what it would be if we were conferencing. The same can be said that using a library in Asia is the same as using the Bancroft Library here. Now all of the journals have been definitely moved on-line. It’s definitely the case that those of us in fields where accessing journals is a key part of our business are not going to go to the library again, we’re going to be using all the on-line journals. But there’s still a lot of important, historical material over there in the Bancroft Library that we’re not going to put on-line. So I think that libraries in some form, in reduced forms, are certainly going to continue for a long time.
Second, the pitch to those of you who have dollars in your pocket, these experiments are very expensive and they’re very risky. And we should expect to see a lot of money spent and a lot of very large, ugly failures. But my point is, we must continue to experiment so that we can exploit this technological opportunity. And somebody else said it earlier, there will be a competitive advantage to the person who figures it out first, and actually integrates it into their university completely. Do I think it will be the University of Phoenix? I didn’t think so before, but I must admit that the comment earlier, that they’ve gotten a dose of reality in terms of their own marketing pitch, says maybe they’re figuring out the world isn’t as easy as they were pretending it was. But I still don’t think they will be the winner in the long run.
© 2006 UC Regents
Last modified: 15 April 2006 | e-mail