A couple years ago, I did a long interview with Paul Hawken about his then-new company, Groxis (now Grokker). For a while I thought about doing an article, or even a short book, on zooming browsers and the next generation of computers interfaces. For various reasons I didn’t, and the interview itself didn’t get circulated.

Recently, though, Paul got in touch and suggested that I should publish the interview. The transcript is published after the jump.

Technorati Tags: , , , ,


In late 2001, my phone rang. “Hi, this is Paul Hawken calling,” the voice said. Paul Hawken? It took half a second to place the name; then another second to wonder why in the world someone as important as Hawken—founder of Smith and Hawken, author of best-selling books on ecologically sound business—would be calling me. It seemed as likely as my wife holding out the phone and saying, “It’s Sting. He wants to talk to you.” Paul had recently started a software company, and one of the members of the board—Paul Saffo, my colleague at the Institute for the Future—suggested that I would be interested in what they were doing. So, now Paul Hawken was calling. His company had been working on a knowledge visualization tool, he explained, and he was it showing to a few people. Would I be free to come see it some time? (“Peter Gabriel and I have a few new songs, and David Geffen gave me your number. Care to drop by the studio?”)

Naturally I said yes. A few weeks later, armed with my copies of Ecology of Commerce and Growing a Business, I headed up to Sausalito. I didn’t really have a good sense of what I was getting into, or what this software was supposed to do. But what Paul showed me blew my mind. Paul’s little company, Groxis, was working on an interface that was so elegant, so beautiful, watching it was almost a spiritual experience. I had never seen anything like it, yet it resonated on a deep level, like a great piece of art. As someone who had worked with gigantic data-sets in the past (I was editor of the Encyclopaedia Britannica during its transition from print to electronic publishing), and was semi-obsessed with charting the history and future of computing, I was thinking of uses for the program after about ten minutes. After half an hour, I was sure that they were onto something amazing. At the end of the morning, I knew I had to write about the company. Writing the history of a product that hadn’t even been finished may seem odd, but as a futurist, I work all the time on things that haven’t even happened yet. Compared to that, a history of an unfinished program is pretty grounded.

It took some time to schedule the interview: I had a new child and a demanding job, and Paul’s efforts to make history tended to give him little time to talk about that history. However, we were finally able to sit down on 23 July 2002, in Paul’s office in Sausalito, for an interview about the evolution of Groxis; the philosophy and thinking behind it; and the connection between Groxis and Paul’s earlier work on sustainable development and gardening.

This transcript is the result. It was transcribed and edited by me, from a cassette tape recording of our conversation. The published version was reviewed by Paul Hawken.

And Sting, I’m free whenever. Just let me know.

Paul Hawken interview, 23 July 2002

[Tape 1]

Background to Groxis

Pang: The last time we met in January, you told me the story of Metacode, and the evolution of Groxis. I thought we might start by reviewing that story.

Hawken: Metacode started as Datafusion, and it really started in Joel Schatz’s head. He came to me with a big map that he and his wife Diane had done, a map of all systems living and human, with the basic typologies and how they connect. He laid that map on the floor, and I was blown away—it seemed so obvious. He was talking about creating a company that, in a sense, created a knowledge map authoring capability for any system. We started working together, and he brought Chris Rathe to form the business, and we started to raise money. But that’s another story.

What happened with Metacode, which was its eventual name, was that we did create a knowledge map authoring tool. We worked with a famous graphic designer from Berlin, Erik Spiekermann of Metadesign, whose famous quote is, “You can’t not communicate.” Every single thing you do is communication, so you’d better think about it. And especially with a UI, no matter what you do, you’re always saying something, you’re always communicating. There are no idle spaces, no idle thoughts, no idle words. So we tried to create an interface that was clean, and powerful in terms of allowing people to see or envision relationships between stocks, inputs, outputs, nodes, whatever entities the map was about. Version 1.0 of that authoring suite or tool was released in 1998… and it didn’t really amount to anything. Nobody wanted to use it. It was elegant but slow with a complex back end in terms of content tagging. It was an ambitious effort, and I think like some things, it was ahead of its time.

So, we did a restart. We had created a back-end content management tool, under the name Metatagger, and that became the core business. It was successful, and it and the company later went on to be sold to Interwoven. In the midst of that, our software engineer Jean-Michel Decombe started playing around with the front end. The front end was turgid; it was not particularly responsive to keystrokes and cursors. Jean-Michel started thinking about how you parse HTML and XML and other links in a way that preserves the metadata, but puts it in graph form that allows the information to be manipulated, distributed, reconfigured, mapped, envisioned, visualized, presented—allowing the user to do it in a quick, agile way. It’s one thing to have cool graphics—a video game can have great graphics—but in this case the graphics really mean something, they convey a lot of information. Jean-Michel came up with a GXML format (a Graphical Extensible Markup Format) that we later patented—it’s really a grammar, not so much a language—with which he started to map things like Yahoo!

We started to see the possibilities, and at that time I convinced my board to spin this off as a stand-alone company called Groxis. Jean-Michel was the head of it, and it was it a Palo Alto garage on Homer Avenue—I mean, it was just classic—it was near where he lived, and why not walk to work? So he started banging away there.

Pang: Where did the name “Groxis” come from?

Hawken: Jean-Michel came up with it, from “grokking,” in Stranger in a Strange Land.

Anyway, that was at the time when we were doing a Series D round for Metacode, and it was a big round, but it was a pretty harrowing term sheet. It quickly became evident that Groxis wouldn’t survive the D round without massive dilution: the investors would have control, and while we’d have money, the company would essentially have had a new owner. And so, at my behest, the board voted to spin off Groxis. But right after that we received a buyout offer, and the board decided not to spin it off after all. It seemed importune, if not inappropriate, to be spinning off a new company, even though the board had approved it, and even if it was only a little skunk works funded with $150,000. So Groxis was acquired—at least the part that was owned by Metacode—was acquired by Interwoven at the end of November 2000.

Then, it just sort of got lost in the shuffle. Interwoven’s stock was on the way up, it had a huge market cap, and it was busy acquiring companies, and there was this little thing buried over here that no one knew about. It took us three, four months to get their attention, I think? By that time, Groxis had run out of money, though it only had one person on the payroll—that was Jean-Michel—he had spent money on lawyers, and payroll, equipment, and stuff like that. Basically, we said we wanted to buy it back, for the money that was on the balance sheet, the $150,000. They said yes, and they were very nice about it. It really had no value to them. Almost all the value was locked up in Jean-Michel; we had a vision of the company, and we had some trademarks and stuff, but that was it. They were quite right to sell it, because if they hadn’t we would have just turned around and started a new company up anyway. That was in April 2001. In July R.J. Pittman joined as CEO, and the rest is history.

Seeing Information: The Motivation Behind Groxis

But all along, the thread that runs through all this is not so much the narrative of the ins and outs of companies, but a desire to create a powerful, visual way to display complex arrays of data. We’ve seen, as you have and many people have, everything that’s out there—there’s some great stuff out there, but none of it really has been taken up in a useful way in the consumer market. There’s programs that NSF has, and some programs done at Sandia National Labs, complex, extraordinary things—what Wired calls Infoporn, using Nvidia chips and graphics cards—and they’re creating these nebulae of data. It’s stunning, and it may be good for the NSA. But the normal user—student, professional— needs something else, and we’ve always aimed at having something that’s useful but not in a dumbed-down way.

We’ve always been about changing the way people see things. The way you see information has a huge impact on what you do with that information, and how it affects your thinking. The way computation processes are presented today on all our laptops and desktops has understandable roots. It comes from a time when bandwidth and memory were expensive, and coders were forced to come up with clear, alphanumeric ways to encode and present data, and that continues to this day. You start to see people peeling away from that in terms of OS X, and other apps, but these are still gestures: making something aqua, or making something show shading on the desktop. Sure, it’s visual, and it’s a way to distinguish something on a desktop, but it’s just teasing around the edges. No company has successfully looked at the way we handle and manipulate and process information, and reconfigured the interface in a way that makes sense. That’s what we’re aiming at.

It goes back, really, to the thing that computers do, which is at the heart and root of our problems: what they do is linear, disparate, inchoate, and disconnected. The whole industrial system is riddled with that kind of thinking—indeed, it’s the result of that kind of thinking—where every solution creates ten more unintended problems. People are not taught, nor do they normatively get information in such a way that they can see the obvious connections between what they do and other sectors and activities, and the impacts of their actions—whether they be social impacts, environmental impacts, financial or economic. Lacking that perspective, we continue—on a governmental, institutional, and even individual level—to continue to create activities and modalities of development that are well-intentioned, sometimes too expensive, but always seem to have consequences that we didn’t anticipate and always something we have to create new solutions for. And so in a small way, but hopefully in a meaningful way, Groxis is about presenting computational processing in a way where you have to say, “Well, I don’t want to see that connection,” instead of not even being able to see it at all. You have to turn it off, in your brain, in your mind, on the screen: you have to make the choice and say, “I don’t care that what I do affects the world,” but at least you know that you made the decision. That’s different from thinking that your decisions have no effect whatsoever. Of course they do.

So what Groxis and GXML and Grokkers are all about is inserting into the quotidian life of computers a way of seeing the world that’s new. It doesn’t mimic what you see outside—you can’t do that, and besides you can always just walk outside and look at the world, and there’s no reason to mimic that. All computational processes are symbolic, they’re all representational, and they don’t mean anything on their own. It’s one big illusion that you can play with, but it’s an illusion that in turn is determining how we see the world. We want to break that. We say, “Break the page barrier.” Information is scattered around in lots of different places—on your computer, on your intranet, on the Web—and we think that for you to be able to go only backwards and forwards, and disappear into the maw of the browser, is outmoded.

Groxis versus the desktop

Pang: Jef Raskin argues in his book The Humane Interface that a zooming paradigm is the obvious replacement for the desktop, which he and others have contended is an increasingly outmoded, antiquated way of interacting with computers. That metaphor is already struggling, and is going to become even more antiquated as we do more with mobile devices, and move into an age of ubiquitous computing, or pervasive computing. Do you want Groxis to be seen as challenging the primacy of the desktop?

Hawken: Well, Grokker maps a desktop, and it does so well, and quickly, and you can keep it mapped in a dynamic way. It has clear applications in terms of remote applications—that is to say, Blackberries and Palms—and it will give you the fastest, most fluid, most intuitive access to your desktop. That we know. We also map the desktop because you can create your own maps, and so information that is outside the desktop, on a company intranet, or on the Internet, oftentimes all belongs in the same map, even though it just happens to be in a different place. Groxis allows you to configure and put that information together however you want it. Groxis also now can make Web sites, so you can have a Groxis map of a 100,000 page Web site, and you can rearrange it by dragging and dropping in the map. So it allows you a lot of fluidity to see and handle information.

But Groxis is not a displacing technology. We don’t intend to replace anything that’s out there now; we’re trying to make everything that’s out there more useful. We’re not saying, “This is not useful, but Groxis is useful;” it’s not comparative or competitive at all. We want to take that inchoate mound or swarm of information that’s swirling inside and outside our desktops, and give users the ability to collate, publish, update, present, offer, or link it to other coalescences of information that other people have created using Groxis.

It’s almost as if the world wide web were a population of six billion people, but there were no cities and towns. And so, when you want to go somewhere, you can certainly go, but you can’t fly to “Chicago.” There are no hubs. If you want to see your friend at Amherst, you just go, but there’s not even an Amherst, because that’s almost a hub, too; but he or she is studying there, and you go across the country, and try to find them, and you can probably do it.

What we’re talking about with Groxis is the ability to create these living hubs of information, to take the 2-3 billion pages that are on the Web, plus the billions that aren’t on the Web but are accessible through the Web—that are in newspapers or libraries or government databases, but aren’t directly searchable—and gather those under different subjects, like “photonics,” or “onset diabetes Type II,” so that a person going to the Web and running a search will find a site that has a Groxis map, where a group or a person has said, “This is it, these are the 20,000 URLs and sites out there that are current—or historically important as well—and will inform anyone who’s serious about this subject,” and lets you go as deep or as broad as you want: You can skate around the surface if you want, but if you’re diagnosing a disease, or if you have it, or if you’re researching or writing a paper, or if you just cannot get enough of Chaucer in your life, whatever—no matter what it is, there will be somebody using Groxis who has said, “This is it, this is the state of the art.” Now, that’s in terms of what’s linkable. If it’s a book and it’s not digitized, all you can do is say, “There’s the book,” and you can link to the fact that the book exists. But if it’s linkable, and even if it exists in a database like New York Times or Lexis-Nexis for which you have to be a subscriber, you can still show the link, and in effect say, “You can subscribe or not if you want, but I’ve made the map, and I’m telling you, that’s were a great article is.”

So what Groxis does is take information gathering and searching and raise it to another level. We are not saying existing resources aren’t good, they are great. We’re not making the past wrong, we’re simply trying to make the future better.

Groxis and mental maps

Pang: That’s a great line. One of the interesting things here is that people talk about how “the map is not the territory,” but in this case it really is: the Groxis map is a kind of representation, but you talk about it as if it’s—

Hawken: It’s a mental map.

Pang: —also a thing.

Hawken: It’s a kind of mental map. If you talk with our advisory board member Peter Senge about mental maps, he will say that we have lots of mental maps, we just don’t recognize them. Yet they determine how we see, perceive, read, understand, hear, and act. And if we don’t understand our mental frameworks and mental maps, we’re not really conscious, reflective people. If we understand those better, we can understand other people’s mental maps. Groxis can be a type of mental map. It’s not the only way to represent a mental map: there are lots of ways to frame things. Some of them are emotional or psychological, and they cannot be mapped graphically.

But in the case of Groxis, they are maps of a topic, or a subject, a person, a place. If you’re looking for something specific, search engines work well, and they’re getting even better as time goes on. If you go out to the Web, and you’re looking for AlexPang.com, and you go to Google, you get your result quickly and it’s fantastic. But if you’re researching, and need to know more about Alex Pang, and you don’t want his Web site only because you want to know what the rest of the world knows about him. In other words, when you start to do research—conventional search engines can break down. There’s only a couple ways, mostly relevance ratings, to rank the order in which search results appear. But to me the methodlogy not entirely relevant, and it can be a positive feedback loop because the more hits a site gets the more “relevant” it becomes. In Google, for example, if you do a search on myself, you get 14,000 hits, but Google actually only displays 671, it only goes down that far and won’t go any further. So you have to think, what would be the Boolean trick to getting the other 13,400? That string might only get you 100, or 200. You wouldn’t actually know what it would be that would get you the other 13,000.

But with Groxis, you could theoretically pull all 14,000 up, and then you could decide what’s relevant and what’s irrelevant, by applying filters to help you do exactly that. Furthermore, you can annotate and comment on the maps. The maps can be syllabi of university courses, that over time change and matriculate, where people can annotate links to papers or monographs, and teaching assistants, students and professors can actually have an archival record of what a researcher thought of a paper. Things change over time in terms of how we see data, events, people, and history.

So there are so many possibilities for these maps. They’re living documents; that is to say, they’re not frozen in time, they can be automatically updated, but the fact is, instead of the 3 billion disparate pages whose only real access is a search engine, which is going to be a semi-random walk, you’re talking about the nodes in the system—in this case being the individuals and institutions that have taken ownership of a topic—starting to create real information hubs in the World Wide Web.

I also imagine these maps being linked to each other by wormholes, so when you go to a map on Chaucer you can get to a map on Ivanhoe, you go to Lady Murasaki’s Pillow Book, which was the world’s first novel but wasn’t written in England but in Japan, and by a woman, not by a guy. When you start to do that, we still individually can toss things into the World Wide Web, but we can count on people gathering this up, just like in nature, like detritivores, picking up all litter that falls onto the ground of the web, and digesting them, reconfiguring them, placing them, and repurposing them.

In a sense, for the first time people may start to think of the world in terms of categories and schema and ontology and taxonomy, to thing of date and information in terms of relationships. And you think, well, what’s a relationship? And what trophic level do these belong on? Is this a subset, or is it even hierarchical at all? And if it’s heterarchical, does it belong in multiple nodes, or containers, or something else? How do we draw the lines between these different topics? Do we want to show which of these sites is a hub, and which of these is what we call a satellite site, in the worlds of Chaucer or Japanese literature or any other subject?

So what happens is that as this spreads and goes throughout the world—if in fact it does—the ability for us to be able to inform ourselves will grow, especially as libraries are digitized over the next ten years. There are projects that are talking about digitizing a million books a year.

Target market and use

Pang: To what degree are you aiming Groxis at business users and institutions of other kinds, versus individuals? I can imagine a company using it as a front end to their site, but that’s different from a person who’s an enthusiast on some subject who uses Groxis for themselves. Do you have to make choices at this point between going after the one market versus the other?

Hawken: We have an enterprise and a client product. The specs on each are different. One’s on a server, and obviously is more powerful and robust. The client is downloadable to the desktop and at present can map up to 100,000 documents or URLs. Going back to the earlier question about whether we see this as replacing the desktop: it is not designed that way, but we think the desktop application is fascinating, compelling and interesting, and we really don’t know how it will ultimately play itself out.

Though there’s another word I’m looking for here, which is a kind of permeability or openness, where we’re comfortable with the fact that we don’t actually know at this point all the ways it could be used. We know what we’re doing, because we can talk about it, and tell you what we’re doing, but I think part of us is always stepping back and looking in, and saying, “We’re not really sure how Groxis is going to play out.”

We have lots of ideas. One is that it is being taken seriously as an e-commerce platform, because you can have the whole product catalog, whether it’s b2b or b2c, on the right-hand side of the screen, and you can navigate and browse infinitely, and you never have to change the page. The left side keeps changing depending on what you select or click on. You’re never lost, because you’re not going back and forth between pages; you’re essentially on one page. You can accumulate things, in a checkout or shopping cart way, and then order and be done with it.

Another function is what we call “Who knows what?” People spend an extraordinary amount of time in big companies—Sun, HP, big law firms, GE, BP—trying to find out who knows what. What you can do with Groxis is have a map of every person in the company by geography, or by function, or expertise; and you can switch views, or combine them: you can say, I want to see everyone in Australia, then everyone by function in Australia. But it doesn’t really matter, because the map isn’t actually so much what you use, you actually navigate the opposite side of the screen, on the left side, and you use the filters. By the way, we now have dynamic filtering, so the filters are actually determined by the metadata itself.

Pang: That’s very cool.

Hawken: It is cool. For example, we search Amazon.com for books on information technology, and then we have filters for customer ratings, list price, etc., because that’s the metadata that’s available from Amazon. We didn’t assign that filter beforehand; the filtering now is dynamic.

Going back to Who Knows What, suppose you are an agency or company and you have a nuclear spill in Peru in a violently contested geographical area.. You need a Spanish-speaking person, who is an expert in conflict mediation, who is comfortable at higher altitudes in the mountains., You can just hit those keywords, and that’s all metadata: either HR has forms that people filled out about their education, language, background, vocations, whatever, and then maybe added some free text about “This is what I do.” So you fill out the left side, set the filters, and the only thing that’s left on the map are the people who fit your criteria: you can say, “Well, I’ve got one in Sydney, and one in Bonn, and one in Montreal.” Then, you click on the node, and on the left side of the screen, you get his or her HR record—their photograph, the bio, their contact info. And you call up Sydney, or Bonn, or Montreal, and say, “We’ve got a situation in Peru, can you help?”

Same for law: this court, this antitrust section of the law, this judge. You may find someone, or no one. Either way it is good to know. Either way, I know within seconds. And this can be live, in the sense that your data may be constantly updated. It tells you who knows who, and who knows what.

In terms of diagnostics, you could map every disease in the world. There’s a category that has metadata. Symptoms, contraindications, the whole nine yards, it’s all tagged: almost every single word you see in Merck’s Index is metadata. So you can just query the map, for example type in, intense frontal lobe headaches, high fever, chills, etc. and it’ll eliminate every disease that doesn’t conform to those categories. You might see right awaythat it could be encephalitis or malaria. Somebody with a wireless Palm Pilot out in the bush can hit the map with a stylus, and quickly aid the diagnostics and treatment.

These are all institutional applications. Institutions need to build up their knowledge base, they need a way for new employees to get up to speed. If you have Groxis maps, and you’re updating them and archiving them, then people can say, “I want to see this unusual thing,” or “I want to see what was happening five years ago,” or “I want to look within this 30-day period,” because it was a critical period in terms of data. We know we’ve got a haystack, we think there could be a needle there. With Groxis, you can say, “Here’s the needle,” or say, “Actually, there’s lots of needles in there.” You need the brass one, not the gold one? You refine the search.

You always have tools right there: it’s not like you’re saying “Search,” and choosing a database, and entering this code word, and getting a list of results. There’s no pattern to that, and there’s no way to recognize the patterns except to say, “Okay, I got them, now I’ll read them.” There’s no dynamic filtering. But with dynamic filtering, you can start to see patterns and relationships. That’s what Groxis can do.

Having said all that, it’s safe to say that we don’t really know how people are going to use Groxis for.

Pang: The medical diagnostic example is one that depends on a lot of metadata, use of XML. But what about knowledge that isn’t tagged? One former head of H-P once said, “If only H-P knew what H-P knows,” suggesting that groups of people possess a kind of collective knowledge that none of the individuals possess. On the Groxis site you talk about this as a collaborative tool. Is this a kind of supercharged Lotus Notes, or are there things you can do with Groxis that you can’t otherwise?

Hawken: We’ve had people from different professions come in, and I’ve sat them down in front of Groxis, and they’ve mapped an area of interest, and created five or ten thousand-node maps. Rather than me show it, I’ve sat on the side, and just watched and listened. And so far, without exception, everyone has discovered things that surprised them, things that they didn’t know. Now, everything they’ve discovered has always been available on the Web. It’s not like we had any new material. What we had was a way for them to browse, to see and look quickly at a lot of different information, and to browse in a way that allowed them, in an interesting and compelling way, to discover. Again and again, people would say, “Gee, look at this,” or “I had no idea about this,” but then, how could they, given the way the Web is organized and search results are normally presented.

So I don’t know how HP would use Groxis to know what HP knows. I do think that the best use of it is going to come not from institutions deciding to adopt it: I don’t think IT departments are going to run out and buy 5,000 seats of Groxis. But somebody is going to buy a seat for themselves, get the reader, and start saying, “Hey, I’ve done this thing, check it out,” and send a map around to their friends. And their friends are going to say, “Hey, that was really useful, how did you do it? Shall we use this as a collaborative tool on our project?” “Okay, let’s use it.”

So I think it’s going to be a ground-up phenomenon, not top-down. That’s another reason a lot of knowledge management has failed, because it’s top-down. The only way people will use a tool is if it adds value, and is compelling and interesting. If it doesn’t add value and isn’t compelling, people aren’t going to use it. That’s been the whole history of knowledge management in institutional use.

Other visualization tools

Pang: There been things like Perspecta, fish-eye views for Web sites, and any number of other visualization tools in the last few years. A lot of these have come out of Xerox PARC or the MIT Media Lab, and here you are, six people in a garage beside the water. The greatest minds in computer science have been working on this stuff for years. What is it that you see in Groxis—in the way that it’s designed, in the visual tools—that you think will make it more useful than these predecessors?

Hawken: The existing sorts of ways to envision or graphically portray information usually animate something that we have done manually. You get the flow chart animation, or the hyperbolic browser, which is Ramana Rao’s great interest. I think they’re all useful. But they don’t scale in the same way, they aren’t dynamically filtered, or have the other features that we have created in Groxis.

One of the problems we all deal with is the ubiquity of information. Ubiquity creates traits that starts to determine our patterns and habits. Turning your computer on is like threading yourself through a crowd at party. If if you are convivial and chat along the way you won’t get anywhere, and you’ll be there for hours. If you avail yourself of too much of the information that comes to you through the Internet, or even through corporate intranets, what happens is that you get inundated, and your brain feels like mush because it’s just too much unfiltered data. Essentially, what some of the visualizers do is mimic a meeting-with-whiteboard session, and then added some scrolling, so you can open up a link below.

But I am not sure what I have seen is what I would call intuitive, in the sense of relating to how we think. Many are designed around the way we have learned to conduct ourselves with information in a meeting with a magic marker. But the way we think is contextual and visual at the same time. The best example I can give is driving, which is highly contextual, and highly visual. It’s amazing how many people of different intelligences—I don’t mean in the IQ sense, but in the Gardner sense, people who have many different types of intelligence—different levels of hand-eye coordination, different somatic intelligence, operating in different vehicles configured in different ways, can do so much driving with so few accidents.

What is it about the brain that allows people to do that? While you drive, you can see what’s in front and behind and to the side of you. You can look at the gauges and stay within the limits of the law. You notice the effects of gravity as you go uphill and downhill, go sideways, and make turns. You can listen to NPR, you can talk on a cell phone, you can think about your meeting, you can drink from a latte, you can talk to the person next to you, you can daydream. Obviously something is going on there that is untapped by a desktop. If somebody were to calculate the number of computations you have to do to, it’s just stunning. But from an experienced driver’s point of view, it’s a simple thing. It’s a visual thing, it’s a context thing. The brain has a tremendous ability to make decisions among variables instantly and quickly, and you have a sense of place that provides context. You latte is there in the cup holder, the road is there, the steering wheel is there.

Maybe the most noticeable shortcoming of some visual maps is that is that they don’t scale. If you look at a hyperbolic tree, you see 50 or 100 entries at the same time, so when you warp around—warp meaning shift the center of focus—you’re in a new place, but where are you? And what happened to the things that were just there? The screen is flat, it’s not a circle, it’s not an orb. You don’t want to lose your context. If you look at your latte, you want to be able to look right back up at the road. So without context, without scale, maps can unintentionally create just another way for people to get overwhelmed, because there’s too many things to pay attention to. It’s like if you were driving and you had to consciously think to yourself, “Take right hand off the wheel, put hand on latte, make sure latte is not too hot, keep left hand on wheel,” you’d be dead. If you have to read too much to know where you are and what to do, then you’re right back at in front of a search engine screen: it’s a different shape, and a different relationship, but you’re right back there. But if you can actually stay within the same visual and contextual realm, and then you choose when you drill down, when to open up a site or go back up or open a spreadsheet, and then go back up—you always have a choice, you always know where you are.

The “intuitive” interface

We can go to one hundred thousand links on one page, and users feel comfortable, they don’t feel overwhelmed at all, in fact they feel happy—”There it is, I’ve got it all right there, thank you.” I’ve got four million bookmarks on my desktop using Groxis, and when I’ve got a hundred on my browser, I feel really lousyI’ve saved the site, and I know it’s there, but that won’t do me any good. No one wants to be a slave to their browser Whereas my four million bookmarks already have twelve thousand folders—probably more, actually, there are folders in folders in folders in folders, until you get down to just an alpha list—and I can go right to “Sustainable Forestry,” or “Urban Forestry,” “Clear Cut,” just click-click-click, and I can get to any other of the one thousand categories, and subcategories, because I have this encyclopedia on my desktop.

Pang: You’ve spoken of Groxis being intuitive—

Hawken: I should clarify that. We’ve never said that about ourselves. That’s what people say who’ve used it. We never set out to make something that was intuitive, because we never knew if it was intuitive for us, whether it would be intuitive for somebody else. But what we’re getting from people, remarkably, over and over again, is that they say Groxis is intuitive.

Pang: Making something intuitive is the holy grail of UI design, but when most people say “intuitive,” what they mean is “this looks a lot like something I’ve already seen.” And Groxis isn’t. But maybe what they’re responding to is that while it looks really really different, it uses cognitive abilities that we employ constantly without having to think about them.

Hawken: We’re tapping into something innate, we think. We think there’s something intrinsic about it. And we’re fascinated with pursuing that. Some things in the beta moved away from basic functionality, and not everyone agreed that we should pursue that course.

As we add more functionality there’s always the danger of getting away from that—Grokker Pro, at the high end, can really scream, it’s got functionalities that are really complex, like multiple root-node generation, and how figuring out how to provide that power while keeping it intuitive is a big challenge.

The Groxis team

Pang: Can you tell me some more about the division of labor between you, Jean-Michel, and RJ?

Hawken: Well, Jean-Michel is the technical genius, the in-house genius. In the early days, I think my function was more about use case and metadesign and function. I represented the user, Jean-Michel represented the creator. When RJ joined, we would work together in a generative way. RJ is the CEO, the person who makes sure that we make a decision, and besides being the CEO, he is also VP of engineering. Jean-Michel and I can go back and forth, but RJ reminds us that we have a schedule. Of course I’m a businessman, too, but I try to focus more on use cases. On the other hand Jean-Michel has what he calls the idea book, it’s about an inch thick, and it contains the future of the company.. All three of us will talk about colors, about whether it should be anti-aliased, about many things—we really drill down.

Details, details

Pang: Designers will end up spending a lot of time on little details that most people won’t notice, but which can have a big effect on usability. Are there experiences like that that you’ve had with the development of Grokker?

Hawken: Absolutely. We think its all details. Some are more important than others, but…. The details have to do with the fact that in order for it to pass the “grandmother test,” it has to be not only be intuitive, it has to be something you can readily use without prior information, knowledge, skills, or training. Anybody should be able to sit down in front of Grokker, take hold of the mouse, and whether they’re six or 96, should be able to make it go. And if they don’t, we’ve done something wrong.

Another part of that involves not details but elimination, because anything that’s superfluous or extraneous is going to be confusing. People have said that the interface is beautiful—that comes out a lot. We didn’t set out to make something beautiful; we didn’t set out to make something ugly, either, but I’m just saying that it wasn’t in the mission statement. And we actually believe it’s beautiful too. And I think that part of its beauty is that there’s nothing extra. There’s nothing you can take away that doesn’t remove functionality. When you see iTunes now, it’s neat to see that it looks like brushed aluminum, and the equalizer looks like the dashboard of a real equalizer. It’s all pretty cool, but from a Groxis point of view, it’s like printing brick on linoleum. We’re working in a space where every bit has a purpose, and if it doesn’t, it goes out.

Now first time they see a triangle, a user may not know what it means, as opposed to a square or circle or pentagon, but it does mean something. We’re always asking, “What is it that we want to differentiate or distinguish?” Because it’s always about making differences, and the more quickly and easily people can perceive or create distinctions, the happier we are.

This is something fundamental. All of education, every book, every novel, every scientific paper is about distinctions. And we always ask, How do you do that in a way that doesn’t make people feel overwhelmed? It’s like going into nature. You can take a walk through the forest, and the more you can make distinctions—the more trees you can identify, the more birds or plants you understand—the more you enjoy it. But you will not enjoy that forest in the same way if somebody says, “Here’s a list of every species of animal and tree and plant.” Same with Groxis. If you can just go there, and open it up, and browse, and walk though the forest of a topic, you’ll enjoy yourself. The longer you stay, the more distinctions you want to pick up, great. The point is, you’re in control.

Pang: Two of the things that really stuck me when I saw this first were the colors, which are dazzling, and second, the movement between levels, the transitional motion as you move up or down in the map. Did it take time to figure out how fast you should zoom down, and whether it was important that the movement be smooth versus instantaneous?

Hawken: Yes. What happened is that in the prototype, we tried everything from zero to 20 frames per second—a range from no animation, an instant transition, to a very slow movement. We found independently that around 8 or 9 frames per second made sense. But if you want to set it so there’s no animation, you can do it, but then you lose some context. You actually leapfrog the visual-contextual modality. Now, as the maps get bigger and more complex, and as you move from just hierarchical mapping to relational mapping—that’s not something on the board [points to board listing features] but is for next year—it’s even more important to have cues that tell you where you are, because you begin to get into a two and half D situation. In fact we’re already doing some stuff where depth is a factor; in the current version you can zoom down, but you don’t get quite the same animation will be more important as you start to get into these more complex spaces.

Groxis, ecology, and sustainability

Pang: When I was reading Growing a Business, I was struck by your description of the founding of Smith and Hawken, and your line that “Tools have always been an interest in my life.” I had an epiphany: “Groxis is Paul Hawken’s second tool company!” It’s another kind of tool–

Hawken: It’s definitely a tool.

Pang: –though it’s a very different kind. How do you draw the connections?

Hawken: Well, I do a lot of work in the area of sustainability: I write about how absurdly we treat the earth. Our resources systems are so badly designed and comported. But so what? It’s so easy to say that, cast stones, lay blame and point fingers. That’s edifying for about ten seconds. But then you have to go back and say, “Why are we doing this, we’re all human beings?”

And that gets you back to mental map and framing issues: how do you see the world. You have to assume that most people are trying to do the right thing. I don’t assume that humanity is stupid or bad, not at all; I see quite the opposite. And yet, our collective activity is pretty damaging and sobering, And so with respect to sustainability—which is really in the broadest sense of the term about ecological and social justice, not just recycling—the only difference I can detect between those who proffer sustainability and those who ignore or think it meaningless, is that the former had at some point a shift where they began to see the world systemically, to see the world as a system. It was like falling through the looking glass.

Now, what they do with that knowledge can vary, depending on what their background is, or their education, or whatever; but once somebody has that view, you can’t go back. You can never go back and say, “Well, it really is a linear, mechanistic, Cartesian world.” Once you can see the relationships, you’re stuck. Because that’s reality. We know that from physics, and the biological sciences. Then you ask, “What are we going to do about it?” The Bushes and Cheneys and MBAs,the way they see the world is a result of their education. They do not see those connections. You can be a medical doctor and not study nutrition; You can be a CEO of an oil company and not know a whit about climate, and have no responsibility for learning about it.

So Groxis is really about trying to introduce tools to people that increasingly will allow them to see the systemic connections. Now, what they do with that is up to them. It’s not a hammer to hit someone with, it is a tool to allow someone to see better. Everything I’ve done in my life has been about trying to change the way people see the world. It’s not about tools; it’s always been about seeing the world. Because once you see it differently, you act differently; you don’t draw the same conclusions, you can’t act in the same way. Groxis is definitely about seeing; it’s more like an offer, we’re offering the world a way of seeing itself. That’s the goal.

Pang: People who work on interface design sometimes talk about ecologies of knowledge; Bonnie Nardi and Vicki O’Day wrote a book with that title. I was wondering if that was a metaphor that you had spent time thinking about? Would it be reasonable to describe you as having become interested in information ecologies, as well as biological ecologies?

Hawken: Yes. I wrote a whole book around the term “ecology of commerce;” it was an oxymoronic term at first, but I think it’s less so now. The idea of ecologies of knowledge or information is self-evident to me; what was missing was a way to naturally construct that in a way that took advantage of new technologies.

Pang: It’s another kind of gardening?

Hawken: Absolutely. There’s a great book, Right Plant, Right Place. I’ve always loved it because if you put the right plant in the right place, it’ll thrive. You don’t need to do much. But if you don’t you got problems. To me, what happens is that we’ve got the technology of the Internet, which created this explosion of data, which has overwhelmed our ability to gather information. So we’re just trying to give people the tools to see that ecology, and create it. So again, it goes back to tools, but we don’t know what that ecology is.

When you do a search on Groxis, what you get is an information landscape of the Web. What you see if the topography that’s out there. You see that the world is amassing Web site and papers, and chat groups, and conferences, around this subject: there it is, on the map. So the very second a search is done—even before, because while it’s searching it starts building categories—you have some sense, an ecological sense, of what’s out there.

[Tape 2]

Paul Hawken and computers

Pang: One other thing I wanted to ask about what your own experience with computers: when did you get one, what was the first one you loved…

Hawken: I got a PC in 1980. Smith and Hawken was highly computerized. We didn’t have a mainframe, but we had a mini. I was interested in real-time data, so we set up a systems where I—and any employee—could look any time at order flow, and we could do a daily P&L. And it helped me to understand that information could be politicized, or it could be free—not free in the sense that you don’t pay for it, but in the sense that you had access to it—and I loved the fact that any employee could go online and look at our numbers, and see how we were doing, what items in the catalog were doing well, or all kinds of other things. We had a brilliant programmer, Lou Richmond. He hadn’t written a line of code until he came to work for us, but he was a brilliant pianist and composer, and I thought, “That’s good enough for us!” I figured, if he could do those things, then he could write code, and he did: I could go to him and say, “Well, how about this?” and—bam!—we’d have it.

So we always saw computers as something plastic, and malleable, as things that you could change to make them work better with you. So I watched with interest how over time, with the emergence of Microsoft Windows and Office, and corporate legacy systems have grown, how much of a stranglehold has developed.

You had mentioned Lotus Notes earlier. We’re actually talking to a company that is a conglomeration of 40 smaller companies, all stitched together with Lotus Notes, because they have all these different legacy systems that they can’t afford to replace or harmonize. What they have are data pockets, cul-de-sacs, and it’s clear that, the way the company is, the data that’s over here relates to the data over there, and the only way you can get it all together is with Lotus Notes. Now, Lotus Notes has an interesting taxonomy, and some interesting classification abilities, but this it is very tedious to get what they want. What they’re going to do is use Groxis to map the whole systems; and then each department person, secretary, they can just drag-and-drop and rearrange the whole IT system, according to what they need access to, what’s meaningful to them, and what relationships they think should exist between data. Groxis maps the Web of the underlying systems Because Groxis is a bolt-on, it is not disruptive to an enterprise’s IT ecology. We’re seeing the possibility that with Groxis we can return some flexibility to people, to customize an existing IT system a lot more reasonably than we can right now.

But my own experiences were just like everyone else: I got Lotus 1-2-3, and thought that a spreadsheet was just the best thing. I was a Mac guy in the end, and thought it was hot stuff. I’ve had the same ISP and e-mail address for over 18 years. I helped start the Well with Stewart Brand. So I’ve done e-mail for a long time, so that wasn’t so much of a revelation. But I’ve never been a techie, neither a technophobe nor a technophile. It’s a tool, and when it works well it’s fantastic, but it doesn’t fascinate me for it’s own sake.

Pang: It sounds like you probably started writing on a computer fairly early—

Hawken: I didn’t. I still don’t like reading things onscreen. I think it’s deceptive. The same text reads differently onscreen than on text, and I do not know why that is. So for a long time, I would type it, and then someone else would enter it into a word processor. I wanted to see the text on the page right away, and I missed the sound of the typewriter. But now I write on the computer, but still I find that when you print out a hard copy, you have a way of seeing, reading, editing, things that you just cannot do on a screen. There’s something about hard copy. Because if you’re a writer, that’s how people are going to read it anyway, in a magazine or a book, so I think it’s important to get to that format quickly.

I think it’s made me freer, but I think some people use it too much, in the sense that you can write forever. It requires a discipline in terms of brevity and editing, because it’s too easy to become prolix when you’re using a computer and word processor. A lot of books demonstrate that.

Pang: People with professional training spend a lot of time mastering the tools—bibliographic tools, databases, whatever—that organize a world of knowledge for them. Learning those tools is part of the process of becoming a member of a scholarly discipline. My sense of your background and your work is that that kind of professionalizing experience is not one that characterizes your work. It seems that in reading Ecology of Commerce and Natural Capitalism that those books made connections between what are normally completely different and unrelated kinds of activities. Driving up here, I was thinking that Groxis helps other people do exactly that kind of thing: you can use it to get at standing bodies of knowledge, but you can use it for identifying, articulating, and mapping out relationships between things that people don’t normally see.

Hawken: It’s a good observation. Of course, I’m not peer-reviewed, so I never suffer from worrying about what my peers think. I’m not saying that I don’t have peers, I do, it’s just that they do the same thing I do, and they have no problem with what I do, and we don’t review each other. Actually, I guess we do review each other in a way, but it’s a subtle thing that goes on. But I never had to work within the Victorian obsessions of the categorization of information where if you stepped over your boundary, you were roundly chastised and reproached; or if you did so, you did it with such humility. I never had that.

Now specialization has its place, but we also need people who can connect things. We don’t educate the connectors, so much as we do the specialists. So Groxis has a strong similarity to my work, absolutely, but I hadn’t made that connection. [Pang laughs] So thank you.