The Tech - Online EditionMIT's oldest and largest
newspaper & the first
newspaper published
on the web
Boston Weather: 73.0°F | Fair

WEB COMMUNITIES AND FREE TUITION

Philip Greenspun talks about the Future of the Web and MIT

By Aileen Tang
STAFF REPORTER

Laboratory for Computer Science researcher Philip Greenspun has been through the complete MIT life cycle. He graduated from Course 18 with a BS at the age of 18, received a Course 6 SM in 1993, earned his PhD in Course 6 last year, and is now teaching the first hands-on class at MIT about building database-backed web sites: Software Engineering of Innovative Web Applications (6.916).

As an entrepreneur, Greenspun has started six companies and buried three. His current company, ArsDigita, builds database-backed web sites for Fortune 500 companies for about $1 million per project. He also runs his own web site at http://photo.net/, which receives 700,000 hits a day.

As a software engineer, he built an open-source Web-based collaboration toolkit, which is now used by thousands of Web publishers and millions of users. In addition, Greenspun’s photographs have been featured in magazines and books. His most recent work is Philip and Alex’s Guide to Web Publishing, a coffee table Web-nerd book in 4-color printing with Greenspun’s own photographs interspersed throughout the text. He does not sell his photographs, but gives them away for free on photo.net if a person is willing to donate money to animal charities. photo.net generates about $13,000 a year for charity through amazon.com referral fees and orders for photographic prints.

Book tells how to harness the Internet

Everybody knows the Internet is “the future.” The Internet Service Providers industry index has reflected this common wisdom by more than quadrupling in the last 12 months. As people scramble to log on to E-Trade so they can buy 500 shares of Amazon or Yahoo stock on margin, few actually stop to remind themselves that companies like Amazon have a market capitalization of $21.5 billion but lost $124.5 million on $610 million sales (for the fiscal year ended 12/31/98, source: Yahoo Finance).

Why? Because although it is now conventional wisdom that the Internet nourishes the biggest market potential of the next century, no one yet knows the best way to milk this cash cow (at least not until a company like Amazon surprises everybody with its price-earnings ratio).

In his new book, Philip and Alex’s Guide to Web Publishing, Greenspun offers his own insight about how to build useful Web-based services and harness the Internet technology to benefit users, if not to make profits. He predicts that the ubiquitous Internet will provide a medium for “computer-mediated collaboration on a scale we can’t imagine.”

In Chapter 2 of Philip and Alex’s Guide to Web Publishing, Greenspun writes, “For example, in 1998 if people from Companies A, B, and C need to work together, they’d expect to be able to call up the phone company and ask it to set up a conference call in 15 minutes. In 2018, it is possible that cross-company collaboration will be far more prevalent. In that case, people will expect to be able to ask the phone company to set up a Web-based collaboration environment, in 15 minutes.

Thus an end to the current paradigm of large system administration budgets, the personal desktop computer, and selling boxes of software at computer stores much like "selling tables and chairs.” Commenting on today’s narrow-minded concept of the Internet, Greenspun writes, “Now we have an Internet and any computer in the world can talk to any other. But sadly it turns out that they have nothing to say.” He proposes to automate a lot of tasks online through the use of collaboratively evolved data models.

Greenspun paints the scene of a home in the not so far away future where every appliance has an IP address and is therefore a Web browser. “My GE Profile range already has a tall backsplash with an LED display. If GE had put a 10base-T outlet on the back to provide technical support, the next logical step would be to replace the LED display with a color LCD screen. Then I would be able to browse recipe Web sites from my stove top. Once I’d found the desired recipe, I would press ‘start cooking.’ A dialog box would appear: ‘JavaScript Alert: Preheat oven to 375?’ After I’d confirmed that, the recipe steps would unfold before me on the LCD.”

Interview with Philip Greenspun

Greenspun in person has a unique sense of humor and assertively expresses his own opinions. In this interview with The Tech, he discussed why he kept coming back to MIT, his vision for the Web, his company, his views about the Scott Krueger incident, his gripes about Bill Gates, and why he gave out $100 bills to each undergraduate student he guest lectured at MIT last year.

The Tech: How did you get the idea about teaching a class like 6.916: Software Engineering of Innovative Web Applications?

Greenspun: The problem was getting the idea out of my head, not into it. Since 1993, I haven’t been able to stop talking about Web-based collaboration. I spent 23 years learning how to program and I finally feel that I’m able to do something useful with that skill. The traditional 6-3 curriculum doesn’t cover the fundamental technologies necessary for supporting Web services, e.g., the relational database management system (RDBMS). You need the RDBMS to keep concurrent users from colliding. The RDBMS requires students to program in a declarative language (SQL), which is nothing like the procedural languages they might have used until now (Scheme, C, Java).

The Tech: What do you want students to take away from this class, besides the material from the syllabus?

Greenspun: I want them to learn how to focus on the user. You can’t be a great engineer unless you measure your creations against the actual user experience.

The Tech: In what ways is 6.916 different from a traditional MIT Course 6 class?

Greenspun: I dragooned five or six expert programmers, each with 20 years of experience, into serving as TAs. Thus with one staff person for every 3 students, students are exposed to how great software engineers think. There is no substitute for sitting down next to a great programmer to attack a problem.

Given the tuition that we’re charging, it ought to be the case that a Course 6-3 grad is worth an extra $200,000 per year. Yet companies that need software engineers are happy to hire grads from non-CS majors and pay them nearly as well, because they know that after a year of industrial experience, someone who was, say, a physics major would probably be a better software engineer than a raw Course 6-3 graduate.

Does this happen in biology? No. A biotech company would not hire a math major. Does this happen in medicine? You probably wouldn’t ask me to take out your tonsils.

Computer science education at MIT starts off very strong with 6.001 and gradually fizzles out. There is simply no evidence that a person can become a great software engineer by taking classes. So I’m trying to take us back to the Middle Ages with apprenticeships. Of course, we don’t have enough great programmers on campus to provide 1:1 instruction for everyone. However, some day, Web-based collaboration may allow us to take advantage of all the MIT alums that have become great software engineers and have Internet access.

The Tech: In “The Book Behind the Book Behind the Book” (http://photo.net/wtr/dead-trees/story.html), a story about what you went through to publish your first book, you described how idiotic the “Teach Yourself Blah Blah Blah in 21 Days” computer books are today.

So what suggestions do you have for the MIT student who wants to avoid buying books that are “written by idiots for idiots?”

Greenspun: Actually I think the most interesting thing about “The Book Behind the Book Behind the Book” is that it is about 30 pages long. No magazine would ever run an article that long. No book publisher would ever produce a book that short. It is a tale that could not be told in the commercial publishing world and only exists because of the Internet.

Anyway, if you really want book shopping advice I think there are two classes of good books. The first has a step-by-step tutorial for doing what you need to do today. Some of these are “I stole the program and now I need a book on how to use it” books. You shouldn’t care who writes these. The second class of good books is written by someone who is describing his or her life’s work. Edward Tufte’s books on information design come to mind. He spent seven years on each one. For example, Tufte’s Visual Explanations, in four pages (pp. 146-149), manages to set forth everything important about Web design.

The Tech: What made you like MIT enough to keep coming back despite having owned six start-up companies? Many MIT students would leave this place in a jiffy for a start-up.

Greenspun: To me, MIT is the Nerd City on the Hill. We are a community of people passionate about pushing science and technology forward. What I love about MIT is that if you said that you were staying up all night to find a more elegant solution to a problem, nobody would say “Why work so hard when you can just go to business school and have an easy life?”

That said, I think we can push ourselves to do better. We could have apologized to Scott Krueger’s family instead of leaving the matter to the administrators we’ve hired. We should get out of the business of asking students how much money their parents have. We should get out of the business of expert witnessing in lawsuits. It does not help MIT [nor] help their students find jobs. It does not build their tech skills (since patent disputes are usually about technology that is 15 years old). It does not push society forward. We should ask ourselves “If it is just about money, why don’t I work at a higher-growth higher-profit organization like Microsoft or Oracle?”

Campaign for Tuition-free MIT

In a Web page titled, Tuition-free MIT, Greenspun argues why it is morally questionable for MIT to charge tuition at the maximum level a student’s family can afford, and how MIT can change its infrastructure to run without charging tuition at all.

Greenspun put his words to action last spring when he guest-lectured an MIT class on designing database-backed Web services. He calculated that the students were paying about $80 in tuition per lecture-hour, and to “stop personally participating in the system of extracting money from MIT kids and their families,” he handed out a $100 bill to each undergraduate in the class.

The Tech: What is the theory behind your campaign for Tuition-free MIT?

Greenspun: It is explained at some length at http://photo.net/philg/school/tuition-free-mit.html and short summaries tend to distort my argument, so I’m chary of ripping the bones off the argument. To encourage folks to visit my essay, though, I’ll ask a few questions:

1) If we are such great engineers, can’t we find better ways to raise money than beating it out of 18-year-olds and their parents? If we aren’t such great engineers, why are we teaching?

2) If we can’t get as much money from the Feds and the Fortune 500 and the crotchety old rich nerds as we’ve gotten from bleeding students and families, then perhaps we can run MIT for less. Do the students need administrators to draft alcohol policies for them? Do the students need MIT-managed dorms? Do they need MIT-supervised dining services? Athletics? If they weren’t paying $24,000/year in tuition, maybe the students would be willing to manage some of these things for themselves.

3) If we were charging $1 million/student, would any rich person give us money? If not, why don’t we think that we’re already losing a lot of potential donors because rich people think that maybe we can take care of ourselves? What we don’t consider is that each extra dollar collected from students makes it more difficult to collect from donors.

Building online communities

Greenspun’s Web site, photo.net, fosters an on-line community about various topics that he is interested in: photography, building database-backed Web sites, travel, Bill Gates’ wealth. The front page begins with this quote, “I built this site in 1993 to share what I knew. In 1995, I expanded the goal to also share what some other folks know.”

The Tech: What inspired you to build a site like photo.net?

Greenspun: It started in 1993 after I returned from spending the summer driving to Alaska and back. I wrote a 200-page online story about the trip, Travels with Samantha. I illustrated the book with 250 photos and it generated a lot of emailed questions about photography. Thinking that I could reduce my email burden with an FAQ, I started building up some photography tutorial pages at http://photo.net/photo But photography is sufficiently open-ended that answering three questions will raise seven more. So I needed to develop technological means [such as online discussion forums] by which users could answer each others’ questions.

The Tech: Many people who read your books and/or your stories on photo.net appreciate the humor and knowledge they obtained from these writings. What motivated you to go and write about all these things?

Greenspun: I think it was my programming background. Due to a combination of altruism and vanity, great programmers tend to want to share their source code so other programmers get a leg up and don’t have to reinvent the wheel. The original programmer thus gratifies his or her Hacker Ego. Richard Stallman [the author of emacs] and Project GNU [the main source of the software behind Linux] are the ultimate examples of the power of this [open source] tradition.

So when I started writing for my friends or to clarify my thoughts on a subject, it seemed natural to distribute it over the Internet as I had done with my source code.

The Tech: Despite having written and published books, you claim that you are not a writer. Why?

Greenspun: A “writer” is someone who spends time looking for a publisher (I spend time thinking of creative ways to hide from my publisher.). Someone who sets down on paper what he happens to be thinking about at the time is just a person.

The Tech: Conventional wisdom says making Web pages is a hobby of those who are either too nerdy, have no life, or “just have too much time on their hands.” One reader of photo.net posted a conclusion that “Greenspun [must be] independently wealthy” to have spent so much time putting up a site that does not even generate a profit. How do people who wish to build Web sites justify the heavy time/monetary investments without clear returns?

Greenspun: Americans these days assume that one ought to be striving to make money 24 hours per day, 7 days per week, 365 days per year. We are fortunate to live in a rich society; we don’t have to struggle for our crust of bread every waking hour. Could I make more money as a bond salesman for Goldman Sachs? Probably. But why worry about that if I can make enough to live comfortably doing the work that I love: building Web software and giving it away?

The Tech: Why don’t you advertise photo.net so more people can join the community?

Greenspun: Michael Dertouzos, Director of the Laboratory for Computer Science, has a great expression when talking about an activity that takes time: “It takes more than money, it takes life.” If photo.net had twice as many users, I’d get twice as many email messages per day from users. The goal of photo.net is not to dominate the Internet photography scene and crush competing sites. It is to have a community where people help each other become better photographers. I’d rather another publisher downloaded our open-source community site toolkit and started another photography site.

The Tech: You publish even the directions, not to mention the phone numbers, to your house on your home page. Most people worry about revealing too much personal information over the Internet.

Greenspun: [Publishing this information] just reflects the practical reality that my phone number and address are listed with Bell Atlantic. So I’m not going to inconvenience my friends who might want to mail me a package in exchange for some illusory privacy. Probably about 1 percent of my unsolicited phone calls are from readers of my Web site. Surprisingly enough, people who find me on the Internet seem to send e-mail instead. Go figure.

Views about Internet commerce

The Tech: What are your thoughts about the current craze in Internet e-commerce? Seems like everything people are doing on the Internet these days is intended to make money. Is this phenomenon of commercialism: Inevitable? Pathetic? Healthy?

Greenspun: E-commerce as in “distribute a catalog of stuff” isn’t so interesting to me. I think the good e-commerce sites have yet to be built. A good clothing site knows what you’ve got in your closet, what fits, what looks good on you, and what’s about to become worn out. A good shopping site knows what brand of vacuum cleaner you own and what bags you might need, it knows that you’ve got a DVD player and not a VCR, it knows that your dishwasher is about to run out of that rinse-aid stuff that comes with dishwashers.

I’m not too concerned that e-commerce will drown out interesting Web applications. Most people don’t spend all day shopping or trading stocks.

I do laugh sometimes when I hear Internet entrepreneurs talk about how great their company is going to be. Then I turn around and find that they’ve sold out for $100 million or $1.5 billion. So I don’t laugh so much anymore.

You don’t need to be smart to make money these days.

The Tech: What do you think of the MIT homepage at web.mit.edu?

Greenspun: web.mit.edu is as good as it can be given that it is static .html files. That said, web.mit.edu isn’t nearly as good as it needs to be. A large organization like MIT needs a dynamically generated site. Suppose the server knows that you’re a high school kid; it should highlight stuff about admission and science tutorials appropriate for people without university backgrounds. If the server knows that you’re on campus, it can point you to talks that are happening today. If the server knows that you’re a biology professor at Stanford, it should greet you with a list of the most recent publications from Course 7 professors.

Note that this isn’t just a technical challenge. To make web.mit.edu really work for users, we’d need to define publishing standards for groups within MIT. For example, a researcher here needs to be able to tag something “of interest to high school students.”

The Tech: I’m curious about what you think is a potential solution to “dead links” on the web. Not by demanding perpetual links, I suppose, since it has a scalability problem and is not practical to expect of the zillions of web publishers.

Greenspun: Actually search engines like AltaVista are already 99% of the way there. They’ve got a database of content. If you couldn’t find the link live, you ought to be able to ask AltaVista “show me http://foobar.edu/yow.html as it existed on June 1, 1998”. Someone has proposed the death penalty for those who create dead links. Probably a combination of these two would be a good solution.

The Tech: Tell us about your company, ArsDigita, and what you envision to do with it. If the ArsDigita Community Software is open-source, how does the company make profit?

Greenspun: Computers don’t solve interesting problems out of the box. In fact, a freshly unboxed computer creates problems. The computer has made more consulting firms rich than any other technology. ArsDigita builds and operates online communities. We might lose some revenue by giving away our toolkit for free, but we’re too busy doubling every year to notice. The goal of ArsDigita is to solve the problem of Web-based collaboration, to distribute an open-source solution (so we give away our software), and to distribute knowledge about how we built the solution (so we give away my books for free). In doing this, we collect some of the best programmers in the world, pay them as well as McKinsey pays its consultants, and feel good because we don’t have to live in hotels the way management consultants do.

Our typical customer has between $7 billion and $100 billion in revenue. They want a site up and running in six weeks. Although they could download our toolkit for free and operate the site themselves, they’d rather pay us $1 million per year to operate our 200th database-backed Web site than risk running their first. ArsDigita frees the company to focus on their problem at a price that, to them, is insignificant.

Open source probably makes it easier to sell these $1 million jobs, actually. Big companies don’t like to depend on proprietary closed-source software unless it comes from another big company.