English | français | deutsches | italiano | portuguese | español
Outline for February 17, 1998:
Hardware and Software Issues
The problems to address:
- Multiple users must be able to access the same information
(conferences, interactive chat) at the same time.
- Some information (e-mail) must be kept private, so identities
must be verified.
- Some information (e.g. e-mail, conferences, Web pages) needs to be
shared with other computers.
Strategies for addressing the problems, (in order of structural
simplicity):
- The Mainframe Mentality solution (example: Chebucto Community
Net)
- Users log into a "shell" of the actual OS (e.g. UNIX), so security is built-in. Unfortunately, a hacker who guesses the "root" password can
destroy the system.
- Internet-standard software (lynx for menus and Web pages,
pine for e-mail and newsgroups, talk for interactive chat) allows 100%
compatibility with other Internet servers. Conferences and Web pages can
be shared, and no e-mail gateway is required.
- The only interface available is plain text, so users can get by
with very inexpensive hardware.
- The text-only interface requires users to be literate.
- Since all connections are continuous, the more users are
online, the slower everyone goes, and there is a finite limit to the
number of simultaneous users the server can handle.
- CCN's ChebuctoSuite package uses all free software (including
the OS, Linux) and can run on inexpensive hardware, so setup costs are
minimal.
- Maintenance can be expensive, since UNIX has a steep learning
curve.
- The Web-based solution (example: Twin Cities Free-Net)
- In addition to the standard UNIX software, use special Web
conferencing software (e.g. Caucus) that allows conferencing, e-mail, and
interactive chat from within a Web browser (e.g. Netscape or lynx).
- Since Web connections are not continuous, the same server can
handle much higher traffic when users come in through the Web.
- Users may feel pressured to use expensive computers so they can
enjoy the Web interface.
- Logging in is accomplished through the Web browser, which sends the user's name and password every time it requests a new page, so security is dubious.
- Interfaces of Web conferencing packages vary widely: one
message at a time with slow response, or many messages on a page with slow
loading? Frames or no frames?
- Discontinuous Web connections do not permit the server to contact
the user with new information such as chat requests, so both users must
enter a chat room (if such a service is even offered) in order to chat.
- Personal opinion: Web conferencing software rarely feels interactive and may discourage people from posting. It also may turn people outward to the Internet instead of inward toward the community.
- The souped-up BBS solution (example: Cleveland Free-Net)
- Users log into a BBS program (e.g. FreePort) instead of
directly into the OS. This makes hacking much less likely.
- Work can be distributed among several servers without the
users' knowledge. Each server may be running a different flavor of UNIX,
so great flexibility in hardware is possible.
- The text-only interface is accessible with very inexpensive
hardware.
- FreePort can cooperate with Internet-standard software (lynx,
pine) to allow Internet access, and UseNet standard newsgroups can be
shared over the Internet for a larger audience.
- FreePort is very old (last officially modified in 1991) and
not always stable, sometimes losing e-mail and conference posts.
- Users spoiled by the Web are sometimes disappointed by the
text-only interface.
- The text-only interface requires users to be literate.
- The classic BBS solution (example: Great Lakes Free-Net)
- BBS software runs on a PC, so logging into the OS is
impossible and the risk of fatal hacking is reduced. Most
administrative tasks may be done through the client software, but others
require sitting down at the PC itself.
- BBS servers tend to have shallower learning curves (that is,
they're easier to learn) than UNIX-based servers, so system operators
may need less training.
- BBSes may require a separate mail server for sending and
receiving Internet mail. However, they may have built-in support for
exchanging e-mail and conferences with other BBSes running the same
software.
- If there is a graphical client, it is designed specifically
for CMC and is potentially very user-friendly. FirstClass Client
allows users to work in multiple windows simultaneously, and sounds alert
the users when background windows need attention. FirstClass Client also
stores many graphics on the client computer, making data transfer very
rapid.
- Some BBS packages that focus on a graphical interface may not have
a fully functional text interface, so poorer users may be discouraged;
however, client software will generally run on a much cheaper computer than
will a full-blown Web browser such as Netscape.
- FirstClass conferences are simply public mailboxes, so a single
subscription to an Internet mailing list (e.g. NPTN's cybercasts) may be
shared by all users.
- In practice, client software can be difficult to install on the
wide variety of computers that users own, so phone support is
ubiquitous.
- Many BBS packages offer no incoming or outgoing Web support,
so investors may see them as a technological dead end. Some other BBS
packages, such as ResNova's NovaServer (no longer available) and Mustang's
Wildcat do offer limited Web support.
- FirstClass has been discontinued in favor of FCIS, a Web
conferencing package.
Community Networking Rule #3: Technology must be appropriate to
the needs of the community.
Optional workshop: the World Wide Web
- Obstacles to sharing information over the Internet:
- Each brand of computer originally required a different
protocol (set of rules for who says what when) and a different
encoding (method for turning intelligible information into numbers
and back again, e.g. ASCII text). Cross-platform standards evolved by
necessity as people demanded ways to connect their computers together.
- Early Internet-standard protocols (FTP, telnet, IRC, talk), with the
exception of e-mail and UseNet (SMTP, POP, NNTP) required continuous
connections between computers. Such connections became less reliable
over large distances and as Internet traffic increased.
- Gopher
- An Internet protocol named gopher was introduced in 1991, the same
year as the World Wide Web. Both were discontinuous: one
computer requested a document, the other supplied it, and that was the
extent of their communication. Gopher intially received more attention and
support than the Web because its reliance on menus rather than prose seemed more useful for cataloguing information.
- Gopher is menu-based. Each gopher server has a set of menus
listing and describing documents (mostly ASCII text files) it offers and
links to related menus on other servers.
- A program called Veronica allowed gopher servers to search for
specific information by querying each other.
- Enter the Web
- The World Wide Web is the implementation of the HyperText Transfer
Protocol (HTTP), which allows a computer to request any document without
first navigating through menus. This means any document can link to any
other document, anywhere in the world, forming a decentralized, web-like
structure of information similar to the neural networks of the brain.
- HTTP can be used to view any type of document encoding (e.g. text
files, GIF or JPEG images), but it's really designed to work with
HyperText Markup Language (HTML). HTML ensures that a document
can be formatted complete with text styles and colors, nested lists,
embedded images, and links to other documents, and it will look acceptable
on any computer. This is accomplished by making all formatting
instructions very explicit, removing all ambiguity.
- HTML is a subset of the Standard Generalized Markup Language
(SGML), which theoretically allows any kind of document to be
represented as plain text. SGML has never caught on, but another subset,
the eXtensible Markup Language (XML) looks promising.
- The Web didn't have a clear advantage over gopher until graphical
browsers became available. A browser is a program that makes HTTP
requests and displays the resulting documents. Almost all browsers must
be on the Internet (that is, their computers must have IP addresses) in
order to function; the notable exception is ResNova's now-defunct
NovaTerm client.
- Because HTML pages need not be strict menus, the gopher/veronica
strategy of searching was no longer effective. Search engines are
services that maintain enormous databases about what information can be
found on which page. When you use a search engine, you are not searching
the Web itself but rather a database generated from the Web.
- Because HTML is based on the ASCII (American standard) character
set, it is best suited for the English language. Those who write
in other languages must use clumsy HTML tags to obtain the special
characters they need or expect their readers to switch to an entirely
different character set (e.g. Japanese, Cyrillic, Greek).
- The World Wide Web Consortium periodically announces new versions
of HTML, but the language's development is currently being driven by
Netscape and Microsoft, which keep adding new features to their browsers
whether they're officially supported yet or not.
- Notable developments in browser technology include frames (allowing
information to appear in multiple panes within a window), support for
embedded animation and sound, ability to display PostScript files
(encoded for a printer rather than for a screen), and ability to run
programs written in Java, a platform-independent programming language.
- Some BBS packages and most Web conferencing packages generate HTML
documents on demand, allowing customization for each user. Java programs
also allow more interactivity than a browser would.
Return to the course syllabus.