An Interview with CNI's Cliff Lynch

Matt: Hi, this is Matt with EDUCAUSE, and joining us now is CNI's Cliff Lynch. Cliff, it's a real pleasure to be with you here at CNI's Spring Task Force meeting. What are some of the major themes that you see coming out of this month's meeting?

Cliff: Well, there has been a lot happening lately, and one of the things that I find striking when I have looked at the program with that in mind, trying to extract themes, is how much a number of things seem to be progressing. And this goes really across the board, everything from the deployment of the shibboleth authorization infrastructure, which has been a slow and fairly painful process, but seems to be picking up steam, all the way through to a set of public policy kind of initiatives that seem to be finally starting to approach some conclusion. We've got the so-called orphan works inquiry which has been going on for I guess close to a year now.
Orphan works are fundamentally things that are still under copyright, because remember copyright today lasts an incredibly long time, 70 years after the death of the author. But that doesn't seem to have any real commercial value at present. So you can think of things such as a book that has been out of print for 40 years where heaven knows where to find the author or his or her heirs. Heaven knows where to find a publisher who has been merged, acquired, or gone out of business so that you can find a contract from 40 years ago to see if the publisher, its successors, or the author or his or her successors own the rights.

These are important intellectual and scholarly resources which are laying unused at this point, and is worth mentioning that while people perhaps talk most about orphaned books, it is not just books, it is music and photographs.

There was a tremendous story in the Wall Street Journal of all places a few months ago, but they described a couple who had been married for 50 years, so the children were planning their 50th wedding anniversary party for them. There was a photo that had been taken by a professional photographer at the wedding 50 years ago, and they wanted to have enlarged into a poster, and they took this to a photo shop that refused to do it because they presumably did not have the copyright in the photo, that was held by some professional photographer long vanished 40 or 50 years ago in another city.

These are the kinds of trouble we get into in this event. We have a couple of sessions looking at the results of the public hearing on the proposed legislation that has come out of the copyright office to address this. We also have a session on something with the remarkably geeky title of "The Section 108 Inquiry". For those of you who don't know your copyright law sections, this basically deals with special exemptions for libraries and archives and particularly, for example, exemptions around preservation. This is a cross industry group that includes both content providers and cultural memory organizations, that is looking at what needs to be changed in there are to meet the needs of the digital world. And hopefully will deliver some consensus recommendations. They have just finished a round of public hearings and will get a report on that, so lots of progress on all kinds of fronts. Institutional repository deployment continues to roll forward and more and more institutions are dealing with that. We see e-science and cyber infrastructure taking on considerable additional urgency in various settings.

Matt: Cool. Can you give us a sense of CNI's priorities these days? What is on the agenda and looking forward do you see the organization's course shifting in the future?

Cliff: Well, let's see. I think the last time we talked was October of 2005, and certainly in the period since then we haven't made any radical changes, and I don't honestly expect any near term radical changes. Our program is changing more now in terms of emphasis, focus, tactics. One of the areas we are spending a tremendous amount of time on is the whole set of issues around e- science and cyber infrastructure and e-research. As I talk to member institutions, a lot of them now have planning initiatives under way to try and work out institutional strategies in this area.
Since we talked in October we have finally seen the National Science Foundation appoint a director of their office of cyber infrastructure, Dan Atkins from the University of Michigan, and the chair of the committee that wrote the blue-ribbon report back in 2003 -- who I just think is a splendid appointment in that role, and he will be going to the National Science Foundation in June when he finishes up his teaching commitments for the semester at Michigan. I think that will really make a difference in terms of getting things moving.

When I look at what's going on in the UK as well, there is a lot of action in this area. I think that this whole question of data stewardship and data curation as part of the construction of cyber infrastructure services has really caught a lot of attention. The National Science Board, the National Science Foundation. The conversation is now really starting to spread in various directions into discussions about disciplinary data sharing policies and so-called open data. I see a lot of action there, and that is one of our major focuses.

We continue to worry a lot about institutional management of content assets, institutional repositories, digital presentation in those kind of issues. Learning management systems and learning spaces continue to be a significant part of our agenda. Joan Lippencott, our associate director, is doing some wonderful work studying different behaviour patterns and skills and preferences that are appearing in the new generation of students that are just entering our institutions of higher education. I think some of that work is very significant and proving to be very useful.

Matt: Cool. Before we began our interview I remember you spoke briefly about your international travels, and I know that internationally, CNI certainly has a role collaborating with a number of institutions. Do you see that accelerating or growing in prominence over the next few years?

Cliff: Well that is a tough issue for a CNI to manage. Let me give you the organizational parameters of this first ... we have a goodly number of members from Canada, we have some members from the UK and a couple scattered around continental Europe, but really not more than four or five I think. Then we have one or two members elsewhere. So we are very much a United States and Canada organization in terms of our membership footprint.
Having said that, we have enjoyed since almost the first days of CNI, a wonderful collaboration with the Joint Information Systems Committee (JISC) in the UK. For those listening to this who don't know that organization, the JISC is basically an organization that centrally manages a good deal of funding out of the United Kingdom higher education, and uses that to procure a set of services ranging from high-performance networking to national site licenses for content, to various kinds of infrastructure development programs and things like digital libraries across the whole UK higher education sector.

It is a bit more centralised over there than it is in the United States, and because of that they are able to put together programmatic things that are quite different than the sort of things that we are capable of doing in the US, where everything is much more bottom-up. And then organizations like CNI or EDUCAUSE or ARL may come in a kind of coordinating and best practice sharing role, after a lot of bottom-up activity gets moving. As we have enjoyed that collaboration for 15 years now.

We have more recently been working closely with the SURF Foundation in the Netherlands, an organization somewhat similar to JISC and the DFG. The problem really is that there are lots of developments going on abroad which can inform and advance the work here in the United States and Canada. We do not always hear about them enough. We have things in our work here to contribute to other organizations and other projects around the world, but it is a big world and we really don't have either the resources or the membership base to try and act as a global organization. So we act really as a North American organization, with an eye towards a very select set of international developments. Recognising that a select set isn't comprehensive, but certainly takes in many significant projects.

There is a lot going on particularly in the UK at this point, they are making very large investments in institutional repositories, for example. They had a bit of a head start I would say on some of the e-science and e-research work, as compared to the United States. The Netherlands is doing a very ambitious nationwide institutional repository deployment, where they are really designing what I would characterise as a system of institutional repositories, rather than just seeing individual institutions make the choice to do it as we are seeing here, and then later on they we'll figure out how to put various kinds of scaffolding on it to turn it into a system. So lots of interesting stuff going on there.

Matt: As you mentioned, it has been some months since we last spoke, and a lot can change on internet time. So I am wondering what piques your interest most these days, and what has changed since our last interview ... what are some of the most pressing issues of great concern to the CNI community?

Cliff: I mentioned a couple of the things that have, if not changed radically then certainly progressed in important ways, like the orphan works inquiry. There are some new things that are starting to emerge that I am very concerned about, but I think could have implications for the CNI community. On the public policy side there is this whole issue of so-called "net neutrality", or put another way about penalising either financially or performance-wise organizations that want to deliver bandwidth intensive applications out on the net. I am actually a little surprised at the way the press has cast this so far, which is really sort of the phone companies and the cable companies on one side, fighting out with people like Google and that was on and maybe some of the media companies on the other. I haven't heard much mention of higher education here, I haven't heard much mention of cultural memory organizations or of public broadcasting. All these institutions need to reach people off campus with increasingly sophisticated high-performance applications, and it is not just distance education scenarios. It's scenarios where you are going to have students and faculty trying to reach institutions from home to do their work.
And the prospect of higher education institutions, libraries, other organizations might have to pay a penalty, an additional penalty for taking on the public service, in essence, of being suppliers of important content is very disturbing to me. So I am watching that whole set of discussions with a lot of interest and concern.

We talked a little bit about cyber infrastructure, and those sorts of issues, and I really see that is gaining momentum as an enormous sea change across higher education. I will give you a couple of other bits that happens, I believe since we last talked, with the Commission on Cyberinfrastructure and the Humanities and Social Sciences of the American Council of Learning Societies has sponsored. That commission has issued its draft report for public comment, and it has stirred up a considerable amount of public discussion about really the whole future of the humanities and how digital technology is going to affect them.

I was very interested to see that the Modern Language Association has actually got an inquiry going on about how they believe tenure and promotion evaluation and scholarly work should be changed in response to kind of new opportunities that the digital environment offers, and in fact we're going to have a breakout session at this CNI meeting that they are going to present, describing some of that work.

But I think seeing the scholarly societies engage that is, very, very important, and as part of getting this whole transition to e-research really kind of deeply absorbed by disciplinary practice. Roy Rosenweig, who is giving the closing keynote, is a historian who has been doing marvellous work in digital history and how digital technology is changing both the practice and the communication of history. So there is another example of a discipline that is really starting to address these kinds of changes.

We are also seeing libraries step up to this in many different ways. For example, the Association of Research Libraries has chartered a task force looking at the role of research libraries in supporting e-science.

Specifically, how research libraries can situate themselves to be helpful within their institutions, and addressing some of these data curation and data storage challenges. I think that is an important development. I am seeing alliances now that involve libraries and presses basically looking at ways in which libraries can become more deeply involved in dissemination of digital works. And again, I think that is a significant recognition of some of the changes that are taking place in the whole scholarly process.

Matt: And you mentioned the potential for scholarly societies to recognise the potential impact for digital technologies on evaluating tenure and promotion procedures of organizations. How long do you think it will be before we start to see significant change in institutions or even for AN INSTITUTION to really begin to change significantly ?

Cliff: I think we're already seeing signs of that. People are so conservative of course about the uncertainties associated with promotion and tenure, and you can understand why. You know, you go out there as an assistant professor, you work incredibly hard for five or six years, and then you come up for tenure. And it is sort of like your whole life is on the line there, and what a terrible time to discover that well, the specific tenure committee at your institution isn't really very enthralled with digital technology as a means of communicating or documenting scholarship.
That is a pretty awful place to be, but I think we can point to now to a growing number of people who have been recognised for their work in this area and recognised among other things through tenure and promotion. Some of it is a generational process, some of it is a confidence building process frankly, and I think it happened step-by-step. I don't think we're going to be able to point to a specific institution and say "that institution has really fully accepted this in every regard". It is a much more diffuse process than that, I am afraid. But it is one that I do see making steady progress.

Matt: Cool, cool. Well, I don't want to dive too deeply into buzzword bingo here, but I know there is a lot of talk these days about Web 2.0 and Library 2.0. What is your assessment of this activity, and what are some areas that we might want to pay special attention to?

Cliff: I guess first I need to confess a little complicity perhaps in this 2.0 business in the sense that a couple of weeks ago CNI co-sponsored a day meeting called the Reading 2.0, where are we were trying to foster a conversation between some of the major organizations doing, if you'll allow me the terms of digital libraries and library automation, some of the folks on Google and Yahoo and Adobe, O'Reilly publishing, groups like that. And it was really quite interesting, I think everybody learned a great deal about other participants and how they are approaching some of this next-generation technology. And I think it was pretty helpful.

Web 2.0 is a lot about service decomposition, as I think of it, and being able to develop composable services of various kinds. The fundamental ideas there really aren't that new. You can trace some of this thinking, not necessarily on the Web context but in a broader context of network-based services, back into the kind of stuff that people like Bob Conn and Vint Cerf were doing back in the 70's and 80's but seeing them work out in a Web context, is I think, pretty significant.

And at the most fundamental level what we are doing, I would say, is moving away from the assumption that the human is always in the loop and always at the front of the loop, pounding away at the Web browser. We are trying to build a Web that is more hospitable to the construction of cascaded or automated services, agents if you want to use that kind of pretentious term, but really what I am thinking of is computational processes of various kinds that can continue to add value to other people's computational processes. I think that is an important long-term development, and pieces of it come under Web 2.0, pieces of it come under the semantic Web, I am going to come back and draw another analogy in a minute to something else.

But before I go there and let me just say a word or two about library 2.0. That I would say is a much more nebulous concept at this point, is something that a few people have thrown around I think as much as a challenge to folks and library community to re-conceptualize libraries and library services, particularly in a network in Web-based world of the sort that I was just describing. I guess I would think of it more as a challenge to the creation of visions than anything else at this stage.

Now let me circle back to the observation I made about getting away from this picture of the nature of the Internet is fundamentally a user at a terminal, or a laptop, or a workstation banging on his or her a Web browser, interacting with things. That is so deeply ingrained in the way we think about so much having to do with computing, networked information and network services. It is also very deeply ingrained in the way we think about interaction with texts, with collections of literature, with scholarly information, and I think that going beyond that is going to be one of the great challenges of the next decade.

Think about, for example, the discussions that we are hearing about large-scale digitization projects. Things like Google's arrangement with those five libraries to digitize enormous parts of their collection, or the work that the open content alliance is doing.

Now what that is doing, presumably, is putting large amounts of public domain information up online for reading. Obviously the copyrighted stuff is another matter, but one of the things this is going to do with the large amount of public domain material up for reading. Now, what they are going to let you do with this, what all the debate is about and all the speculation about the moment, is that you can sit there and page through the stuff.

Or maybe, imagine this, you can even download a book in digital form, a whole book! Now the hard drive on your laptop will hold thousands, many tens of thousands of these, and it is not just going to be about individual humans interacting serially with one text at a time, or a very small number of text. We are going to see, I think, an explosion of computational data mining, text analysis, indexing. What Google really does it compute on an enormous corpus of text. That is what it does when it indexes the Internet. That is what is going to be doing with these collections that it is creating from the library indexes it is digitizing. And they are not going to be the only player there. We're already seeing conversations in many scientific communities about supplementing generic XML markup in their scholarly literatures to flag specific objects of interest, maybe the name of the gene or species, or a specific disease, or a chemical sequence or a spectrographic signature.

Matt: Ala microformats?

Cliff: Precisely. Basically these are all these specialized ML's of various kinds, that people are starting to work within the various scientific communities. Basically all that does is reduce the ambiguity and given extra leg up to being able to compute over the literature and link it to factual databases. I can't resist mentioning, by the way, on the business of micro-formats, that is not a term that I was familiar with until a few weeks ago when it came up at that Reading 2.0 conference where someone from Yahoo was giving a short talk on their use of micro formats, and was absolutely unaware that this whole thing under the guise of specialized disciplinarian markup was starting to play out through a whole set of different scholarly disciplines. So those are exactly the kinds of crossovers that we were hoping to get out of that meeting, and that we did get some out of.

Matt: Do you have any thoughts on ways that libraries might employee the use of social software? In our last interview I kind of hinted at that ... do you sense any urgency in understanding this issue?

Cliff: I see a few libraries experimenting with this. There was a project which I think we have a session on in our meeting last December at, out of the University of Minnesota.

Matt: The Group Lens Project?

Cliff: Yes, the Group Lens Project work. Where they were trying to deploy recommended technology. There is a session here at this meeting describing some of the work that the University of California has done trying to think through how you might graft a recommender system on to the Melville catalogue.
But these are a very few and far between experimental projects. In general I don't hear a whole lot of urgency about this, I think that if anything the recent activity around the Patriot Act and other sorts of things has made libraries feel even more defensive about the privacy of their patrons, and to believe every more deeply that the best way to protect it is not to collect the data. So that is -- I am not seeing a lot of systematic stuff. I have seen a few cross library projects. OCLC for example, has done some work in this area.

But it is not really full bore social stuff. I am beginning to think that if we see that we will probably see a lot of that coming from outside of library settings, just because the whole sensitivity and privacy issues are so raw there.

Matt: You mentioned the importance of the great degree of sensitivity offered by libraries in this regard. Doesn't that in some respects prompt the need for it, even more? When I talk about a sense of urgency, you have the private sector doing lots of activity on this front without, up until the recent Google decision, regard to some of the privacy elements for their search engine utilization, not a whole lot of regard on that front for care taking of personal data. Any thoughts on that front? Do you think that there needs to be more urgency, or some sort of vision for how we can join together as these kinds of institutions, to provide a protected, sort of trusted alternative?

Cliff: Well, let me respond to a couple of different parts of that question. First off, I think that there is a growing recognition among the public broadly, and not just people and higher education, but really the broad public, as a response to this recent business with Google and the Department of Justice.
There is a sense that we really don't have a lot of privacy in terms of the activities we do on the Internet, the search queries we put into the net. And I think the thing that is particularly striking about the most recent business with the Department of Justice is this: this was not limited to Google. If I recall correctly, I believe I saw information from the American Civil Liberties union that disclosed that the Department of Justice had gone after a whole range of Internet service providers, search engines, filtering software providers, all kinds of folks on this big fishing expedition.

And they are gathering data here at a statistical level. This is exactly the sort of thing that I think is also making people nervous about some of this warrant-less eavesdropping that is going on as part of the anti-terrorist activities under the Bush administration. You start with one known number, and then you look at all the numbers it calls, and all the numbers that it called, and all of a sudden you are doing statistical analysis on these enormous, enormous networks of things, which is very far removed from a traditional ideas about getting warrants for looking into an individual's behavior based on probable cause and things of that nature. So I think there really is a growing concern about privacy in the networked world in general, and this really opened up a whole new dimension for the public that goes way beyond the traditional concerns about things like identity theft.

Having said that I think that, you know, there sort of two ways you could go in response to this. One would be to argue we need to build a search engine that has entirely transparent algorithms that keeps no records, and that would be a good thing to do or that only keeps records when you ask it and does it in a very transparent way. Building this sort of a thing would be terribly expensive, there are people trying to do it I believe, but you know, matching the level of investment that a Google or a Yahoo has put into this area, I think is pretty unrealistic. So my sense is that libraries rather than getting into a "we're going to do informed consent and be very careful and talk a lot to patrons about that specific tradeoffs between privacy and the ability to gain traction in a social software setting", they instead seem to really be backing away from this more and more to just "were going to protect people's privacy as best we can and not go into this sorts of nuance things". I don't know weather that's an ideal call or not but I think it's certainly a pragmatic one, especially for the public libraries. I think the other thing to remember, particularly in the research kind of a setting, is even if you did a recommender system around your circulation system, this would be of somewhat limited value because so much of the most important reading that many scholars do is based on the journal literature, and even now the pre-print literature and being able to capture their behavior streams there, is almost impossible because in general because you're accessing licensed journals that are scattered across dozens perhaps as many as hundreds of different publishers and the readers interact directly with those different publishers so there's no way to get a uniform click stream. There are a few places such as Los Alamos National Lab for example that really have followed a strategy that emphasis local mounting of content. They're in a much better position to do those kind of recommenders, but for the average research library, all they can see is the book circulation traffic and that's going to be a very limited part of the information use for many of their scholars so that's another challenge there.

Matt: Interesting. Well, the final report of the University of California Bibliographic Services Task Force seemed to create a lot of attention and I like to ask a few questions about it from your perspective. First and foremost what were some of the highlights for you and why do you think it has proven such a popular document?

Cliff: I don't know about popular, but let me talk to that a little bit. I think that it was wonderful for the University of California to undertake an inquiry of that scale and really sort of going all the way back to first principals about what they're doing and it's particularly significant to me to see the University of California doing that because they've got enough scale as a system to make a lot of different choices. They have alternatives that a single research university, except in very few cases, probably doesn't have economically and so in a sense they can look at collective thinking that may reflect out into shared services that other research universities can invest in down stream. It's worth noting by the way that the Library of Congress has also undertaken, I don't want to say similar, but related kind of an inquiry and there's a draft report for that, that's going to be coming out shortly as well. Now, I think that if you look at the kinds of questions they were asking and the kind of conclusions they were drawing at least as I interpret it, they're going a couple of different areas. On is about degree of centralization, historically every campus set its own online catalogue, and then there's also been a union catalog, Melville. One question is whether that continues to make sense, particularly in an era where the online catalog is becoming less and less important as an end user information discovery tool. It's not being replaces by things like Google but things like Google are drawing traffic away from it. As long as we have big book collections especially big physical book collections, scholars are going to need online catalogs, but clearly there's....

Matt: Or APIs to them.

Cliff: Yeah. Clearly there's a shift in emphasis here and I think it begs a question that other libraries are asking about, do we really need public online catalog for our collection or is some is kind of a union catalog through OCLC WorldCat or something good enough. And if we do need it for our own collection how much resource should we continue to sink in to this? Should we at this point say you know we've got a system that works well enough and we're going to just put it on maintenance? So that's one complex of questions they're dealing with. Another one is whether they should do more centralized things in terms of cataloging, and that again is a very important economic tradeoff. Historically we've moved more and more to centralized cataloging over the years, more and more use of copy cataloging, and I think they're probably going to get even more explicit about it. At the same time you've got those trends happening. Here are a couple of other trends which I think are also very much implicated in their report and I think are very significant. The first one is that we're increasingly recognizing that things other than published books and journal articles are important to scholarship. For example, image databases are playing a much greater role in both research and teaching, and you've seen the rise of systems like ArtStor recently too, start making larger collections of images available to the community. Now, if we're really going to have image databases as sort of first-class research and teaching resources, we need to describe those images. Right now there isn't the same kind of cannon of images that's been created by the system of publishing over the last umpteen years, which lends itself to centralized copy-cataloging. We've got a lot of people holding collections of images, that have partial overlap, and describing these is a hugely expensive project. Yet, I think there's a recognition that catalogers are going to spend more of their time doing description of things other than published books or journal articles, and that we need to deal with that. Another piece of it is that for textual materials we are steadily moving away from a world of surrogates, in other words, you have a physical book which is a considerable amount of nuisance to lay your hands on, and then you have a little digital record describing the physical book with a few subject headings and an author name and things like that. If you move away from that world to a world you where you have the whole book as a digital file, do you really still need that bibliographic description or can you drive it from the book? If the book is appropriately structured, so that suggests a whole reallocation of effort and expense away from kind of classic bibliographic control over time, particularly the expensive things like the generation of subject headings. If you can do full text search and computations on the whole book and its index, there's some question about how much it's really worth investing in subject headings, so they frame that question as well. These are, you know, sort of big questions that look at gradual but inexorable changes in the kind of world in which traditional bibliographic control operates, and it's really important to ask them. I don't think you're going to see really abrupt cutovers as a result of that report. You don't do abrupt cutovers when you've got stewardship responsibility for eight or 10 million physical volumes and the process of converting these to digital is going to play out over a period of decades or longer. But were starting to see things like that report challenge us to look at gradual reallocations of emphasis and reallocation of resource away from the old and towards the future. And I think it's a very important report in that sense.

Matt: Yeah. Cool, cool. And how long do you think it will be before we see realization of one or more of the more aggressive options that were laid out on the report? And lets take a look at the OPAC side in particular.

Cliff: I don't really have a good sense of that, and I'm not trying to be evasive, it's just that there's a whole process that going to run inside the University of California about sorting thought what do with that report. I'm not close to that process.

Matt: Whether it'll be at the University of California or elsewhere perhaps. I mean do you see instances where others are sort of leaping on to this and saying "this is something that we have to do" or is it everybody sort of saying "hold steady, we don't have the budget to even consider anything like this"?

Cliff: I think it's more gradual. I think that within UC it will result, over time, in a reallocation of resources. But I think, you know, they'll probably go through a pretty complicated internal process of figuring out how to implement those recommendations and which ones they're going to implement. There are probably not a enormous number of other institutions who can do things along those lines other than around the margin. You are clearly seeing more and more of the research libraries investing in the description of non-textual materials. And I think, actually you know, ArtStor may be an important development in this context because what are store has done is built up a reference collection of described images so to the extent to which, as a library, you want to license that collection, it comes with all it's description. So you're sort of back in the economics of copy cataloging, right? You describe it once centrally and piggy-backs on that. And as the ArtStor database grows and becomes more widely accepted, I expect that some institutions may start to think about it as a framework for developing a shared pool of described images. It's just a slightly different economic model than an organizational model that the one we did for books.

Matt: And you talked a little bit about the marginalization of the catalog a little earlier. Meanwhile we've seen what NCSU has done with Endeca, certainly not all libraries can afford to try and reproduce that kind of technology. What do you think the future hold with regard to taking library systems like these to the next level? Are they... are we going to continue to see people experiment in these kinds of ways, or are we going to see scaling back like you mentioned earlier that people have to consider the tradeoffs of building value-added other layers versus trying to innovate on these sort of more familiar technologies? Any thoughts on that?

Cliff: Well, I mean the Endeca thing is a very nice piece of work, and I'm not sure that I entirely agree with you that other libraries can't do that. I think that certainly NCSU paid some price for leadership and trailblazing as leaders and trailblazers always do. But I think you could see that kind of technology moved much more into these kind of mainstream offerings over time, and it may look a lot less exotic three years for now, five years from now, than it does today among library catalogs. I think that the really hard question, and I don't know that I really know the answer to it is, how much resource to continue to put in to enhancing your online catalog in a world where the collection of information you're trying to manage and provide access to, continues to diversify and multiply and move away from physical books. And that gets into questions as well, which don't have crisp answers about what's the appropriate scope of an online catalog. You know, there are some institutions who say it's books and journals, not journal articles, but journals, you know, the names of journals that they subscribe to. There are other institutions that have started to put in their manuscripts and photographs and things like that and, you know, this is very disconcerting because the scale of the objects is off. A book seems like a bigger object like an individual photograph somehow. So, you know, there's some argument maybe the right thing to do is to catalog your photographs at a collection level and put them in the online catalog, but then that's not always that useful for finding things. So there's a big scope issue on which people are struggling about what are you putting on your online catalog, what is it a catalog of these days. Is it a catalog of things you provide access to? Is it a catalog that includes free things on the net that you think might be important to your user base? And I think your choices there are probably interact with your choices about how much investment to make in advanced retrieval mechanisms and interfaces for your online catalog going forward.

Matt: And do you expect the consolidating base of system vendors to provide more compelling solutions at a faster pace or do you expect to see a consortium of libraries joined together to work on open source alternatives that try to emulate a given set of features? Any sense of how the market is going to develop on this front?

Cliff: I don't know about the vendor marketplace, it has always been a pretty fragmented marketplace with a reasonably large number of, relatively speaking, small vendors. One of the consequences of that is that it's been a marketplace that has a lot of trouble finding money for R&D, particularly for the R part. I mean there's plenty of development but it's development that's targeted at something that turns into a product one or two releases out. It's not, you know... Let's take a flyer on this research thing and, you know, maybe it'll turn into a product in ten years and maybe it won't. There just isn't, by and large, the financial base to do a lot of that in the library automation community. I don't know that there's going to be a lot of heavy consolidation in there. I think most of the vendors that are in there right now are fairly viable. There was a considerable round of consolidation back in the... probably late 90s I guess, the latter half of the 90s, and I don't see that we're necessarily ripe for that again. I also don't see a lot of new vendors coming into that field though. Now the open source question is really interesting, and I'll say some things that probably a lot of other people will disagree with. There are two sets of reasons purposed for doing open source in higher education. One reason, and it would be I think the reason why you'd do a system like Sakai for example, or a system like D-Space, is because this is a system that is doing a bunch of core functions like teaching and collaboration or like stewardship or institutional assets, you need to be able to adapt and evolve those system in a very complex environment, that involves lost of moving parts and lots of external systems, and you need to be able to continue to innovate really on a distributed basis, because nobody really fully understands the problems and the solutions yet. We're still finding our ways in term of how technology can transform teaching and learning. We'll be doing that for, I imagine, 50 years or longer. Certainly we're just at the beginning of understanding how to do stewardship of things like complex data sets. So for those things open source to me is compelling. It's something we need to do because we have to do it and we have to find the funding to do it. Now there's another set of argument that are advanced about open source which basically say that we've got some kind of reasonably stable system that we're purchasing from one or a group of vendor today, and it's expensive, and if we could only do an open source version of it, it would be cheaper. I am much less confident about those arguments, that's the argument that's being made around KUALI, the administrative system. I think that's the kind of argument that's typically advanced when we talk about "let build an open source library system". Many of the functions of the an open source library system we understand very well now. There are functions like circulation and acquisition and inventory management, and this kind of stuff which 30 years ago we were trying to understand how to do those in an automated world. Now we have very stable systems for that even the online catalog, unless you really want to invest in a whole new cycle of research and R&D there, if fairly stable among these vendors. So I am, you know, deeply skeptical that doing an open source library system would really save much money, and more to the point it seems to me there's an enormous opportunity cost. I think that place that we do best to focus our efforts on open source is the places we have to be. Talent is scares. And so I'd much rather see that talent and that time go in to contributing to these mission critical systems that we're just trying to understand, like our stewardship system than in recapitulating what looked to me to be mostly pretty viable and well thought out and solid commercial products.

Matt: And what about the recent d-LIB issue that was built around the "what would you do with a million books" theme? Any especially interesting information to relay from that series of articles?

Cliff: Well this is the March 2006 issue and you know I think it very much ties into the conversation we were having a little earlier about moving away from thinking about the future of text as still things that humans read one at a time. And many of those articles explore themes about translation, correlation, cross-referencing of very large corpora of texts and what that can contribute to scholarship and, you know, I think their well worth reading on that basis. I think that people who are interested in this might also find some of the literature about text-mining in the life-sciences particularly pretty eye-opening. And some of the work that's going on there.

Matt: And what would be some pointer to some of that for our listeners?

Cliff: Well, in the UK, the National Center for Text Mining that was just set up is hosted at the University of Manchester and I happen to know about this because I'm their advisory board. Their remit primarily focused on biological and life sciences so if you hit their site that would be one place you could pick up a lot of pointers to that. In the US there are a lot of research groups working in this area, the National Library of Medicine has done a great deal of work in this area as well, so I think a couple Google searches will get you [headed in the right direction].

Matt: Cool! A few months ago you spoke to the folks at SURF about the implications of digital libraries. Can you share a few highlights about that presentation?

Cliff: Sure. Actually what I was specifically talking about there, they asked me to talk about, and I kind of welcomed it as an opportunity to organize some thoughts, was specifically what impact libraries were likely to have on education, and I think I may have taken this in a direction that they weren't entirely expecting. But let me just give you a couple of bits of the stuff I was thinking about there. One of the things I'm very struck by, and we may have talked about this a bit in October is personal storage is getting very big and very cheap, and the notion, for example of when you issue each schoolchild his or her first laptop, which you know probably ought to happen some time -- grade five, six, somewhere around there, I don't know, why not give them a good starter library of ten or twenty thousand books and select a journal articles and a comprehensive encyclopedia and a few other things to fill up some of the disk space on there, why not? That's clearly doable; this decade. And the only things that are getting in out way are some copyright things or licensing things, and the will to do it. Now, next point, wireless is getting very ubiquitous, particularly in schools and universities. You know, you can point to some funny situations now where we've actually achieved the wired classroom, although we've done it wirelessly rather than Ethernet to every chair in the lecture hall, and having achieved this, some faculty are really quite unhappy about it, and are banning, you know, laptop use, or the ability to turn off wireless in their classroom. Others are adapting to it and thriving with it, but the net effect of this is that a lot more information is going to be persistently available, either on your laptop or through your network connection. And you're going to have things that computer on this, that index it well, that known what you've seen and what you haven't seen. Now, flash backwards to the... probably early 1970s, the great calculator controversy. Right? We started to do cheap hand-held calculators and so the question is how much time should we spend children to be really fluid in long division and things like that, when they can punch it out on these calculators. Should they be able to use these calculators in class? Should they be able to use them in exams? Well, now the question is going to be like this, except its going to be about are you able to use the World Wide Web and all the digital libraries in the world in exams, in class? I think it's going to create intense pressure on some of our assumptions about mastery versus memorization, about when the goals of specific education processor, about the ability to find facts as opposed to the memorization of facts, as opposed to the ability to analyze and evaluate facts, so I think this is going to kick off, I hope it's going to kick off, honestly. Some really hard thinking about the role, the extent of memorization, and the necessity of memorization in various educational settings. I have to share this one with you, and you should talk to Rozensweig more about this, he did a wonderful, wonderful article a couple months ago about something he calls H-Bot, which I basically a history robot. And one of the things he did was he gimmicked up a version of this to try and find answers just by doing some computation on the web to, I think it was the high school American history standardized test. And he did pretty well, which suggests to me something where that falls on the analysis verses memorization scale, and I'm not an expert on teaching of American history but I think these things start to underscore the questions there, and they're really hard questions. I don't know what the right answers to them are.

Matt: Yeah, yeah, indeed. Very cool. And what can you share, circling back to research now, regarding the... Toward 2020 Science publication recently issued by Microsoft Research and the offshoot of that, Nature's issue on 2020 computing?

Cliff: I'm still working my way through some of this. There was a lot of material there, especially in the nature special issue, but I think that what I took away as just kind of a compelling message was for all that information has transfigured the practice of science, over the last 20 or 30 years, the changes that are coming or going to be at least as dramatic. There is a, I don't remember weather it's an attachment or auxiliary file or what, but there's a file that comes with the Microsoft Science 2020 Report that's sort of a timeline roadmap for things that they think can happen in biology between now and 2020. And keep in mind were talking 15 years now, we're not talking, you know, decades or centuries, we're talking... you know this is going to happen in the lifetimes of most of us. If you look at this stuff they map out about the increased understanding of biological processes and then out ability to control them, it's really pretty staggering. You start seeing that the distance between that and some of the more speculative stuff that people like Verner Vinge and Ray Kurzweil talk about singularities, you know, really maybe isn't that far apart after all. It's pretty striking stuff. I mean, I can't image what a biologist of the 1950s or 60s would think of biology today. And the way in which it's turning into, in some ways, an information science.

Matt: Cool, cool. Well, Cliff, it has once again been a really great pleasure speaking with you, and I'm sensitive to the time commitments here, but do you have any closing thoughts?

Cliff: It's always fun to catch up and I'm struck by how widely our conversation both last time and this time have ranged from, you know, things that are fundamentally public policy questions all the way through things involving the practice of science. I think that, I would just urge I guess, people who are interested to, keep your eyes open broadly for these changes. I hope in future if we talk again and we'll have a change to perhaps probe further into some of the social implications of this.

Matt: Absolutely. Well thanks again for taking time to speak with us and we wish CNI continued success in addressing this wide range of issues.

Cliff: Thanks, always a pleasure.