Interview date: March 3, 2010
Denver, Colorado

Clifford Lynch

Summary: By background and training, Lynch is a computer scientist, although he has spent most of his working life dealing with issues around information retrieval, library automation, computer networking, and those kinds of things. Lynch has been involved with various phases of libraries since about the mid 70’s and has watched the evolution of technology as it has gotten less expensive and more accessible. Lynch’s working career began at the University of California in 1979 to around 1997. He did various things there, starting with the building of the online catalog, Melville, a catalog covering the UC system and the computer networks needed to support it. Lynch helped grow Melville from an online catalog to a multifaceted kind of information resource that held various kinds of primary and secondary materials.

Lynch left UC in 1997 to take his current position as director for the Coalition for Networked Information (CNI) where he sees a tremendous amount of digitization opportunities. However in his personal experience, Lynch has primarily dealt with the delivery of digitized page image material. He gives the example of the University of California initiative as a pioneer in a number of projects to take page images of scholarly journals and deliver them across the network to display terminals scattered around the UC system and beyond. Lynch saw this as a substantially technical undertaking, as it was the early 90’s, when all the publishers could give him was bitmap images instead of the more user friendly .xml files and without the web browsers that became later available. He wasn’t much involved in the digitization side of it, that was done by the publishers and we worried about delivery. UC was involved with the TULIP Project with Elseveer and IEEE which were two of the larger imaging projects Lynch was directly involved in the deployment of.

There were many things lynch was involved with on a consulting basis. Back in 1983, he was involved in a series of projects that the National Agriculture Library (NAL) initiated. They were delivering all kinds of material from the topics of sustainable agriculture to third world countries. NAL set up a pretty significant line for setting up the CD-ROM. They were more concerned with readability, rather than perfect reproduction and fidelity of the material. Lynch also had peripheral involvement with several projects involving the imaging of medieval and renaissance manuscripts in the late 80′s.

Later, in the 90′s, various UC entities got quite involved with the NSF Digital Library Program, the Geo-spacial project at UC-Santa Barbara which is, in some ways, going on today still, as the center of digitized material: aerial maps and that kind of thing. There was another project at Berkeley that had a lot of material about the floral and fauna of California. Those were some of the early digitization things that Lynch was involved with in one capacity or another.

The other thing Lynch mentions, which he says he cannot take any credit for, other than doing a bit of consulting for them, was the [IMLS-funded] big image database that the New York Public Library launched about 4 or 5 years ago…consisting of about half a million images from their collection. It represented a fairly early effort of theirs to deal with a large and extremely heterogeneous collection of imagery.

Lynch saw the challenges as divided in to 3 periods. The first problem was the period before CD-ROM got stable, then there was the period where you were doing things on CD-ROM for most projects, doing work station delivery, then there was the period when you were putting stuff on to the network. Lynch says you can think it as the 3 epochs. Pre-CD-ROM was horrible; there were no standards for anything. There were early forms of tiff for dealing with images, compression wasn’t well standardized, people were using some variation of Hoffman coding, at least for bi-tonal images, but it was pretty messy.

People couldn’t afford to use magnetic storage and there was an endless supply of crappy and unstable optical storage solutions some with the “extra bonus” of jukebox-like technology that constantly had mechanical and optical problems. Proprietary file formats were used, which were insufficient for the data. You had one silo where you could do metadata stuff, another one where you were doing imaging stuff and a real workflow problem connecting the two, reliability. And trying to find something to connect the two. Things were pretty bad back then. People wanted use multiple large mainframes that were a particularly bad mating for various reasons. People would wind up with a mini computer feeding larger mainframes. There were basically no metadata standards and zero standards for delivery. Lynch wound up building custom delivery things for different platforms that had a fairly short life because the platforms were unstable themselves. One of the things people tend to forget now is the manageable rate of the sufficiently large deployed base of technology in which the vendors still have to worry about compatibility. There was a tremendous churn in the industry and a great deal of discontinuity. Companies would go away and sunk investment data would be stranded. This is something than is far less commonplace today.

Once Lynch got to the CD world, rapid standardization in regards to image handling and Apple’s QuickTime began to occur. JPEG standards then became well established. Metadata standards, relevant for images other than things like art, cultural, and scientific objects…still faced an integration problem. There were not good platforms for building a system that combined text and imagery nicely. The early things, like the Voyager, had a certain hand crafted quality which were beautiful for these closed content systems, but many of these cultural memory institutions were interested in not just snapshots, but in growing collections.

People wanted to build up these network based servers and do delivery across the network. That was hard for a lot of reasons. Coming up with the right kind of architecture for distribution and how to function between various operating systems and picking the right kind of standards for the distribution of functions was difficult. It was quite a while before the network grew into the kind of promise that it immediately offered and you could sense.

Those were really complicated issues that were very poorly understood. Such things as do you use something like X Windows or Z3950 and move the images around as data objects and then deal with them out on the workstation. One of the problems with using a client serve or style protocol where the object being delivered is an image, but it’s really the software that is understanding the image. People forget how slow the networks are…you can probably faintly remember the earliest days of the web, where there were graphic web browsers like Mosaic. For most people, access to the web was a dial up modem operating at 32 kilobits, may be 50. And that was not enough capacity to support fluid imagery work. It was quite awhile before the network grew into the kind of promise that it immediately offered and that you could sense, like the first demonstrations of Mosaic.

Browsers of course were an absolute godsend for all kinds of imaging things because they provided a common platform. Prior to web browsers and things like Gopher, there was still a tendency to have custom clients for certain databases, something that really impeded the use and reuse of images in a really serious way.

One very important shift for Lynch wasn’t one that happened in a moment, but was rather a change in thinking, that Lynch started working through in the late 90s was the realization that we were rather quickly going to get to the capability to image objects and other materials in such a way that they were as good as the underlying objects for not all purposes but for many purposes. The whole strategy of imaging important collections of cultural heritage was really a stewardship and survival strategy.

It was something that institutions charged with stewardship were obligated to do to be good stewards. And the strategy going forward would be to have digital records of the material and then the underlying material would give you protection and leave you in a much better place, given the ugly multi-mullenia history of wars and natural disaster.

In terms of the underlying technology, probably the thing that Lynch believes he didn’t see coming was that how much trouble color fidelity was going to be. Maybe he was naïve about this, because Lynch had started to hear early on from people who worked for example in film that it was going to be an issue. Lynch heard this too from people who published art books, where they were trying to do high quality reproductions of prints and paintings and things and believed that it was going to be an issue for Lynch.

In terms of the amount of grief we have had with standards are color space, color management, and calibration schemes, this is still a big headache for work flows that are concerned about capturing color with high fidelity. Another area we would have been so well served to get out in front of with a good standard 10 or 15 years ago, is the situation when you have textual material imaged and an OCR transliteration attached to it and you want to work with those two as connected objects and people have way too many ways of doing this today. It is a disgrace. If you look at for example, at the projects that Mellon has been funding to digitize manuscripts. That is the kind of thing that if we had made a strategic investment in some standards and software early on, it would have saved a lot of pain, some of which is still to come as we get things in to a homogeneous form. So that is another problem area.

Beyond the technology though, and keep in mind that a lot of what Lynch did in the 80′s and 90′s was thinking about building big deployable systems. When you look at the history of digitization work, there was a lot of over promising. It is still very hard for people who make large scale funding commitments. And Lynch doesn’t mean a small digitizing project but, the “I’m going to make a commitment at scale.” It is hard to figure out when the right moments are for cost performance. To move from little projects to doing something of scale. There has been a history of doing things of scale too early. When technology was over-promised to funders. The kinds of things that one deals with in pilot projects are very different than the things one worries about at scale. It is easy to make nice looking pilots of image collections. It is much harder to do if you have 50,000 concurrent users beating on this thing from around the world, where you have no control of the devices they are using and the control patterns, when you want it to be robust. That is a different kind of environment. Collectively, everyone didn’t do as good as they could of in thinking in strategic terms and when to fund pilots and fundamental research and communicate the outcomes of these to policy makers who would drive the decision.

In terms of digitization for two dimensional things, in regards to the capturing of it, Lynch thinks the questions are really about color management strategies (where that is applicable), and about levels of quality and how one judges quality. Lynch finds there are a couple of tricky decisions to make…like whether you are going to keep uncompressed reference images or whether you are going to go with something like .jpeg 2000 and let it do a bit of compressing. Lynch says the question of how much resolution is enough is an important one. You need to understand that people have very strange ideas about the future of displays. The nature of displays really hasn’t change very much except they have gotten flat we have gone to an LCD technology but the pixel densities haven’t changed very much.

Have you seen the work on the octaputer? There is a professor at UC-San Diego named Larry Smar. He ran the National Center for computer applications back when they built mosaic. He has been a pioneer in high performance computing for about 20 years. When he went back to San Diego, he was really interested in high performance visualization of models and biological phenomena coming out of super computers. He decided what they needed to do was get serious about display devices. He started building these things called octaputers, rolls of LCD’s ganged up 20 by 4 with a Beowulf cluster, and IO drivers on each of the parallel machines of the clusters. It is a machine of power and storage, where you can zoom and drag and drop stuff and have screens. We are going to start seeing stuff like that, as display devices, in high end imaging applications. Like medical imaging or simulations, people are starting to work with it. Even in cultural heritage too. You will see large painting reproduced, where you can zoom in on detail. We need to be cautious about that “How much is enough” resolution. Back in the day, we underestimated that. This really only applies in cultural heritage to things that people really need to see the detail of. It would be silly to do this for hand manuscripts and stuff like that.

Lynch thinks there are questions that people should be thinking about in regards to multispectral imaging, something that is starting to be used in manuscript scanning, for example, setting you up for nice image enhancement that you can’t do with mono-spectral because of the light imaging. Lynch uses the example of different things people have done with the Archimedes pall-um set. Having those scans in different wavelengths are very useful, but how you correlate and align them can be messy. As a side on OCR, one of the things we are going to need to spend some R&D money on is OCR for more kinds of texts and hand written texts. It is interesting to hear this rhetoric about engaging people with primary sources, but the fact of the matter is that most kids aren’t taught handwriting anymore, the sort of Victorian copper plate kind and you show them handwritten manuscripts or letters and they may as well be looking at a 10^th century manuscript. Things we can do to help with OCR and transliteration of those, you know, maybe another way of saying it is that we have a bigger paleography problem than we are willing to admit.

The digital world is starting to see a lot more work on imaging 3D objects and there are a whole lot of different strategies where you document it from all 4 sides and the top if you need too. For statues now, we are doing laser scans…if you look at stuff like the Michelangelo project or what Stephen Murray has been doing with French churches, we can do whole building and statues..the whole question of digitizing 3D stuff is very much on the table.

One of the things we need to get a lot smoother about is zooming in and out and the rendering of 3D things. One issue is with the individual object itself. Another is the whole business of color calibration upon delivery, which is a mess and needs some thinking about: how you contextualize objects in an apparatus that involves layers of annotation, that involves the multispectral imaging, that involves transliteration and translation attached to that.

We don’t have good standards for that…certainly people have built good individual systems but we need to make this a routine process that allows databases to inter-operate freely. And that is much easier for people on the capture and delivery side. Lynch thinks one thing we are running into as we digitize materials is that people want to work with them in different ways.

One thing textual scholars do a lot is compare texts. And you know it is not uncommon in the days before computers to see a scholar working with 2 or 3 manuscripts trying to understand them to build a critical edition. Even working with 2 or 3 is clumsy in most systems right now.

In reality, it is not just 2 or 3 texts it is just that is all one could work with. So you have a project like “The Romance of the Rose,” at John Hopkins. So they had a 100 to 150 different manuscripts they were using. So now the question is, how do you build tools to allow people to understand the variation among 100 objects? Think of examples. Ancient Greek vases. You have a real challenge as you think about a big museum that can exhibit 5% of its holding most of the time. You will see a whole line of development changing around these things, to open up large encyclopedic museums. How do you indicate uncertainty and the reconstruction of damaged things and represent that so people understand what they are looking at? This gets you to the London conventions for the reconstruction of architectural models, where you might have the ruin of a building and you want to start with the ruin and represent what you think the place looked like. And in trying to understand what parts are what based on the evidence. It becomes really critical and scaffolds on top of the actual digitization, that provide evidence around the areas of archeology and architecture and damaged manuscripts that are being reconstructed. There is not near enough thinking about how to do that.

When one is dealing with pictorial material, it becomes impossible to comprehensively describe because it has so much in it and is so rich and has interpretations that are influenced by cultural context in it that is involved with so much context: allegories and historical representation, all in really complex ways. Everyone is really bad at describing images even when we can throw huge amount of human time into them.

Most of the time we don’t have the money to throw the human energy at it and do the elaborate cataloging. It is not uncommon to find things on the net like 20,000 photographs of New York City street life in the 1950′s. What happens around these collections, whether they are well documented or sparsely documented, is that…basically they turn into a conversation with the audience. The impact of this phenomenon is rather underestimated.

The library of congress, putting of photographs on Flicker commons and dealing with the conversation around it, you start realizing that there are lots of people around that are interested in aspects of material culture, interested in airplanes, trains, machines, cars…there are lots of people interested in genealogy, family, and local history and a ton of material in private hands. The place where this reaches critical mass and ignites is where you are dealing with photographic collections, because photography is a relatively recent technology. So unlike putting up images of 16^th century painting, the property of most collections, big ones especially, is that is touches the edges of living memory. You know, you make connections between what you see and yourself. That was the story in the 1930′s, and you create this really deep story telling. When these kinds of things become available, you have to figure out what you want to educate with this conversation and lend authority too it. How do you distinguish between assertions you make and that other people make about the object? How do you deal with the acquisitions piece about the object? It could be a treasure trove or a disaster for a cultural heritage institution. As a community, we need to start talking about it in a much more sophisticated way then we have been. We have some pilot projects that some institutions have done good thinking about, but we should not be reinventing the wheel on this one. In terms of making large collections of photographic material available, it becomes a monster issue.

One last thing, since we are focused on cultural heritage. One of the places where Lynch has been paying a lot of attention lately and talking with a lot of people is what happens with cultural heritage going forward? As archives and special collections in libraries and other groups acquire personal papers and collections from people, these have more and more of a digital competent. You are always a little bit in the rear-view mirror there because you get things after they have been sitting around for awhile. If you look at what people are doing today, it is absolutely clear that the nature of the personal collections is changing dramatically and the amount of still and moving image material. A family from the 1950′s, somebody might have a little bit of 16mm film from someone’s birthday.

We’ve not only got things digitized to our standards of cultural standards but will have some pretty funky images coming in off of cell phone cameras and stuff like that, that leave a great deal of something to be desired. I think there are a whole new collection of issues as we start thinking about what are collections are going to look like in 2050 as documenting and understanding the lives of individuals in our culture, as it continues to be an important activity.

Interview date: March 3, 2010
Denver, Colorado

Summary: Lynch says it wasn’t a sudden revelation, but that it became clear early on that the potential was there to really just completely blow open the doors of cultural heritage institutions and fundamentally change the equations about the way things were typically used. And it was largely the question of how long would it take for the cost curves and the deployment rates to really make it possible to do it. One very important shift for Lynch wasn’t one that happened in a moment, but was rather a change in thinking, that Lynch started working through in the late 90s was the realization that we were rather quickly going to get to the capability to image objects and other materials in such a way that they were as good as the underlying objects for not all purposes but for many purposes. The whole strategy of imaging important collections of cultural heritage was really a stewardship and survival strategy.

In terms of the underlying technology, probably the thing that Lynch believes he didn’t see coming was that how much trouble color fidelity was going to be. Maybe he was naïve about this, because Lynch had started to hear early on from people who worked for example in film that it was going to be an issue. Lynch heard this too from people who published art books, where they were trying to do high quality reproductions of prints and paintings and things and believed that it was going to be an issue for Lynch.

In terms of the amount of grief we have had with standards are color space, color management, and calibration schemes, this is still a big headache for work flows that are concerned about capturing color with high fidelity. Another area we would have been so well served to get out in front of with a good standard 10 or 15 years ago, is the situation when you have textual material imaged and an OCR transliteration attached to it and you want to work with those two as connected objects and people have way too many ways of doing this today. That is the kind of thing that if we had made a strategic investment in some standards and software early on, it would have saved a lot of pain, some of which is still to come as we get things in to a homogeneous form. So that is another problem area.

Interview date: March 3, 2010
Denver, Colorado

Clifford Lynch

Summary: In terms of digitization for two dimensional things, in regards to the capturing of it, Lynch thinks the questions are really about color management strategies (where that is applicable), and about levels of quality and how one judges quality. Lynch finds there are a couple of tricky decisions to make…like whether you are going to keep uncompressed reference images or whether you are going to go with something like .jpeg 2000 and let it do a bit of compressing. Lynch says the question of how much resolution is enough is an important one. You need to understand that people have very strange ideas about the future of displays. The nature of displays really hasn’t change very much except they have gotten flat we have gone to an LCD technology but the pixel densities haven’t changed very much.

Lynch thinks there are questions that people should be thinking about in regards to multispectral imaging, something that is starting to be used in manuscript scanning, for example, setting you up for nice image enhancement that you can’t do with mono-spectral because of the light imaging. Lynch uses the example of different things people have done with the Archimedes pall-um set. Having those scans in different wavelengths are very useful, but how you correlate and align them can be messy. As a side on OCR, one of the things we are going to need to spend some R&D money on is OCR for more kinds of texts and hand written texts. It is interesting to hear this rhetoric about engaging people with primary sources, but the fact of the matter is that most kids aren’t taught handwriting anymore, the sort of Victorian copper plate kind and you show them handwritten manuscripts or letters and they may as well be looking at a 10^th century manuscript. Things we can do to help with OCR and transliteration of those, you know, maybe another way of saying it is that we have a bigger paleography problem than we are willing to admit.

The digital world is starting to see a lot more work on imaging 3D objects and there are a whole lot of different strategies where you document it from all 4 sides and the top if you need too. For statues now, we are doing laser scans…if you look at stuff like the Michelangelo project or what Stephen Murray has been doing with French churches, we can do whole building and statues..the whole question of digitizing 3D stuff is very much on the table.

One of the things we need to get a lot smoother about is zooming in and out and the rendering of 3D things. One issue is with the individual object itself. Another is the whole business of color calibration upon delivery, which is a mess and needs some thinking about: how you contextualize objects in an apparatus that involves layers of annotation, that involves the multispectral imaging, that involves transliteration and translation attached to that.

We don’t have good standards for that…certainly people have built good individual systems but we need to make this a routine process that allows databases to inter-operate freely. And that is much easier for people on the capture and delivery side. Lynch thinks one thing we are running into as we digitize materials is that people want to work with them in different ways.

One thing textual scholars do a lot is compare texts. And you know it is not uncommon in the days before computers to see a scholar working with 2 or 3 manuscripts trying to understand them to build a critical edition. Even working with 2 or 3 is clumsy in most systems right now.

In reality, it is not just 2 or 3 texts it is just that is all one could work with. So you have a project like “The Romance of the Rose,” at John Hopkins. So they had a 100 to 150 different manuscripts they were using. So now the question is, how do you build tools to allow people to understand the variation among 100 objects? How do you indicate uncertainty and the reconstruction of damaged things and represent that so people understand what they are looking at? This gets you to the London conventions for the reconstruction of architectural models, where you might have the ruin of a building and you want to start with the ruin and represent what you think the place looked like. And in trying to understand what parts are what based on the evidence. It becomes really critical and scaffolds on top of the actual digitization, around the areas of archeology and architecture and manuscripts. There is not near enough thinking about how to do that.

When one is dealing with pictorial material, it becomes impossible to comprehensively describe because it has so much in it and is so rich and has interpretations that are influenced by cultural context in it that is involved with so much context: allegories and historical representation, all in really complex ways. Everyone is really bad at describing images even when we can throw huge amount of human time into them.

Most of the time we don’t have the money to throw the human energy at it and do the elaborate cataloging. It is not uncommon to find things on the net like 20,000 photographs of New York City street life in the 1950′s. What happens around these collections, whether they are well documented or sparsely documented, is that…basically they turn into a conversation with the audience. The impact of this phenomenon is rather underestimated.

Interview date: March 3, 2010
Denver, Colorado

Clifford Lynch

Summary:

Lynch saw the challenges as divided in to 3 stages. The first problem was the period before CD-ROM got stable, then there was the period where you were doing things on CD-ROM for most projects, doing work station delivery, then there was the period when you were putting stuff on to the network. Lynch says you can think it as the 3 epochs. Pre-CD-ROM was horrible; there were no standards for anything. There were early forms of tiff for dealing with images, compression wasn’t well standardized, people were using some variation of Hoffman coding, at least for bi-tonal images, but it was pretty messy.

People couldn’t afford to use magnetic storage and there was an endless supply of crappy and unstable optical storage solutions some with the “extra bonus” of jukebox-like technology that constantly had mechanical and optical problems. Proprietary file formats were used, which were insufficient for the data. People wanted use multiple large mainframes that were a particularly bad mating for various reasons. People would wind up with a mini computer feeding larger mainframes. There were basically no metadata standards and zero standards for delivery. Lynch wound up building custom delivery things for different platforms that had a fairly short life because the platforms were unstable themselves. One of the things people tend to forget is the sufficiently large deployed base of technology in which the vendors still have to worry about compatibility. There was a tremendous churn in the industry and a great deal of discontinuity.

Once Lynch got to the CD world, rapid standardization in regards to image handling and Apple’s QuickTime began to occur. JPEG standards then became well established. Metadata standards, relevant for images other than things like art, cultural, and scientific objects…still faced an integration problem. There were not good platforms for building a system that combined text and imagery nicely. The early things, like the Voyager, had a certain hand crafted quality which were beautiful for these closed content systems, but many of these cultural memory institutions were interested in not just snapshots, but in growing collections.

People wanted to build up these network based servers and do delivery across the network. That was hard for a lot of reasons. Coming up with the right kind of architecture for distribution and how to function between various operating systems and picking the right kind of standards for the distribution of functions was difficult. It was quite a while before the network grew into the kind of promise that it immediately offered and you could sense.

Browsers of course were an absolute godsend for all kinds of imaging things because they provided a common platform. Prior to web browsers and things like Gopher, there was still a tendency to have custom clients for certain databases, something that really impeded the use and reuse of images in a really serious way.

Interview date:  March 3, 2010
Denver, Colorado

Clifford Lynch

Summary:

By background and training, Lynch is a computer scientist, although he has spent most of his working life dealing with issues around information retrieval, library automation, computer networking, and those kinds of things. Lynch has been involved with various phases of libraries since about the mid 70’s and has watched the evolution of technology as it has gotten less expensive and more accessible. Lynch’s working career began at the University of California in 1979 to around 1997. He did various things there, starting with the building of the online catalog, Melville, a catalog covering the UC system and the computer networks needed to support it. Lynch helped grow Melville from an online catalog to a multifaceted kind of information resource that held various kinds of primary and secondary materials.

Lynch left UC in 1997 to take his current position as director for the Coalition for Networked Information (CNI) where he sees a tremendous amount of digitization opportunities. However in his personal experience, Lynch has primarily dealt with the delivery of digitized page image material. He gives the example of the University of California initiative as a pioneer in a number of projects to take page images of scholarly journals and deliver them across the network to display terminals scattered around the UC system and beyond. Lynch saw this as a substantially technical undertaking, as it was the early 90’s, when all the publishers could give him was bitmap images instead of the more user friendly xml files and without the web browsers that became later available.

The other thing Lynch mentions, which he says he cannot take any credit for, other than doing a bit of consulting for them, was the [IMLS-funded] big image database that the New York Public Library launched about 4 or 5 years ago…consisting of about half a million images from their collection. It represented a fairly early effort of theirs to deal with a large and extremely heterogeneous collection of imagery.

Greg Crane

Interview date: February 8, 2010

Medford, Massachusetts and Centennial, Colorado

Summary: Crane’s primary focus is open content. One of the biggest changes he has noticed is the shift  of using multiple digital resources in modern work.

An innumerable amount of different resources are available for access. Crane sees a need for people to be able to analyze and transform the sources so it can be annotated and published in a database. Crane, too, sees a need for people who may be editing texts, semi-automatically, and creating name identification systems (like Crane did with London). Dealing with databases that are hidden behind subscription walls becomes crippling to scholarship because they cannot be downloaded reused.

There is a real shift in the paradigm of how resources are used and created today and how they were used years ago. Crane sees society at an incunabular phase. The first generation of digital work, with its new tools, is doing the same things that they have done before. There is access now, because of the rebuilding of the infrastructure of the humanities. Crane sees a need for people to help do what machines can automatically do: at the simplest level, correcting optical character recognition (OCR) or at a more complex level, taking the output of text mining and classification to try and create order and analyze the results. With Google Books on the horizon (that is, if they follow the settlement and the legality passes and it becomes accessible), everyone is going to access. And it will be a completely new world.

Interview date: February 8, 2010
Medford, Massachusetts and Centennial, Colorado

Greg Crane

Summary:

Crane’s involvement with digitizing goes back to 1982, when he was a graduate student and had access to the first chunks of electronic texts and magnetic tape. Crane remembers something like 60 megabytes of classical Greek texts that were typed in and available for use, flooding the storage available at the time. Harvard Arts and Sciences, where Crane was a graduate student, had two 60 megabyte drives that “looked like washing machines.” Crane remembers how getting any text at all was a big deal, even if it couldn’t be transcribed.

In 1983, Crane visited Xerox Park and saw his first digital color image projected and realized anything could be done; the opportunities were endless. Students of the ancient world have had to deal with information sources in a very disciplined way. Everything had to be analytic and the text, reconstructed. There was a contained set of linguistic data, studied very intensely, where one needed to come up with methods to find patterns in the text.

For the philologists (literary studies and scholarship linguists), computers were a natural extension of what they were trying to do. Crane’s field really started building resources in 1972 when they decided they wanted to digitize this corpus of Greek thus creating the Thesaurus Linguae Graecae (TLG). Crane began working on the project 10 years later, in 1982. It was clear to Crane that if you wanted to study the entire ancient world (the archeological, linguistic record all together) you couldn’t’ do it very well in print because you couldn’t get the data. Movement toward the digital environment allows for the possibility of having this accessible model. One of the key motivations of the project was to have an environment where everything was available. It was clear to Crane in 1983, that anything you could do in a normal library could be done better and with more clarity in a digital library. The Thesaurus Linguae Graecae made information sharing possible and easy to access.

Interview date: February 8, 2010
Medford, Massachusetts and Centennial, Colorado

Greg Crane

Summary:

Crane sees a lot of things that can now be done via the World Wide Web, but if the boundaries are pushed, problems arise and you run in to things that

cannot be done generically. Crane thinks that the people in Google are inexorably holding in on a lot of the core functions that create a modern intellectual life. Crane sees Google as more aggressive than the digital libraries have been; hebelieves that the cyber infrastructure will come out of the minds of Google.

Crane believes content is key and that we are driven by data. But content and services all had to be built from scratch. Remembering ten years ago, Crane was working on a London project where 40 books were put online. Crane spent a semester cleaning up these books by hand and extracting information. Now, with Google Books (or just the internet archive) you can instantly start data mining and working with them. All of that information is just right there, for access; it is incredible.

Greg Crane

Interview date: February 8, 2010
Medford, Massachusetts and Centennial, Colorado

Greg Crane is professor of Classical Studies at Tufts University and editor-in-chief of the Perseus Digital Library. Crane’s first exposure to digital technology in the humanities came when he was still a graduate student at Harvard, and developed a full-text retrieval system for the Thesaurus Linguae Gracae.  From 1998 through 2006, Crane directed a grant from the Digital Library Initiative (DLI) studying problems related to digital libraries for the humanities. Under the DLI-2 program, he worked on a range of topics, varying from the city of London to the history of mechanics to the American Civil War. Each collection provided new insight to electronic tools for learning and scholarship.

With the rise of the Google Books project in 2004, Crane began to focus on the problems and opportunities that arise when whole libraries (rather than curated collections) become available online. The broad range of projects that he undertook with support of the  DLI-2 program, the Institute for Museum and Library Services (IMLS), and the Mellon Foundation provided a strong foundation within which Crane could frame his current generation of research projects on Classical Studies at Perseus. Crane oversees the overall research program at the Perseus Digital library and is especially interested in finding ways the emerging cyberinfrastructure can serve the needs of the humanities and classical studies.

Interview date: February 8, 2010
Medford, Massachusetts and Centennial, Colorado

Greg Crane

Summary:

From the start, the Thesaurus Linguae Graecae designed all the data with concern for preservation and sustainability. The first data discs didn’t look very good and Crane and his colleagues knew it shouldn’t be made into their archival medium. In the 1980’s, it took years to get things encoded. Crane remembers thinking how he could have released the data much earlier had different technology been used. At times, it seemed like a wasteful investment and expenditure.

David Smith, a computer science professor at UMASS/Amherst was able to get the Thesaurus Linguae Graecae on the web without more funding.  The basic services of Perseus emerged in 1994. They were selling CD’s and the TLG was moving on the web, and it was clear that there was no future for the CDS. Because of the design of the data, thinking like libraries and librarians in terms of scale, the TLG was able to make this transformation and maintain collections for over 20 years. Crane saw the whole nature of scholarship change. The thing Crane didn’t appreciate 20 years ago that he thinks about now is that the TLG is a publication medium that you can use to make ideas more generally accessible. Ancient texts of Greek and Latin became more available through the linking of texts and dictionaries.

Crane did not appreciate the degree to which people are able to contribute (on sites like Wikipedia) and the dissemination of the power to create and participate in intellectual life. Crane did not anticipate the importance of this new phenomenon and its impact.