Overview, progress, and goals

by Adam Thornton
June 1994

Overview

The Galileo Project is a hypertext textbook that is being written for use in Professor Albert Van Helden's new course on Galileo to be taught in the Spring of 1995. The course will focus on Galileo's life, work, and social environment, and may eventually be integrated into the more general history of Early Modern science.

The text itself is being written by Professor Van Helden; images have been and are being acquired from various and sundry sources: scanned from books, downloaded from Internet ftp sites, or accessed via the World Wide Web. All of the increasingly sticky copyright issues, and much of the resource location and acquisition, is being handled by librarian Elizabeth Burr. Adam Thornton is providing the technical expertise--for the most part, knowledge of HTML[1] and Unix[2] tools--that furnishes the "glue" holding the project together. Two more students will be joining the project next year: Nell Warnes, who will be performing essentially the same functions as Adam Thornton, and Adam Lasics, an image manipulation specialist charged with taking scanned images and massaging them to professional quality. Martha Turner, currently studying in Israel, may also be recruited for the Galileo Project.

The Galileo Project is funded by a grant from the Council of Library Resources to Fondren Library.

Philosophy Behind the Galileo Project

This could, perhaps, be more succinctly phrased as "why are we doing this?" The short answer is not very informative: "serendipity". Expanding this theme, we find that an unusual combination of circumstances that happened to set off the right sparks at the right time, combined with a need for a core text for the History of Science courses.

The project also exists because of an admittedly radical belief I hold:

A cheaply and widely accessible global information network is to the printing press as the printing press is to the hand-illumination of manuscripts.

This dogma forms the ideological substructure of the Galileo Project; in the section on "Distribution of the Project", the implications will be made much clearer.

Chronology of the Galileo Project

In the spring of 1993, Van Helden had just finished teaching his modern history of science course, covering the period from 1700-1960, and was somewhat dissatisfied with it: simply put, the exponential expansion of knowledge since the Scientific Revolution makes it almost impossible to reasonably teach an overview of the history of science. The problem is less acute with ancient and medieval science, since the primary texts are--at least in translation--relatively accessible. By the modern period, each branch of science has developed its own unique language and primary sources are completely impenetrable to anyone not part of the scientific community that forms the intended audience. This makes teaching the history of science after Newton extremely difficult.

We still have come to no solution to this problem; however, Van Helden was also dissatisfied with his Early Modern course, which covered the period from 1400-1750. It is during this period that knowledge first begins to get out of control. However, while the volume of scientific material skyrockets, most of the texts are still accessible to the layman, as a specialized scientific language has not yet developed. This combination makes it an excellent period to teach, as one can cover the core texts (e.g. the Siderius Nuncius and the De Revolutionibus), give a reasonably detailed overview of the scientific topography of the period, and still leave plenty of nooks and crannies for students to do their own exploring in.

However, no particularly good introductory overview of the beginnings of the Scientific Revolution exists. While coming up with excerpts from primary documents for students to read is not terribly difficult, finding a coherent text to present this material in a basic conceptual framework is. While writing an entire such text might be difficult, writing the text for a course specifically about Galileo seemed a reasonable project for a year of sabbatical.

Van Helden discussed these concerns in late spring and early summer with me. At that time, Rice had just acquired a new Vice President of Information Systems, Tony Gorry, who had proclaimed the "Year of the Network", and had making funds available to upgrade network connectivity on the campus and to promote the use of computers in the humanities. I speculated that, rather than a traditional text, perhaps what was needed was a hypertext. The idea of teaching via a network-accessible text stems from my belief that the exponential spread of the Internet is the wavefront of a paradigm shift of the same type and order of magnitude that the printing press represented, as well as my conviction that the nonlinear nature of hypertext makes possible a much more individually tailored way to present information than a book.

The idea of writing a hypertext "textbook" bounced back and forth between us for a few months, gathering momentum and a life of its own as it went; because this would be a trial balloon for the use of computing in the humanities--and thus central to the Electronic Studio project, which the Galileo Project became part of--funding was relatively easy to secure. The summer was spent learning the limits of the various systems available to us, settling on a choice of platforms, and evolving a basic structure for the document we would produce.

In the fall, Elizabeth Burr joined the team and work began in earnest; we began writing and incorporating links and images, worrying about copyright issues, and brainstorming for alternate organizations of our information. By the time of writing, we have a fair number of links in place, a basic structure well-defined, and a few images and maps. At the moment, the Galileo Project is basically at a proof-of-concept phase: for most of the areas we wanted, we have working examples that really only need to be further expanded to be complete; it is only in a few areas--such as inclusion of a bibliography--that work has not yet begun.

The document is intended to be complete enough that a course can be taught using it by January of 1995. This is not meant to suggest that it will be finished--one of the most exciting things about the Project is that it is almost infinitely extensible--but it should be fully usable and very useful. We feel this is a realistic goal: many of the difficult design choices are, I think, behind us. What lies ahead is mostly implementation, which will not be trivial, but should be fairly straightforward.

Hypertext

What does "Hypertext Textbook" mean?

This is best answered by analyzing the terms one at a time. Hypertext is simply nonlinear text. The concept takes a little getting used to, as all traditional text is overwhelmingly linear. What is meant to be suggested is just that there is no one correct path through the text; this will be much more fully discussed below.

For our purposes, all "textbook" means is the central non-primary source material for use in teaching a class. Therefore, a "hypertext textbook" is a non-linear document intended for use in a classroom.

Linear vs. Nonlinear Texts

The distinction between plain text and hypertext will be much clearer if we first examine the similarities and differences between linear and nonlinear media. This should also provide insight into why we feel that hypertext is the appropriate approach for the Galileo Project. Initially I will discuss only the relative merits of text and hypertext; later I will move to discussion of non-text (graphical images, sound, et al.) objects imbedded in hypermedia.

Linear Text

Traditional texts are linear: the reader is supposed to read them in a continuous fashion, starting at the first page and reading through until he or she runs out of book. In, for example, a work of fiction, there is usually a linear progression of action: the characters and situation are introduced, an element of conflict is presented, the plot develops around the conflict, some sort of climax occurs, and the work concludes. In a language textbook, a very basic lexicon is developed and the fundamentals of grammar built--perhaps the present tense and simple sentences are presented. Then the student progresses to more complex grammatical constructions: interrogative and imperative sentences, for example, and a variety of useful adjectives. Eventually, the sphere of student knowledge will have expanded to encompass most of the language. This is developed in a linear fashion, where each chapter builds on the material learned in the previous ones.

There are advantages to linear media. One of the biggest points in its favor is that it presents an extremely familiar and comfortable user interface. By the time a student gets to Rice, it can be assumed that he or she is not only highly competent at reading, but that he[3] probably derives some pleasure from it as well. Books are, almost by definition, structured linearly, and have been as long as writing has existed: no new skills need to be learned to read a book.

A further advantage of linear media is that the information is usually highly structured. The author probably has a good reason for presenting the material in the order in which he chooses to do so. For works that cumulatively build on their earlier parts--physics or language textbooks, philosophical tracts, or novels, for example--this is extremely important.

However, there are drawbacks as well to linear text. The first is the converse of the second advantage above: the order in which an author structures a work may not be the order in which it would make the most sense to a reader. Imagine, for example, a discussion of forms of governments in the Classical Mediterranean basin: the author might present it in a geographical way, progressing from North to South, tracing the entire history of each region before moving on to the next. A reader might well find it more useful to read it chronologically: a snapshot of each of the various governments at 500 BC, and a similar picture for each century. A linear medium would prevent him from doing so.

Another problem with traditional text is pointing to additional references. Typically, a book will have a bibliography; the interested reader is then encouraged to go to the library, find the referenced book that interests him, and read it. This is not ineffective, but not all readers are able to have high-quality libraries at their disposal. Even if a reader does, the book he wants may not be in the collection or may be in use by another patron. In any event, he has to expend a significant amount of effort to get to any of the secondary texts. This is, as we shall see, not necessarily the case with hypertext.

Nonlinear Text

There are four major advantages I find in hypertext as compared with linear text. First, the reader decides in what order the material is to be presented. This was discussed above, and while it is inappropriate for some applications, for many it is best to let the reader choose the order in which the information appears.

Second, the bibliography can be active rather than passive. When another source is referenced, the reader does not necessarily have to search for it. Provided the material is also on line, the text can be directly linked into the document. This process can be automated to provide for the automatic maintenance of an up-to-date concordance or index without human intervention.

Third, hypertext is a superset of linear text. There is no reason whatsoever that a hypertext cannot contain passages of linear text--in fact, most do include significant amounts of linear text. Thus, nothing is lost by moving from a linear medium to hypertext.

Finally, and perhaps most importantly, the information does not have to all fit into one book or one series of books. There is no longer a need to put interesting but only marginally relevant material into the text when a link will suffice; size can be arbitrary since there is not necessarily an expectation that the entire document must be digested. If a networked approach, such as the Galileo Project employs, is taken, then even geography becomes irrelevant. Part of the document can exist on a computer disk in Houston and part in Florence; the user need never know where the information he uses physically resides.

There are also some problems with hypertext. The first is directly related to the lack of well-defined structure traditionally imposed by the author. In short, it's easy to get lost; if no structure is given, it is easy for the hypertext to become a chaotic mess of randomly--and tenuously--linked documents.

Hypertext tends to lead to an "everything, the kitchen sink, and your little dog Toto too" approach; developers get so entranced with the freedom to include links to anything and everything that they do so with wild abandon. Thus much of the material found in a traditional hypertext application is often completely irrelevant. There is also a tendency to emphasize form over content--particularly with hypermedia that includes non-textual elements--and thus to throw in a lot of things that look neat or that seem flashy, but do not in fact contribute much of anything to the document itself.

Finally, hypertext presents a very different user interface than the traditional book. A reader will have to learn how to read a hypertext. Since hypertext presentation methods have not had the benefit of half a millennium of printing technology to standardize an interface, a reader may very well have to learn new skills for each hypertext he reads. This learning curve may be steep and certainly functions as a deterrent. Even if the document is interesting, it will, at least initially, be much easier to simply read a book.

When is Hypertext Appropriate?

The easiest way to answer this question is to first show when it is inappropriate. As mentioned above, novels (James Joyce and Thomas Pynchon notwithstanding) are generally poor venues for hypertext. Any subject in which knowledge must build cumulatively is not suitable for hypertext--this includes science and language textbooks, chronologically-oriented material, and recipes.

Hypertext is appropriate for interactive fiction. This includes traditional "text adventure" games, such as Zork or The Horror Of Rylvania: even though these often have a linear plot, in which the player must perform certain actions in a certain order, the order in which he travels between locations in the game or the order in which he picks up items is his to control. A reference manual is a good candidate for hypertext: if it is likely that the user will simply want to look at the section that he needs to consult, and then be able to instantaneously bring up related topics, then this is precisely what hypertext is good for. In general, hypertext works best in a text with many unordered but related topics. If I am consulting on-line documentation, I should not have to read through commands instructing me how to print the contents of a file to get to the section on formatting disks; however, when I get there, I want to be informed that a link exists to the commands that check whether a disk is damaged or not, since that is logically related to the process of formatting a disk.

Why do we feel hypertext is appropriate for the Galileo Project?

The way the Galileo Project is structured is highly topical, but does not usually need to be read in any particular order. It really makes no difference to the reader's comprehension of the subject whether he reads about Copernicus's geocentric universe before or after he finds out about seventeenth-century Florentine politics; thus, we want to give him the option to order his reading as he pleases. For those sections that do require sequential comprehension--the timeline of Galileo's life, for instance--we can make the timeline a single linear text with links branching out from it. One of the great strengths of hypertext is that it can include as much linear text as desired.

We also have wanted Galileo to be easily extensible; rather than doing this with a huge bibliography (although we certainly will include an appropriate bibliography), we can include links to hypertexts about other subjects. For instance, in the section about telescopic astronomy in the seventeenth century, we can include a link a network resource about current developments in astronomy, a link to a presentation on the architecture of Renaissance Florence, or to material about the lives of other scientists.

Digital Hypermedia

The medium we have chosen is not limited to text; it is a digital medium, and anything that can be represented in bit patterns can be represented here. Very nearly everything can be represented as bit patterns.

Bits Are Bits

Books are limited to words and pictures. Occasionally a book will be packaged with a cassette tape (e.g. the hardback version of Ursula K. Le Guin's Always Coming Home), but this usually smacks of a promotional gimmick and is extremely rarely an integral part of the text.

There is no such restriction in hypermedia. We can have text, images, sounds, animation clips, and interactive simulations. These can be bound as tightly to the text as we care to design; we can even make them a necessary part of our--now more broadly defined--hypertext.

Media in digital format take up essentially no space. The compact disc, which is a technology ten years old, can pack roughly 600 megabytes onto one side of a 5-1/4 inch platter. If this were represented as text, it would use up about 150,000 pages. Images have a much higher information density, but still much lower than that which can be routinely expected from a digital format. Finally, digital texts are more durable than paper texts. Paper becomes brittle after 50 years or so; while magnetic tape should be recopied roughly every 25 years, the labor involved in duplicating it is much less than that involved in reproducing a book. The lifespan of a compact disc, barring catastrophe such as melting or breaking, and assuming proper storage, can probably be measured in millennia[4].

Implementation of the Galileo Project

Choice of Platform

Minimal System Requirements

As with anything else in the computer world, a trade-off existed between accessibility and performance. We needed a system that would be capable of displaying crisp graphics, but not something that would be more expensive than students could be expected to have or at least to have access to.

It was fairly easy to decide that the base system ought to be a color computer capable of displaying at least 640x480 pixels with an eight-bit (256 colors) color depth. Most color Macintoshes and PCs sold in the last three years meet this requirement: few of the personal computers on campus fall short. Many of the PCs in Mudd, however, fall short, as they are only capable of 320x200x256. Nonetheless, this is really the minimum screen size necessary for decent image quality.

Distribution of the Project

The scope of the project was obviously too great for distribution on floppy disks: it would be both expensive and clumsy to distribute 20 or 30 diskettes to each student in a course. A CD-ROM is more manageable, but would have been expensive to produce, and would not have allowed customization of the material. Therefore, the most natural approach was to make the Galileo Project available over the campus network. This seems immediately to exclude reaching it from student rooms; at the moment this is mostly true. However, in the next round of campus renovations, individual college rooms probably will be wired. The Electronic Studio project already seeks to provide course materials via the campus network, and therefore, the Galileo Project can and should exist under its auspices.

Most importantly, a networked approach means that we are no longer restricted to Rice. The Galileo Project can access resources literally halfway around the world via the Internet. Conversely, we eventually hope to be able to open up the text to the entire Internet community: an interested student in Sri Lanka would then be able to read the documents in the Project just as easily as someone sitting at a terminal in Mudd. The most marvelous thing about the Internet is that geography finally becomes truly irrelevant. Communities are defined by interest and language rather than by location. I am convinced that the Galileo Project is at the leading edge of the new educational methods that will be explored as global network connectivity grows cheap enough that it becomes like telephone connectivity: an assumed and necessary resource.

Interoperability

Another primary concern was that the Project should run on different kinds of hardware; while Mudd is predominantly Macs and Sun workstations, a great many students use PCs at home; ideally, we wanted a program that would run on Macintoshes, PC clones running Windows or Win/OS2, and Unix Workstations running the X Window System. This would allow us access to all of the major graphical computing environments on campus. This design choice ruled out Hypercard and Toolbook; while both are excellent systems, each is very platform-specific. While tools exist to migrate Hypercard stacks to Toolbook presentations and back, this still leaves Unix workstations out in the cold.

After deliberation, we settled on the HyperText Markup Language (HTML) as our format, with Mosaic the preferred viewer[5]. Mosaic is available from the National Center for Supercomputing Applications and is free. Versions currently exist for Windows, the Macintosh, and Unix with the X Window System.

HTML

HTML has its own particular advantages and disadvantages. One of its strengths is that it includes a very general mechanism to make available files anywhere on the World Wide Web (WWW). The World Wide Web is loosely defined as the set of all computers on the Internet which run a program that makes available resources via a certain communications protocol, called http, for "HyperText Transport Protocol".

HTML has evolved from IBM's Standard Generalized Markup Language (SGML), which is a language designed precisely for formatting and presenting text, so complex text formatting is extremely easy. Since it is basically intralinear annotations in a text, it is easy to write and modify with standard text editors; it also can be easily sent via e-mail. How HTML is displayed is left entirely to the discretion of the program used to view it. This means that the same file can be read by Mosaic, a full-featured graphical interface, or by Lynx, a very basic, text-only, interface to the hypertext; while Mosaic will only run on Windows-based IBM clones, Macintoshes, or Unix systems with the X Window System, Lynx can be compiled and run on any platform that has a C compiler and the curses libraries[6], which should include practically any computer likely to be encountered in a university environment.

HTML does, however, have its drawbacks. Tables are not yet well supported, and creating maps in which a region is clickable is not as easy as it should be. Further, the viewer-dependent display means that the interface is inherently inconsistent: accessing an HTML document via Lynx will be completely different from accessing it via Mosaic.

Mosaic

Mosaic will be our primary viewer, since it, unlike Lynx, supports all of our requirements. It is available via ftp from the site ftp.ncsa.uiuc.edu[7], and is absolutely free. This is perhaps the strongest point in its favor: since it and its source code are freely available, its popularity is immense. It has migrated to a plethora of Unix platforms and has become an accepted standard on the Internet.

Text presented through Mosaic looks wonderful; it's generally nicely formatted and very legible. However, the only way to explicitly control formatting at the moment is to use preformatted text, which is much uglier: it is in Courier[8] font and line breaks are no longer automatic. As long as one is content with the default formatting--which is more than acceptable most of the time--Mosaic does fine.

To do anything besides displaying inline GIF[9] images and HTML-formatted text, Mosaic relies on helper applications. This is another point in its favor, as it simply makes use of the tools appropriate for the platform on which it runs: images are displayed using the program xv on Unix, and by JpegView on the Mac; audio files can be run through /dev/audio on Unix boxes or played through a SoundBlaster under Windows. It is the responsibility of the "helper applications" wherever Mosaic is installed to handle other file formats. This allows the Mosaic developers to concentrate on HTML interpretation and leave the bells and whistles to established programs.

Mosaic has its own share of problems as well. The reliance on default text formatting makes it impossible to create a document with the same degree of control as in a desktop publishing program; the user can always resize the window and have the text flow to fit it in Mosaic. Mosaic is not intended to be a collaborative tool. It is difficult for multiple users to change the same document at the same time, although much can be achieved with local annotation. It is hard to implement action links well in Mosaic. Although it is fairly simple to state a process of execution that certain actions within Mosaic can invoke, it is difficult to do this without compromising the security of the system. The capability for user authentication completely isolated from Unix security could presumably be added to Mosaic to get around this problem, but it would be difficult and perhaps not very useful to do so.

Our Approach

The "textbook" will be a read-only HTML structure. Students can then access the text through Mosaic or Lynx. With this approach we can ultimately make the HTML text accessible to the entire world merely by allowing off-campus access to the document.

Structure of the Project

At the moment, we have three major paradigms of organization, with two more proposed or under development. Galileo's Villa, a geographic metaphor; the Timeline of Galileo's Life, a straightforward chronology; and the Table of Contents Of Links, which is a simple subject-based hypertext that contains pointers to our various resources. All of these point to the same links; it is only the manner in which access to the links is organized that changes. The Bibliography is under development, and an Index, searchable by keyword, has been proposed and will probably eventually be implemented.

The Links

"Links" are the central feature of the Galileo Project. Although they sound mysterious, they are, in fact, quite straightforward. A link is simply a short piece of text--typically 250-1000 words--written by Van Helden, translated to HTML, and put into the document. A link may also be an image, a map, or some other object we want to include in the hypertext. Links are, to a large degree, given meaning by the way they are fit into the superstructure; the framework in which we arrange them, while loose, does give them a certain necessary degree of context.

Galileo's Villa

The Domus Galileanis is a digitized floor plan of a seventeenth-century Florentine Villa[10]. Each room has its own function: for example, one is the Chapel, one the Portrait Gallery, one the Observatory, one the Laboratory. The user can click the mouse inside a room (or select a room name from a list beneath the Villa map, so that text-only users are still able to get somewhere from the Villa), and be presented with the "contents" of that room. Thus, in the Chapel, one might find links to Cardinal Bellarmine, Pope Urban VIII, the Spanish Inquisition, and the Protestant Reformation. The Observatory contains links to Kepler, Tycho Brahe, and comets; soon it will also have a flip-book animation made of Galileo's sunspot sketches and a short history of telescopic astronomy in the seventeenth century.

A link can be referenced from many points in the Villa; Urban VIII, for instance, is equally at home in the Hall of Patronage and in the Chapel. This document will probably itself be found by clicking on the cellar stairs, showing that it details the "substructure" that tells about the design decisions made during the Project's implementation.

Timelines

Currently, there exists a Timeline of Galileo's Life; it, and the more general European Timeline (which covers the period from 1450 to 1700), are linear chronologies which include links to the topics mentioned within themselves. Since internal reference points can be embedded within a document, it has also been possible to put pointers back from the various links to their first occurrence in the timeline. These serve as a good example of how even a fundamentally linear text can be used to good advantage within a hypertext document.

Table Of Contents

This is exactly what it sounds like: a list of links, grouped by type. Selecting any of them will bring up that link. It's handy if a reader knows what he wants to read about and doesn't want to have to figure out where that link can be found in its context.

Index

The index does not yet exist. What I would like to do is to maintain a list of keywords, each of which has a connection to a page that lists links that relate to the keyword chosen. Prentiss Riddle is currently working on the search engine that would make this scheme feasible. The index would be somewhat tedious to implement, and would require a lot of testing to get even a majority of the keywords people wanted to search by, but would also be immensely useful. I feel that the utility of the index probably outweighs the drudgery of its creation.

Bibliography

The bibliography will be a single large document, formatted to conform as nearly as possible within the constraints of HTML to the Chicago Manual of Style's format. Each entry will be marked with its own internal reference so that clicking on any mention of a referenced book within any of the links will bring the reader to the appropriate entry in the bibliography. This is extremely easy to do with HTML, but will prove very time-consuming.

Other Formats

We will have several maps, with regions defined on them such that by selecting a region the appropriate link can be activated. For example, in our map of Italy, clicking in the north should bring up the link about Florence and Tuscany. These maps can be on as large a scale as Europe or as small as Galileo's Villa.

Eventually Galileo's notebooks showing the motion of Jupiter's moons and of sunspots will be made into short animation clips. The effect should be rather like a flip-book in giving the illusion of motion, and will provide an interesting graphic representation of the actual motion of sunspots. This is a technique that simply does not work in a printed book; one must be content to print each picture, and then--if one so chooses--instruct the reader to flip through very quickly. However, by implementing this as an MPEG[11] file, we get to choose the frame rate and ensure that the "pages are flipped" at an appropriate speed.

Vincenzo Galileo was a musician. There is no reason that--should we think it relevant--we cannot take digitized sounds of the instruments he would have played and include those in the Project, although not all of the workstations currently have sound capability. We can also do visual and aural representations of classical physical demonstrations--such as the harmonics of a taut wire--very easily with these techniques.

Galileo in the Classroom

How will this work?

The Project will replace the course textbook, but will assuredly not replace primary sources. Most of the class will probably still revolve around reading translations of actual documents, such as the Siderius Nuncius. I envision handing out a syllabus containing a "core" of must-read links for a given day's discussion, and having the other ones be optional reading. It will not be necessary that all links be hit during the course of the semester.

One of the great things about the HTML format is that it is very simple to translate a document from a popular word processor format into HTML. Thus, extremely good student papers can be rewritten, turned into HTML, and made links within the course text. This will allow the Project to grow synergistically, especially as it starts to overlap with other courses taught using the same technologies[12].

Will it work?

The proof of the pudding is in the eating. Those of us working on the Galileo Project think it's an extremely exciting experiment. However, while we all feel it's a really neat way to teach history, it will ultimately be the success or failure of the course with the students that determines whether or not it works. This may be hard to determine; the students' necessity to learn a new interface before reading the course text will be a strike against us, as they may reject the system before ever coming to grips with the material itself on the grounds that the interface is unreasonably difficult.

This has never been tried before. Standard textbooks are simply not very good in this area: the Galileo Project should have at least as much cogent text as the introductory History of Science textbooks it is trying to replace. I think that the format has two great advantages over paper text: first, it allows the student to approach the material in the order he finds most useful. Second, and at least as important, is that this allows extension and improvement on the course text. There is no need to wait for the next edition from the publisher, when publishing a new version is as simple as editing a file; further, if either a student or the professor writes a good paper on material relevant to the course, that document can become a link in the course text[13].

Finally, we must recognize that globally networked information is the direction in which education seems to be inexorably heading. I find this a great--though not an unmixed--good. Copyright problems are among the first bugbears to appear as traditional notions of "publishing" are challenged by a communications infrastructure that makes "store-and-forward" the only reasonable means of transmission, and anyone with a laser printer is his own printing press. They will not be the last. As the distribution of information shifts from being paper-based to being electron-based, a great many old and valued paradigms will crumble. I believe that, eventually[14], many if not most new "texts" will resemble more closely the Galileo Project than the traditional book. It is and will remain exciting to be at the leading edge of the transition. The Galileo Project is a radical experiment; its success is far from assured. The eventual success of projects like it is almost certainly unstoppable.

Notes:
[1] HyperText Markup Language, about which more later.
[2] Unix is a trademark of AT&T Bell Laboratories. AT&T is the first four characters of a modem test string.
[3] Hereafter, when "he" is written, accept it as notational shorthand for "he or she".
[4] The more interesting question is: will anything besides historians' tools be able to read compact discs in a century? The same applies to magnetic tape: realistically, tapes are not copied because the medium begins to age but because the old magnetic tape format has become obsolete and, unless it is copied onto a newer medium, shortly nothing will be able to read it.
[5] Students accessing the Galileo Project via modem will probably prefer to use Lynx, however, to speed up the process. While Mosaic can be used over a modem, the inclusion of images makes it painfully slow.
[6] A popular screen-handling package. You are not expected either to understand it or to care about it.
[7] If you don't know what this means, and you want Mosaic, find a beginner's guide to the Internet. There are several out there.
[8] This is Courier. The typewriter look is out.
[9] Graphics Interchange Format; a very popular picture format.
[10] No, it's not really one of the houses Galileo lived in. Hopefully someday we can get such a plan and make it into a map, though.
[11] A common animation format. I don't know what it stands for.
[12] There's no reason this shouldn't happen, if the first year is a success and other history or philosophy courses adopt a similar approach.
[13] This does indeed suggest that publishers may not be crucial to the textbook publishing business.
[14] Probably not within twenty years; probably within one hundred.