Team creates LEDs, photovoltaic cells, and light detectors using novel one-molecule-thick material.
Gregory Jackson, director of academic computing at MIT, recently discussed his views on the state of academic use of on-line resources such as Athena and the World Wide Web with Tech Talk assistant editor Alice Waugh.
Waugh: When Project Athena was being planned and getting off the ground, was there any anticipation of the explosion in the World Wide Web and how big it would become?
Jackson: I think everyone anticipated that sharing information across the network would become routine, but no one knew exactly what form this would take. The key to the Web is the idea that client and server machines interact through specified protocols, and Project Athena pioneered and therefore anticipated this client/server model.
But the current explosion has limits. The Web only does some things well, especially moving files from one computer to another for display or downloading. The Web isn't a good way to run other software, and so you can't use it to run a really sophisticated plotting program, for example, or a computer-aided design program; you really need some local computational power for that. Basically the Web is about moving information over the network, and the explosion of academic computing extends well beyond that.
Waugh: What are some WWW academic applications right now?
Jackson: There are two ways to answer that question. There are a lot of subjects that do routine things in new ways using the Web-for example, putting the syllabus up and linking it to problems sets and solutions, and also maybe some readings and the instructor's home page, and perhaps including a simple form for students to communicate with TAs. Now I happen to think that's a spectacular move forward, even though there's nothing all that cutting-edge about any of the pieces, because it just dramatically changes the idea of where you go to get routine information about a course from a chair outside the instructor's door, which for years has been the way to do it. But there are also many subjects that go beyond this "routine things in new ways" use.
Professor Abelson's 6.095 class, for example, not only had all the readings and other materials online, but also set up a discussion forum where every participant had a pen name so that discussion could be both anonymous and linked. Then the course made the full transcript of the discussion available to everyone. Then the final papers had to be submitted as Web URLs rather than on paper, and the collection of student "papers" was linked to the 6.095 class page. It's a wonderful use of the Web to extend classroom boundaries across time and space.
The Web page for the Experimental Study Group's course 7.01 is also very nice in a lot of ways--it's a full textbook, with lots of examples and self-quizzes, links to external databases and so on. The 2.70 page has some very nice video of work in machine shops, which is very useful to watch over and over. There are some animations of current-carrying wires moving through magnetic fields that John Belcher did for 8.07. This takes dynamic mathematics that's always been very hard for students to visualize and makes it visible; there have always been still pictures that people could look at, but they didn't convey how the pictures would change if you changed the mathematical values. By next year you'll probably be able to generate the animations on the fly, so you'll say "move the wire so many feet per second" and back will come the animation of that happening.
Waugh: Do you think it will cut down on the use of paper?
Jackson: Not noticeably. What it'll do is cut down on the use of telephones-phone tag and things of that sort. I already go through days where I don't have a phone conversation with anyone, but I don't notice my mailbox any less full. I keep hoping that will change, but it hasn't yet.
Waugh: Do you think e-mail replacing phones is a good thing or a bad thing?
Jackson: It's a good thing, because it usually takes just one interaction, as opposed to the five phone calls back and forth before we finally talk. Phone calls have to be synchronous, unless you want to leave voice mail, and I much prefer leaving e-mail to voice mail, because I can look and see what I said before finally saying it.
There is some concern that electronic communication will replace richer personal interaction. The converse is that there are lots of folks, particularly among students, who won't speak up in person but who are perfectly willing to ask a question by electronic mail, and they won't call professors or TAs either because they have to talk and sort of halt and hesitate, whereas by e-mail they're perfectly happy to ask questions, make observations and communicate. This is something we're finding is becoming quite well established, not just here but a lot of other places. There probably are some personal contacts that get replaced by mail, and everyone's got to evaluate that and work to minimize it.
Waugh: How fast is academic use of the Internet growing?
Jackson: It's hard to say because it's hard to break it out-it's difficult to say what you call academic. The answer is: very fast. No matter how you look at it, Internet use seems to double every year, and it's hard to believe it's very different for academic use. But there's another more local measure of how fast things change. In a report I wrote less than two years ago [Education and Computing at MIT, fall 1993], I'm pretty sure the word "Web" doesn't appear once. I knew what it was and had played around with it, but it simply wasn't important academically at that point. If I wrote the same report today, every other paragraph would have "Web" in it.
Waugh: Will the World Wide Web eventually replace Athena or TechInfo?
Jackson: It used to be that Athena was the only way of distributing all kinds of information, and the Web's certainly taking over much of that load. But as I said before, much of academic computing is about computing rather than networked information, and Athena still provides that more efficiently and better than any other system I know. We'll see more and more use of the Web, but we'll also see more and more use of Athena-for example, over the past year or so, Athena use grew by about 20 percent. It's done about that every year I've been here, and I'm sure it will continue.
TechInfo is a tougher case. TechInfo today is, by and large, a feed into the Web rather than something in its own right. Very few people look at TechInfo with a TechInfo browser any more. If I look at something in TechInfo with a Web browser or a Gopher browser, it looks the same as it does with a TechInfo browser. There's little reason to invest our money in a new TechInfo browser if the world will provide a new Web browser. But TechInfo also has a server side, and that provides some real value. There are a whole set of controls and authentications and the like on the TechInfo server such that if I go and look at [a document] in TechInfo, I know the News Office (or whoever) put it there and somebody else didn't. If I'm looking at a Web page, there's no easy way for me to get that same assurance. In that sense, the Web doesn't make TechInfo obsolete, but does change its focus.
Waugh: Do you ever find that faculty members are altering the method or content of their courses in response to the on-line resources and computing that are now available?
Jackson: That happens all the time. I already talked about handouts on chairs giving way to Web pages. On the computing side, probably the biggest example is that more and more faculty are giving problems sets where the mathematics can't be done by hand. They'll give a real problem-here's the spacecraft and the real data that came back for signal/noise and Fourier analysis, or a real bridge on real soil, or a real Superfund site, or a real receptor on an AIDS virus. The availability of Athena workstations and professional-quality software make it possible for students to do problems like this; when I was a student most of our problems were "toy" problems.
Waugh: Do you foresee a time when Internet adjuncts will be standard or even indispensable to teaching?
Jackson: Standard, absolutely. They're coming close to that already. They'll never be indispensable, in that you'll always be able to teach as you learned, but it'll be very expensive and awkward to ignore the technologies we've been talking about, especially if you start talking about distance education in any serious way.
One interesting question is what level of computer is "standard" or "indispensable" for, say, an undergraduate or a faculty member to own. Only four years ago, the price gap between a low-end workstation and a high-end personal computer was around $10,000. Today the gap between a high-end Macintosh, say, and a low-end Silicon Graphics workstation at the MIT Computer Connection is maybe $1,000. That kind of convergence is a real challenge. It seems clear to me that at a place like this everyone needs a personal computer and access to more powerful central facilities, but it's harder and harder to define the difference between the two.
Waugh: What feedback have you had from faculty and students? Are Internet applications creating more or less work?
Jackson: More. People seem to want to do more and more-faculty pound down the doors trying to do new things, and students keep coming up with new ideas. Yet except for chairs with handouts in the hall, not much seems to be going away. My online list of Web uses in MIT subjects keeps growing-I update it maybe every two or three weeks, and each time I update it I add add 10 or 15 new subjects. That's only a subset of the subjects that use the Web, and there are other subjects that use computing beyond the Web, both Athena and departmental. Faculty and students are experimenting all over, and they're starting to expect certain technological applications of one another, so by and large it's more work for everyone.
But we're also getting more value, and I suspect there's more of a return to the kind of extra work we've been talking about than to other ways we might try to improve education at MIT. So the extra work is worthwhile.
Waugh: Is there concern about people taking Web documents created by a faculty member at one university and using them at another?
Jackson: Sure. They can't do it legally, of course; basically everything you create is copyrighted as soon as you create it. It's possible to restrict access to materials, but that works against the wide accessibility the Web is supposed to promote. There have been enough requests for us to be implementing a facility whereby someone can restrict things to the MIT network, but I think people will want to think carefully before they use it.
But whether you can restrict material is different from whether it's legal to copy it. You can photocopy an entire book and sell it, but that doesn't mean you may do so legally. We can restrict access to Web materials, but we can't make them impossible to copy. If you try too hard to prevent some kinds of access, you end up degrading what you're trying to do. For example, some institutions put electronic "watermarks" on images they distribute on the Web, but this degrades the fidelity of those images in the course of making them hard to misappropriate. The only way people will treat intellectual property appropriately is if they understand intellectual property rights, and for the most part we don't educate people very well in this regard.
I think if there were an easy way to pay people for the rights to use something one finds on the Web, then folks would generally do it, but no one's been able to figure out how to do that simply and securely.
Waugh: Are there financial constraints that could put brakes on continued growth of academic use of the Web?
Jackson: Absolutely, but oddly enough, they have more to do with infrastructure and support than they do with computers and software. Our academic-computing budgets have been pretty level for the last few years, maybe even declining in real terms, but my concerns are less about available funds than about how we allocate them.
Hardware, when all is said and done, is pretty cheap. If you presume that [an administrator or faculty member] at MIT has to have a machine that costs, let's say, $3,000, and that it has to be replaced every three years, what you're saying is that it costs about $1,000 per year for that person's hardware. For someone making $50,000, that's two percent of salary, which is relatively small compared to the 11 percent in the benefit rate that goes to graduate student tuition or the six percent that goes to medical plans. We need to view this simply as a cost of having an employee, like space and power. Putting in a network connection and buying software add a couple of hundred bucks onto the hardware cost-now we're up to maybe three percent, tops-but the sum total of putting someone on the network with a reasonable computer just isn't that large.
The hard part is, if we're not really careful about encouraging people to do similar things in similar ways that scale efficiently, we can create very complicated help and support problems; when something goes wrong, there's a lot of finger-pointing, and it's hard to figure out what needs fixing. Someone very skilled is necessary to help users fix problems, and if feedback loops don't work right, more people mean more problems. In this case, we can't afford to succeed.
We have to invest a whole lot of attention up front in making things in such a way that a user can get out of trouble in a very reliable way without a terrible cost. For example, none of the Athena clusters has assigned attendants, but when a machine gets totally disabled, which it will do now and then, it typically takes only two keystrokes to restore and reboot it. There are fewer than 20 people who operate and watch over the entire Athena environment of 1,000 client workstations when it's operational, which is an extraordinarily low ratio. We could buy less expensive computers, and invest less in developing automated installation and recovery systems, but we'd end up paying far more to operate them. The great danger is that we'll forget this as academic computing and networking proliferate, and begin thinking that hardware is the major cost of academic computing and that support costs will decline with computer costs.
There's also the simple fact that innovation takes ideas, skills and support. We have five Faculty Liaisons [for academic computing] and a few other people in my group who help faculty with all kinds of things, but their fundamental goal is to help faculty realize that there are interesting things to be done with technology and then to help faculty do those things. (And here's my one commercial: I encourage any faculty member interested in computing for education at MIT to call on the Faculty Liaisons, x3-0115, or
Another great danger, beyond assuming that support needs will decline, is that we'll confuse support for innovation with support for operation, and sacrifice the former to meet increasing demand for the latter. Fortunately I think the current reenginering projects in Information Systems are going to emphasize the distinction between innovation and operation, and to help MIT faculty stay at the academic-computing forefront.
Waugh: What applications or trends do you foresee for the future?
Jackson: The big things people talk about, besides cheaper computers and a fancier Web, are wireless technology, portable computing, advanced visualization, and dramatic increases in home computing and connectivity. Some of these are already affecting us-the visualization projects already going on in [the Departments of] Architecture and Chemistry, to pick two, dazzle me every time I see them-and the others will soon. What we have to keep remembering is that the goal isn't just to use all this technology, but to use it for education.
A version of this article appeared in MIT Tech Talk on March 8, 1995.