© 1996 Craig Feied and Mark Smith
The Internet
What is the Internet? The internet has been variously described as 'a self-healing communications web', 'a community of communities', 'a piece of cyberspace', and 'a nice place to live'. The Internet is a series of local networks and computers along with the wires, fiberoptic links, and satellite channels that connect them together. More to the point, the Internet is the collection of information & communication services available on these interconnected computer networks that now span the globe.
To understand how you fit into the Internet we need to look at how it's constructed, and how it differs from other networks we know. Most of the connections we make in day-to-day life are direct one-on-one connections, such as a telephone call or a fax transfer, a bicycle courier or a Pony express rider. When you pick up your cellular telephone, a dedicated connection is made between your telephone and some radio receiver. When you drive into another 'cell', your call is 'handed off' to another receiver, but your connection is still point-to-point. Even a conference call is basically a dedicated set of one-on-one connections, where one end of all the connections is shared by an operator. We think of the telephone system as a large network, and indeed it is one. A switched point-to-point network.
The US mail delivery system is a little closer to being a different type of network, called a 'packet-based' network. A letter going from LA to New York will travel with a lot of other similarly addressed letters on a series of planes, trains and trucks. The particular pathway taken by a particular sack of mail depends upon many different factors. Price per mile is one of the factors, and transit time is another, but if the NY airport is closed due to poor weather conditions, mail sacks will land in Washington DC and travel by train to New York, even though it's both slower and more expensive.
My parents live 20 miles from the nearest village on a Greek island called Zakynthos. In Zakynthos they REALLY have a packet mail system. When a letter is sent to my parents, if it looks routine then it sits at the post office until they come in to get it on their weekly visit to town. If a letter looks important, though, the postal clerk gives it to a passing peasant who happens to live on their side of the island. The letter gets passed from hand to hand, gradually working its way closer to my parents' house. After a few days, a dusty peasant on a donkey will flag down my parents in their little jeep, reach inside his shirt, and pull forth the sweaty, crumpled missive.
Like the Zakynthos postal system, the Internet is a 'packet-based' network. The nodes on the network are redundantly interconnected in a huge web. Each node on the web has an address, and all of the nodes can talk amongst themselves to find the most efficient currently open pathway from one site to another.
To send a message to somebody on the Internet, you simply put your message in a address packet as though you were putting a letter in an addressed envelope. Then you pass the packet to your nearest neighbor in the direction which you believe is the shortest route to its destination. If your message is too big to fit into a packet, you break the message into pieces, put each piece in a separately addressed, numbered packet, and send them out separately. If one node along your optimum pathway gets busy or breaks down, or if two pathways are equally efficient, then some of your packets will go by a different path than others. They may not get there in the right order, either, but software at the receiving end can read the identification codes on the packets and can reassemble the contents to recreate the original message. If a packet doesn't arrive, or arrives garbled, the recipient sends a message to the sender asking to have the packet re-sent.
Some of the advantages of such a system are obvious. The original designers of this system were trying to achieve a military communications system that would continue to function in the event of a nuclear attack. This design is self-healing because it is self-routing and because there are redundant pathways. There is no central switching station, and no central authority controls the routing pathways. If a segment stops working, its nearest neigbors stop getting replies from it, and they pass that information back to their own neighbors. Messages automatically get passed along alternate pathways, and over a period of time the system will find the most efficient alternate pathways that are still functioning.
Some of the benefits of this web-like structure are not so obvious until you stop to think deeply. For example, it is exceedingly easy to repair this web or to extend it. To become an active node, all you have to do is convince the administrator of another node to let you connect your computers together.
The connection can be by satellite, by fiber-optic cable, by coaxial copper cable, by twisted pair, by telephone, or even by radiotelephone (cellular or otherwise).
Table 1: Connections to the Internet
1960 4
1961 25
1962 40
1963 55
1964 70
1982 213
1985 600
1990 130,000
1994 2 million computers
50 million users
There's a whole subsection of the Internet, called FidoNet, that is made up of people whose personal computers dial up other personal computers late at night, when everybody has gone to sleep. When two Fidonet computers talk, they exchange mail packets and updated information about the cuurent state of the web. Each one then dials up as many other machines as necessary to pass along any mail that belongs somewhere else. In theory a mail packet can pass all the way from Washington DC to San Francisco TOLL FREE, with Fidonet computers making thousands of local calls forever heading west. In reality, many FidoNet computers are also connected to the Internet, so most Fidonet mail today makes only one pair of local hops to get into the internet and a second pair of local hops out at the other end.
Another interesting point that's only apparent on close consideration is that the Internet automatically regards local censorship of information as a 'broken area', and self-heals by routing censored packets around the blight zone. Global censorship on the Internet is quite impossible today, because there are millions of linkages and because each linkage is under strictly local control.
Conversation overheard at a public access computer center:
- "Who owns Internet?"
- "Nobody."
- "What do you mean, nobody?"
- "Nobody owns the Internet."
- "Well, then, who just paid for my message to Hong Kong?"
Internet Growth
Usage of the Internet is growing at a phenomenal rate. In 1960 there were exactly 4 computers connected together on the forerunner of the Internet. For the first decade or so there were about a hundred and fifty new machines connected to the growing network per year (Table 1). Today, about 2 million computers around the world are physically connected together via the Internet, and more than 50 million people can be reached via some form of electronic mail originating on the Internet. The only area of the world where connectivity doesn't yet reach at all is northern and western Africa, and right now we are working with a group in Morocco to bring some degree of connectivity to their medical research community.
Connectivity & Bandwidth
There are lots of different ways of getting connected to the Internet, but some are more desirable than others. High-speed connections are desirable, but speed doesn't tell the whole story. Bandwidth is a much more important factor than pure speed. Some messages are only meaningful if they arrive complete and in one piece. In this case, the 'speed' with which the message is delivered depends both upon the speed of the connection and on the amount of data that the connection can carry per unit of time. For example, the radio link to a space satellite transfers data at the speed of light. In one recent case, several antennas on a space probe stopped working and the satellite is only able to transmit on a single frequency instead of on multiple frequencies as originally planned. The transmission of a single photograph now takes more than ten days, instead of the hour it should have needed.
The bandwidth that's needed depends upon the application. For human speech, telephone lines seem just about instantaneous, but for sending an image file through a high-speed modem, they seem slow. The actual speed of transmission is the same in both cases, but when sending a file, we're limited by inadequate bandwidth.
So, great, here we've got this global Internet and we're planning to hook ourselves up to it. How much bandwidth do we need? Well, that depends on what we want to do. I do some consulting to the film industry, so let's look at what a film director wants to accomplish. The film director wants to point a camera at some action on the movie set in Mexico City, and have the resultant image transmitted in near-real-time back to a lab in Los Angeles. If we want to send uncompressed digitized 35mm film images in real-time from one place to another then we need to push 26 frames per second through our connection. Each high-quality frame is about 8000 pixels wide and about 5400 pixels high, for a total of 43 million pixels per frame. At 26 frames per second, this is more than 1.1 billion pixels per second. But wait, there's more. This is only the spatial dimension. There's also a color dimension to deal with. It takes 24 bits of information per pixel to represent as much color information as the human eye can perceive. Twenty-four bits per pixel times 1.1 billion pixels equals 26.6 billion data bits per second. Oh, yes, there's also a couple of percent of overhead needed to ensure that the data arrives in Los Angeles in the same condition in which it leaves Mexico city. This means that our information superhighway needs to carry a whopping 30 billion or so bits per second point-to-point -- and that's just to meet the needs of one film crew.
30 billion bits (plus change) is about 4 gigabytes, or nearly 4,000 floppy disks worth of information every second. Can we accomplish that kind of information transfer today? Not electronically. There is no currently available installed technology that can provide that kind of sustained bandwidth to an end-user. But I can get that kind of bandwidth anytime I want using old-fashioned methods. A full-length movie, before cutting, runs about fifty hours, at 128 megabytes of data per frame. If I throw that movie into the back seat of my car in Mexico City and start driving, I can achieve 'real-time' if I can get to LA in less than fifty hours. In fact, I can throw hundreds of cans of film into a truck and thus achieve astronomical bandwidths of data transfer.
This is not so farfetched as it seems. I did a project once with Control Data Corporation in which medical billing data needed to be transmitted from a computer in San Diego to another computer in San Francisco overnight. When the project started, the volume of data was small, and we sent it by modem. But when the volume of data became large, the only way to get enough bandwidth was to write the information out to computer tapes and have a courier fly them up the coast every night for re-loading in San Francisco the next morning.
Those of us who were at universities have had dial-up internet access via a unix host for several years, but dialup access is limiting for two reasons. First, when you dial into a host machine you don't get to choose what software you run. You're stuck with what the host system's operators have installed. This is usually some form of character-based interface with 99 different modes and 8 keystrokes that do different things depending on what mode you're in -- a far cry from the Mac or Windows software we've come to expect.
Table 2: Time to transfer a 16 megabyte file
| 120 Baud | 12.3 days |
| 300 Baud | 5 days |
| 1200 Baud | 29.5 hours |
| 2400 Baud | 14.6 hours |
| 9600 Baud | 3.75 hours |
| 14,400 Baud | 2.3 hours |
| 28,800 Baud | 1.15 hours |
| 56,000 Baud | 38 minutes |
| 1,544,000 Baud (T-1) | 1.3 minutes |
| 10,000,000 Baud (Standard Ethernet) | 13 seconds |
| 45,000,000 Baud (T-3) | 3 seconds |
| 100,000,000 Baud (FDDI) | 1 second |
| 1,000,000,000 Baud (Internet backbone) | 0.1 second |
Secondly, dial-up access is SLOW. A few years ago when I was in the Department of Emergency Medicine at The George Washington University, I had a T-1 line installed and we bought a router to connect our three departmental ethernet networks to the Internet with moderately high bandwidth. It was worth the cost to be easily able to handle the traffic of a busy ftp or gopher site. Note that a T-1 line can transmit a single high-resolution 35mm slide image of 128 megabytes in about 10 minutes. People connected directly to the fiber backbone can send such a file in about 1 second -- only twenty times slower than Steven Spielberg needs. The time to transfer a 16 megabyte file over a wide range of connection types is shown in Table 3.
Internet Resources
Data can be transferred at high speed between any two computers on the Internet, but not with enough bandwidth to send a traditional movie across in real-time. What does it really get used for? Some people use the Internet to submit large data processing jobs to supercomputer sites, and many large companies connect their business networks together via the Internet. We connected our departmental billing people to our hospital IDX system across the Internet. To get from our offices across the street to the hospital our packets went out into never-never land. I was shocked when I first traced the route our packets took to get across the street. Most of our packets went through Bethesda, but some were been routed out as far as Ohio.
Basically, though, most people use the Internet for two purposes: non-real-time interpersonal communication, and information-gathering. The Internet allows access to an incredible number of resources: I spent a couple of days seeking out medical-type mailing lists, newsgroups, and on-line remote systems such as libraries. I found several fairly comprehensive lists of resources that already existed, and by the time I gave up I had developed a list of medically oriented resources that fills more than 200 pages of 10-point text.
Table 3: Common internet services
Electronic mail is the single most commonly used service on the Internet. You can send an electronic message to a colleague in Caracas, Venezuela as easily as you can to your next door neighbor -- and delivery is usually made in a few seconds or minutes. You can answer E-mail just by hitting a single key <R> for <Reply>. Addresses@gwemed.edu is under development as a name and address server for email addresses of emergency physicians. Try sending email to Dr. Ilya V. Zakharov, the director of the Center for Emergency Medicine in Moscow. His email address is zakh@home.vega.msk.su (it may take several attempts, and depending upon your connections, you may need to send eastern european mail via a gateway).
Mailing Lists
Mailing lists are just what they sound like: lists of people who all receive the same mail. Sometimes mailing lists are used to distribute an E-Zine, or electronically published magazine. More often lists are used like a long, drawn-out conference call among people with a shared interest in a topic. When you send e-mail to a mailing list server, your message gets automatically re-mailed to every single subscriber on the list. There are mailing lists to suit every interest. The National Center for Emergency Medicine Informatics offers an auto-subscriber that can get you signed up for several emergency medicine oriented mailing lists fairly painlessly.
Newsgroups
Newsgroups are the community bulletin boards of the Internet. They're a little like mailing lists, since if you post a message it will be read by a large number of people, but in some ways they are much better than mailing lists. They allow many-to-many discussions over a prolonged period of time. Messages are grouped in loose associations by topics (called 'threads') so that you can follow the flow of an online discussion on a particular topic of interest. You can 'tune in' any time you're in the mood, and can scroll backwards or forwards in any thread of interest.
Newsgroups are organized in hierarchical categories. Table 4 lists the 'big 7' most widely recognized categories.
Table 4: The 'big 7' newsgroup categories
Talk & Internet Relay Chat
Talk opens a channel of communication between two distant computers and sends a constant stream of empty packets in order to keep the route active and minimize transmission delays. When you type at your keyboard, each character may go out in its own packet. You can see that this is very wasteful of Internet bandwidth, and a few people running talk can really clog up a data pathway. By and large talk is tolerated because there's a lot of unused bandwidth just sitting around, and because it doesn't really 'cost' anything unless it is actually slowing down something else.
Internet Relay Chat (IRC) is a great big worldwide conference call -- a real-time CB-radio style of interaction. IRC actually behaves like a real-time, one-line-at-a-time version of a mailing list. You sign onto a 'channel' and as long as you're logged in, a server computer somewhere sends you a copy of every message it receives. Messages are processed one line at a time, rather than one character at a time. Many people share a 'channel' all at once, and your message appears on the screens of everybody who's currently logged into the same channel as you.
Telnet & Rlogin
Remote login programs allow you to sign into a remote computer and run programs on it as though you were sitting there at a local terminal. Of course, what you're allowed to do depends entirely upon your security level at the remote computer. Most machines won't let you log in unless you have an authorized account on that machine. Some machines are specifically set up to provide some public service through telnet. Telnet is one way to run online card catalog programs at remote libraries, for example.
FTP
FTP stands for 'file transfer protocol', and is one of the most valuable protocols on the Internet. This is what lets you download files from some other computer on the other side of the world. If there's a file you need sitting on a hard disk in a computer somewhere, and if you know where it is and have the right security access level to allow you to download that file, you use FTP to tell the remote computer to send you a copy. At the remote site, FTP breaks the file into packets and sends them out on the Internet addressed to you. At your end, your FTP program collects all the packets and re-assembles them into the original file. As with all well behaved Internet-aware programs, if a packet arrives garbled, the two FTP programs negotiate for that packet to be re-sent.
Anonymous FTP is a way of making selected files available to anybody who wants them, without bothering about security access. If I'm running an anonymous FTP site for emergency medicine, for example, you can ftp to my computer and login with a user name of 'anonymous'. I'll allow you to see a list of files available for FTP transfer, and will let you send any of them to yourself. There are literally millions of free files at thousands of ftp sites, just waiting for you to download them. Some examples of FTP resources are listed in Table 5.
Table 5: Selected ftp resources
Archie
Archie is an indexing program that lets you find files available for FTPing. Anonymous ftp sites periodically send information on the contents of their archives to a few special 'Archie' server computers on the Internet. You run a client version of Archie, which takes your desired keywords and sends a query to one of the archie servers. You get back a list of files and ftp sites whose descriptions include your keywords.
Gopher & Veronica
Gopher is basically an automated menuing program. Gopher servers run software that collects information about what resources are available on the Internet. The identified resources are offered to the end user through menus. You select an item on a menu, and your gopher software connects you to that resource. Most menu entries point to other sub-menus, and some of them point to other gopher servers, so that basically you can find and get into just about any resource you want. The interconnected set of gopher menus on all gopher machines everywhere is known as 'gopherspace'.
Veronica is an indexing system for gopherspace that has every word from every menu in gopherspace listed as a keyword. If a veronica search finds ten menu items at ten different universities all containing your keyword, then veronical will show you a gopher menu of those ten items. Assuming that you have a security clearance for the listed resource, you get access just by selecting the menu item.
WAIS
Wais stands for wide-area-information-server. This is like another version of anonymous ftp, except that in this case the files contents are indexed. You run a keyword search, and you get a menu list of files that match your query. You can acquire any of the files from the list just by selecting it from the menu -- sort of an automated ftp.
World Wide Web
WWW is similar to gopher, in that you can start from one point and tap into dynamically formed menus representing a large fraction of the resources on the Internet. WWW doesn't use traditional menus, though. Instead it uses 'hypertext' linkages. You see a graphic, or a page of text, or a list, and you select an 'active area' to jump to a linked item which can be another graphic, page of text, list, or whatever. For example, a WWW Weather server might offer you a full-page graphic map of the world. You click on the area you're interested in and you get another map, this time zoomed-in. You click on Washington, DC and you get the most recent weather satellite map for this area. On that map are buttons for 'short range forecast', 'long-range forecast', 'earthquake alert', 'and maybe even 'recent headlines'. The 'headlines' button takes you to a database of Washington Post headline stories for the past two years, and you can select stories to read from a list of headlines. Within each story you can select accompanying photographs to view, and you can also select linkages to previously referenced stories on the same topic. And on, and on, and on...
How To Connect
You want connectivity. You want it ASAP. You are willing to pay for it. Here's how to get connected. A dialup modem connection to a large host computer is probably the best way to start. Your host computer system will offer only a subset of the tools available to navigate the internet. Some host systems offer only e-mail access to the internet, but most are expanding their offerings on a daily basis. If you need more extensive access to a wider range of services, you'll need to establish a direct connection to an Internet gateway computer. You may find that a host system offers everything you need, or you outgrow your host system rapidly.
You can get a dialup direct connection via serial link interface protocol (SLIP) or point-to-point protocol (PPP) connections. Newer teleconnection technologies like ISDN and ATM are more expensive, but can bring high-speed connectivity directly to your bedroom. It is often possible to obtain a free end-user account on a local computer system that is connected to the Internet. Your local university or college is certain to be connected. In many areas you can obtain a free limited use account through the local library. Freenet providers are dedicated to facilitating free or low-cost public access points to the Internet. If you can't find a free connection point near you, you can establish an account with Erol's, Microsoft Network, America Online, Delphi, Compuserve, or another large commercial host. The fact is, it's incredibly easy to get connected today. If you're not happy with your current provider, click here to browse a list of more than 1280 ways to get connected to the Internet. Most serious providers today can give you 24 hour unlimited access for a flat fee of $15 - $25 per month. There's no reason to pay more!
Logout
The Internet is a more complicated and more valuable resource than you realized, and it's also more disorganized than you could have believed possible. This is at once its strongest feature and its worst attribute. The time to get connected is now, while it is still easy for an individual to have an effect on the growth and development of the organizational patterns that will define tomorrow's 'information alleyway'.