Table of Contents
It's difficult to write about the future of Internet publishing when it's changing literally day by day. But in the year that I've been thinking about and working on this book I've seen some trends that I think will continue. To state it simply, publishing on the Internet is becoming easier and more widespread much faster than I think anyone anticipated. Graphics are important, more so than people would like to think, but text and useful content are even more important. Fee-charging techniques are here and working, although much more is coming.
Computers do not solve problems, but people do, and computers are linking people together in ways never before imagined. Books and magazines won't go away. Computers, even the tiny ones, are not nearly as portable and inexpensive as books and magazines. If the theory that online versions will stimulate interest in buying print versions proves true, perhaps Project Gutenberg and others like it will stimulate a renewed interest in the classics as well as less well-known literary works.
This chapter is loosely divided into several sections, but it's important to read the sections you're not interested in. Although you might not think so at first, changes in protocols, software advancements, and more sophisticated encryption techniques are interrelated. Something that doesn't interest you may very well affect something that does. A new protocol for search and retrieval can open up commercial and business opportunities. Encryption techniques have social and political implications and vice versa. Software advancements can lead to new publishing techniques. In short, the implications are in the interconnections. Pay attention.
A natural and realistic concern is how well the Internet will hold up under all the attention and massive growth. U.S. government support for large sections of the U.S. Internet is already being phased out in favor of commercial support. Although loss of federal support may lead to higher access charges, it's not clear how much that will affect end users.
Whether the Internet is ready or not, the influx of new users and new hosts is tremendous. These users will affect all services currently available, and publicity can provoke wild distortions in the load on any given server. Consistently popular servers will often be unavailable or require constant retries to get in. Only faster hardware and greater bandwidth will solve this part of the problem.
Whether the growth comes from integrating the cable TV system and the Internet, from ATM (Asynchronous Transmission Mode), or from ISDN (Integrated Services Digital Network) access to home and office, the Internet pipeline is likely to get bigger and faster in the next few years--probably not big enough or fast enough to accommodate everyone and their multimedia dreams but considerably improved over today.
Archie, Lycos, and Veronica, among other centralized indexing services, are becoming overloaded. One alternative that's arising is commercial versions of these services, assuming that people will be willing to pay to get immediate service. However, Internet habitués hope that free alternatives will continue to multiply too. Multiple mirrors of the popular servers will help, a phenomenon already evident in Australia. Programmers already are trying to redesign the way indexes collect their information to spread the load across the system. Corporations looking to contribute to the Internet should consider investing in a few key services such as Veronica and Lycos.
We're now on the second generation of WWW indexing approaches, which are becoming more sophisticated. Developers are learning from their mistakes. One approach, I think, will be for user input to become an important element in adding to the depth and accuracy of the Internet indexes and subject directories. Lycos, for example, makes a strong step in that direction by allowing users to fill out an online form to remove a URL from the Lycos index.
InfoSeek <http://www.infoseek.com/> and Architext <http://www.atext.com/> both offer new searching technology. InfoSeek is a commercial Internet indexing service that indexes the Internet (WWW, Usenet News) as well as commercial databases. Architext was designed to do conceptual searches. The service uses statistical and analytical methods to turn up documents related to the words you used in your search, even if those words don't actually appear in the documents. These and other searching systems will accomplish Architext's primary goal, raising users' expectations for information retrieval.
When you think about how much has happened because a few people at different universities and institutes decided to write some software (Gopher at the University of Minnesota, Veronica at the University of Nevada at Reno, WWW at CERN, and Mosaic at NCSA at the University of Illinois), trying to predict what might happen in the next few years is foolish. But some interesting developments may give us an inkling.
But first let me deal with the competition between Gopher and WWW.
Even if everyone had a fast computer and speedy Internet link, Gopher would have a future. Gopher server software may not be that much easier to administer than a Web (HTTPD) server, but it is definitely simpler to put Gopher documents on the Internet than it is to write HTML and maintain links for Web documents. The irony here--and WWW zealots will loudly proclaim this--is that nothing prevents you from putting plain text documents on a Web server, just as you do on Gopher servers. But the killer is the expectation for WWW servers. WWW with HTML formatting can make online documents look so much nicer (to those with the right equipment) that it would be a waste to ignore that capability--and few do.
But that's just my opinion. Here are some others.
Michael Regoli, administrator of the CICA Windows FTP site, says:
I think that in terms of "bleeding edge" Internet technology, Gopher has indeed seen its day; this was predicted when I first saw the Web in action, some 12 to 18 months ago [as of April 1995]. Gopher, though, still has its users. It doesn't take big hardware to run it (on the client side), and I think it will continue to have a following, especially [among] those using limited computing resources (low-end 386s, 286s, etc.) and limited net access. Gopher is useful, don't get me wrong. Its beauty and usefulness lie in its simplicity. It'll be here for a while.
Paul Hoffman, president of Proper Publishing, an Internet consulting company <http://www.proper.com/>, believes that
Gopher will never go away, in the same sense that FTP hasn't gone away. They are not as 'good' as WWW for some, but they are much better for others. Gopher is inherently better for character-based browsers than WWW and will thus live on well into the next century.
Anyone who thinks that the whole world is going to graphical browsers doesn't understand the large corporate and the academic markets. Look at radio in the U.S.: it is far from dead 30 years after television reached a 90% penetration in the home.
There are no technical advantages to Gopher versus the Web, but that is irrelevant. What will keep Gopher alive is that it is easier to install and support Gopher clients on character-based systems than it is Web clients.
The University of Minnesota Gopher Development Team is also working on the interaction of Gopher and WWW. Paul Lindner presented a paper about their work at GopherCon95. <gopher://boombox.micro.umn.edu:70/00/gopher/Gopher_Conference_95/Papers/WebbedGopher>
Both Gopher and WWW browsers will be enhanced by the
development of plug-in viewers that handle documents with special
formatting. Some already exist and others are coming. Commercial
organizations are sponsoring some, and others are technical improvements
that may or may not be adopted as browser standards.
One essential element of software is the formatting of the data that are being manipulated. In this case the data are, for the most part, a combination of textual and graphic information. The following are some entrants in the competition to become the standard data format for the Internet:
The combination of two different and formerly competing security systems (SSL and S-HTTP) in Terisa's cooperative venture should go a long way toward establishing a system of secure communication across the Internet. Once a secure system is developed and becomes available for other companies to use, browsers with these new features will become available. At that point Internet commerce will become widespread, a development I expect by early 1996.
The Gopher team at the University of Minnesota is working on a three-dimensional virtual Gopherspace system that will offer a new way to browse Gopherspace. Unfortunately, it will require the browser to have a powerful machine, which runs contrary to Gopher's strength in being friendly to low-end PCs. <gopher://boombox.micro.umn.edu:70/00/gopher/Gopher_Conference_95/Papers/3DUnixServer>
The April 1995 WWW Conference in Darmstadt, Germany, saw the announcement of the availability of VRML, or Virtual Reality Markup Language for WWW. VRML apparently allows users of different Web browsers to interactively and cooperatively build a common virtual reality environment. The educational, social, and commercial applications (and implications) of this are just starting to be explored. One commercial example might be an auto company that simulates its new car in VRML and puts the simulation on the Internet for online suggestions and improvements. For more information on VRML, see <http://www.ncsa.uiuc.edu/General/VRML/VRMLHome.html>,<http://www.sdsc.edu/vrml>, and <http://vrml.wired.com/>.
One clear trend is toward improvement in number and quality of such WWW publishing tools as HTML editors and converters. Microsoft and Novell have already brought out free HTML authoring add-on packages for their word processors, Microsoft Word for Windows 6.0, and Novell's WordPerfect. These make it simple to save word-processing files in HTML format. Quarterdeck is planning a product that will work with those word processors as well. The efforts should quickly improve the functionality and ease of use of HTML editors. And as authoring tools become more prevalent and easy to use, the number of sites and amount of content will increase exponentially, especially if quality is maintained.
PDAs, or Personal Digital Assistants such Apple's Newton, are computers and therefore candidates for Internet browsers. One version of Mosaic already runs (in demonstration mode) on the Newton. That means it can display HTML files it has stored in memory but has no facility for easily browsing the Internet. With improvements in communications and wireless technology, it might be possible to look at the Southern California Real-Time Traffic Reporter Web site while you are driving in that traffic <http://www.scubed.com/caltrans/>.
URLs (or Uniform Resource Locators) should be familiar to you by now. They simplified life on the Net immeasurably, but a significant drawback to the URL is that moving a file or changing a host name can invalidate documents around the world that link to it. If the host specified in a URL is not available, or too busy, or has changed its name, that link won't work. Many people have thought hard and long about improving the URL scheme. The results include an alphabet soup of acronyms--URI (Uniform Resource Identifier), URN (Uniform Resource Name), URC (Uniform Resource Characteristic), and URA (Uniform Resource Agent).
URI (Uniform Resource Identifier). URI is a generic term for different methods of describing addresses on the Internet, including URL, URN, and URC. The original term, URL (Uniform Resource Locator), combines the name of a resource with its location. The URI working group of the Internet Engineering Task Force (IETF) is attempting to move WWW and other Internet services to a system by which a name (URN) is assigned to a resource that is described (by author, title, keyword) in a URC and can be found at various locations (URLs).
The IETF draft proposals on URIs are located at <http://www.ietf.cnri.reston.va.us/ids.by.wg/uri.html>. The archive site for the e-mail discussion of URIs is <http://www.acl.lanl.gov/URI/archive/uri-archive.index.html>.
URN (Uniform Resource Name). URNs provide a way to specify a resource without specifying the computer host (site) on which it resides. A URN server would take a given URN and look up the closest site that has that resource. That is, if a document proves extremely popular, many sites around the world might maintain a copy, and the URN server would list all of them.
URC (Uniform Resource Characteristic). URCs provide meta-information about documents and resources specified by using URLs and URNs. This meta-information might include owner, encoding, access restrictions, cost, language, and location. Having a clearly defined way to store information about a document can be crucial to being able to retrieve documents that match your search criteria. See <http://www.acl.lanl.gov/URI/urc_draft.txt> for the URC draft specification.
URA (Uniform Resource Agent). URAs refer to agents that would be programmed to work within our computers to do certain types of searches on the Internet. For example, a URA might be defined to do a "person search" using existing (free) Internet search services, such as Netfind <http://www.rpi.edu/Internet/Guides/decemj/itools/nir-utilities-netfind.html> and Whois <http://www.rpi.edu/Internet/Guides/decemj/itools/nir-utilities-whois.html> to find a long-lost friend who might be online. The user needs to know nothing more than that this URA is designed to find people. If in the future another type of person-searching service develops, scripts that interact with it could be added to the person-search URA. Bunyip Information Systems is developing a proof-of-concept (or demonstration) model of this, called the Internet Resource Discovery Client <http://services.bunyip.com:8000/products/client/client.html>, to show that it works. The link to the IETF working group's draft description of URAs is <http://services.bunyip.com:8000/products/client/draft-ietf-uri-ura-00.txt>.
Developing the URA is important because it could become a building block for more sophisticated tools that could make finding information on the Internet less of a hit-or-miss process. This area of software development would enhance the abilities of Internet clients like Netscape and Mosaic browsers.
Common Client Interface (CCI) is the converse of the already-established Common Gateway Interface (CGI) that provides for scripting or programming links behind the scenes on WWW (and now Gopher) servers. CCI would lead to programs that control Web clients. For example, you might have a program that starts your WWW browser 30 minutes before you get to work each morning and does a standard set of searches for news and information pertinent to your interests. The program would control the Web browser through the CCI. When you arrived at work, your computer would have a list of pertinent documents waiting for you.
Two new innovations from Netscape include Client-Pull and Server-Push scripting in HTML. Client-Pull allows WWW browsers to download an HTML document and then go back repeatedly (with no command from the user) to download a particular document, image, or resource from that same server. The inverse of this is Server-Push, which requires the HTTP server to keep the connection open and to send updates at certain intervals to the Web client. One application might be a stock market service that constantly updates stock prices and downloads them to the Web browser, with no action on the user's part.
Java is a new programming language developed over the last several years by Sun Microsystems as an improvement on the C++ language. Hot Java takes this language and puts it into a Web browser so that it understands (or interprets) programs written in the Java language. Once you have the two together (Hot Java in the browser and Java programs on the server), a Web publisher could create a special program (or "applet," as Sun calls them) in the Java language that would be immediately available to all Hot Java-compatible Web browsers. Hot Java is significant because it opens the door to an amazing amount of innovation and flexibility. For example, if someone wants to develop a utility for users of her Web site, she can write it in Java; users with a Hot Java-compliant browser could automatically download the program and run it on their machine. The obvious concern of controlling against viruses and other destructive software is said to be handled through extensive safeguards. Sun has released a beta version for X-Windows and is porting it to Macintosh and Windows. Hot Java is adding quite a bit of excitement to the WWW development world because of the opportunities it offers. Find out more about Hot Java at <http://www.sun.com>.
Silicon Graphics, Inc. (SGI), is starting to bundle Netscape's Commerce Server, WebMagic Author WYSIWYG HTML Editor, and an enhanced Netscape Web browser with a new line of SGI workstations. The bundle provides turnkey, or right off the shelf, entry to Web publishing. Makers of other platforms are likely to do the same almost immediately. This development could mean that publishing on the Web is as easy as browsing it, if you buy one of these new machines.
Bundling HTTP and other Internet server software with some workstations and PCs is a trend that signals that Internet publishing has arrived. As bundles become a standard feature, administering a Web site will only get easier.
Oracle, the leading database server company (for UNIX, Netware, Windows NT, and OS/2), is offering the World-Wide Web Interface Kit--at no charge. Oracle has a clear fix on the relationship between information stored in databases and publishing on the Internet. Its whole approach is to show how you can easily put your Oracle database on the Internet.
The goal of the Web Mail project, partially sponsored by the Open Software Foundation Research Institute, <http://www.osf.org/RI/>, is to seamlessly integrate mail processing and the Web browsing environment. The idea is to be able to access e-mail with all the normal e-mail functions--reply, delete, forward, and so on--while browsing the Web. In addition, the project also wants to devise a way to give you Web access to your store of archived e-mail without forcing you to convert it to HTML. One exciting result of this project might be to make it possible to control your e-mail archives through Web browsers, which are usually able only to read (or browse) archives. For more information on Web Mail see <http://riwww.osf.org:8001/www/webmail/>.
Latency refers to the interval between a request for a document and when you can retrieve it. Latency is a problem on the Web, and how to solve it is being researched. Simon Spero, a WWW researcher and developer, presented a paper on latency at the second International WWW Conference in Chicago. The basic complaint seems to be that because each document usually contains inline images, all of which are retrieved separately, telephone costs mount because of the delays resulting from the numerous reopenings of new connections to the server. If the images and documents all were sent together, those extra connections wouldn't be necessary. Spero's paper is located at <http://www.spyglass.com/techreport/six/developers_tech_doc6.html>.
Another paper, by Venkata N. Padmanabhan, of the University of California at Berkeley, and Jeffrey C. Mogul, of Digital Equipment Corporation's Western Research Laboratory, describes these problems, methods they devised for resolving them, and statistics showing how their methods improve speed. They suggest including in HTTP a GETALL function that would retrieve a document and all its inline images in one connection. If it works as billed, we can look forward to improvements in speed and access, even without improvements in Internet bandwidth. Their paper can be found at <http://www.spyglass.com/six/developers_tech_doc5.html>.
According to a company representative, WAIS, Inc., is considering making a free version of its $15,000 commercial WAIS server product available for downloading from its Web site <http://www.wais.com/> sometime in mid- to late 1995. It probably would be limited to indexing files of 10MB or less, but otherwise it would have all the features of the commercial WAIS server 2.1, which is Z39.50-V2 compliant. This would be a big improvement for the low-end Internet publisher who wants to do full-text searching.
Z39.50-1995 (version 3) was approved at the end of 1994 and should become final in 1995, hence the name. This means that an international protocol has been established for highly sophisticated information retrieval systems. Internet publishers who want to do really powerful search and retrieval should plan to become Z39.50-1995 compliant. For more information on Z39.50 applications and WWW to Z39.50 Gateways, see Table 5-1. The U.S. Library of Congress, AT&T, the British Library, Chemical Abstracts Service, European Space Agency, Knight-Ridder Information, Inc., LEXIS/NEXIS, and MIT Information Systems are developing or using Z39.50 in their applications.
Commerce on the Internet is alternately seen as its salvation and its destruction. Some have expressed the viewpoint that until commercial services and content arrive on the Internet, "nothing of value" will be available. That point of view ignores the cooperative work that's being done and the real virtue of the Internet, access to millions of people all over the world.
I think that commercial activity on the Internet will take two major forms. One form involves items or information being bought and sold, and the other involves items or information exchanged at no cost in indirect support of a company's business. Given the public relations, advertising, focus group, and customer support aspects of modern business, the Internet provides an excellent medium in which to conduct these activities.
Recent developments in the areas of digital signatures, digital watermarks, document coding, and digital cash are going to make it easier to do business on the Internet.
In March 1995 Utah passed a law making digital signatures legally binding, and the states of California and Washington are said to be considering similar legislation. This movement removes one hurdle for commerce on the Internet. The law in Utah is based on public key encryption and requires certification authorities to register with the state. The law is designed to allow the transmission of court documents.
Digital watermarks consist of an image or piece of text married to a graphic file in such a way that it can be used to identify the owner of the image. If the image is visible, it might deter unauthorized reproduction. If it is invisible until processed in a specific way, it could be used to detect unauthorized reproduction. Systems Research and Applications Corporation (SRA) of Arlington, Virginia, has applied for a patent on a technique it developed and calls Imprint(TM) that is being used by large online publishers of digital photos.
Imprint runs on a Macintosh and marries a company logo or other image file in PICT format to a TIFF image file. The TIFF image file can later be converted to other formats without losing the watermark. The intensity, or "strength," of the watermark can be varied from blatantly obvious to almost invisible. The watermark becomes apparent during color separation. Placement and the color of the watermark are two other variables in the Imprint software (see Figure 11-1). The technique can also be used on black and white images. For further information contact Anirudh Kulkarni, director, Media Systems Initiative, SRA, via e-mail at anirudh_kulkarni @sra.com or at 703-803-1883 (voice).
AT&T has been researching the encoding of small amounts of information (four binary digits) in formatted text. Formatted in this case means that the text is in either an image file or in a format description language such as PostScript, TeX, or troff. The encoding techniques include shifting lines of text, shifting the position of words, and altering the features of certain letters or fonts. AT&T's goal is to ensure that the encoded information is retrievable even after it has been photocopied several times. <http://www.research.att.com/#docmark>
Popularization and trust of digital cash and other electronic payment systems are essential for large-scale purchasing over the Internet. These payment systems are becoming available, and trust is starting to build. It's quite likely that 1996 will see the arrival of many more sites with something to sell and that they will accept electronic payment of some sort.
Subscribing to academic journals is becoming increasingly expensive, and many university libraries are unable to maintain the wide variety of subscriptions their faculty demands. As the financial stability of these journals weakens, online publishing will become increasingly attractive, as it has in the last few years in some disciplines. The desktop publishing technology many of these publications use is easily transferable to Internet publishing techniques. Many have small circulations, and they often can reach their audience more easily and less expensively through the Internet. The delay caused by peer review has led to extensive publishing of preprints in some academic disciplines. High energy physics is one field that has come to rely on Internet distribution of research papers. Other disciplines have taken note. The Journal of Medical Imaging is a demonstration of this concept. <http://www.umds.ac.uk/JMI/ejourn.html>
Seals of APproval (SOAP) is a system under discussion that online academic journals might use to give their "seal of approval" to particular journal articles or documents to show that they have passed a peer-review process. The actual seal would be some sort of digital signature using public key encryption.
Academic journals are not the only ones going online. 'Zines (short for electronic magazines) have exploded. Many are free, and some will experiment with charging for part or all of their issues.
Sometime in 1996 or shortly thereafter you will find it possible to move documents directly from your word processor to an Internet server format such as HTML, PDF, ASCII, or PostScript. Organizing the content and maintaining links will continue to be labor intensive, but that load will increasingly be spread among the actual producers of the documents as well as increasingly sophisticated data librarians.
Several U.S. newspapers, including the Los Angeles Times, the New York Times <http://nytimesFax.com/>, and the San Jose Mercury News <http://www.sjmercury.com/>, have begun to offer online editions of some sort. The Los Angeles Times puts its version on Prodigy. The New York Times offers an Internet version of the paper in Acrobat PDF. And the San Jose Mercury News offers news summaries and updates throughout the day as well as all its classified ads. Full news stories and archives are available for a $5 monthly fee. All offer hypertext links of some sort that can make reading a newspaper online a very different experience. Journalists already have access to a wealth of online information, and newspapers may link their online stories to related stories in their archives. The problem of rising newsprint prices also will affect what newspapers do. Because online publication means that stories could include video and sound, editors will have more options in stories they put online, although newspapers are still trying to learn how to make money this way. The New York Times online is free but carries advertising and requires that you register.
The rise of a completely new, extremely powerful, and far-reaching communication tool is bound to raise political and social issues. The international scope of the Internet adds to the implications.
Many parents, educators, and politicians are concerned about the lack of safeguards to what children may encounter on the Internet. Pornography, enticement by pedophiles, and bomb-making instructions are listed among the potential dangers to children using the Internet. Federal legislation that would make distribution of such material a crime was approved by the U.S. Senate in 1995 as an addendum to the telecommunications deregulation bill. The bill as written was so broad that enforcement would be difficult and threatened a huge censoring effect on Internet publishing. Yet it would have little effect on what comes to the United States over the Internet from other countries. Another problem inherent in such legislation is the application of moral standards from one area to another as the Internet makes geography less of a factor.
Another approach to safeguarding children is to develop restricted access systems that can reach only certain limited (pre-approved) sites or newsgroups. Although commercial servers will almost certainly take this approach, it requires a body that would approve sites for the rest of the Internet. The sites would be left with the problem of checking all their material--wading through an unimaginable amount of material and then filtering all new material the same way. The sheer volume of Usenet News prohibits this unless it's done with some sort of computerized filter. Even so, a site approved for children could add something unseemly later.
Someone may be able to develop a set of Internet tools for limiting access to a certain range of "approved" servers and refuse to link to servers that are not on the list. (This might work as the inverse of the security scheme common on servers that already restrict access to certain servers or Internet subdomains. A restricting mechanism might be built into clients.)
Some critics of these solutions reason that free speech brings with it risks and responsibilities. They argue that parents are the rightful monitors of what their children have access to. Part of the problem is what children might find accidentally; another concern is that children will actively search out forbidden material.
If limiting software is developed, organizations might offer a highly controlled (censored) subset of Internet resources to their users. This has its own risks, even if the censors are monitored.
The next few years will see many and varied solutions to these problems. One can only hope that the solutions allow access to all that is useful and educational on the Internet.
Another problem is who defines what is appropriate? Because almost anyone with Internet access and a computer can publish on the Internet, I think everyone will have to define his own standard. What people post and the items on other servers to which they link will reflect their standards and values and may or may not be appropriate for their audience. Perhaps it will become commonplace to state somewhere on a server exactly who the server is designed for.
The other immediate political problem on the Internet is encryption and who has access to it. Encryption technology is the key to many of the fee-charging systems in use or under development. However, a huge impediment is the U.S. government's ban on the exportation of this technology. The United States considers encryption technology a matter of national security. Although some efforts are being made to change this policy, other countries have even stricter laws concerning the use of encryption techniques. France has made use of unapproved encryption techniques illegal. Other governments are apt to have similar, if not more severe, restrictions.
The movement to make the Internet and its resources available to all is strong. In many cases that requires that a country first acquire sufficient technology and infrastructure to gain access and take advantage of the Internet. In the United States the debate centers on whether the government should license commercial Internet and telecommunications providers, requiring them to connect and provide a certain minimum level of service to all schools and libraries in their communities in exchange for licenses. None of these issues is easily resolved, and you should expect to hear a great deal of discussion about subsidized access.
Even as some people argue about censorship and whether to license commercial gateways to the Internet, others are worrying that the Internet, like television before it, will lead to the "dumbing down" of society. These critics point out that because abstracts are much easier to distribute and archive on the Internet than full text of articles and books, reading the abstract of an article will begin to replace the reading of the full text. But the popularity of Project Gutenberg and its massive effort to publish the full text of works of literature (described in Chapter 10) is eloquent testimony that the Internet is unlikely to carry television's baggage.
I think the Internet will increase the level of exposure to high culture, the cultures of other countries, and different points of view. My hope is that the profusion of varied points of view on the Internet will be greeted enthusiastically by the majority.
No one book can cover all the technical, philosophical, social, and legal aspects of publishing on the Internet. As I review what I have written, I find many more issues and techniques that I would like to discuss. But as my seven-year old daughter Riley said, "Just stop writing and tell them, 'And that's how it ends.'" Given the amazing rate at which Internet publishing techniques are evolving, and that your next step should be to explore these resources on the Internet, I'd have to paraphrase her. That's how it begins.