Definitions vs examples.

From: R. Clayton <rclayton_at_monmouth.edu>
Date: Mon, 24 Feb 2014 11:25:54 -0500
When asked to define X, it's important to remember that giving examples of X is
not defining X.  While it may be useful to give examples to illustrate a
definition, examples do not replace the need to give a definition.

For example, if someone asks for a definition of "well-known port numbers," the
response "HTTP is port 80" does not define "well-known port numbers," it gives
an example of a well-known port number.

Similarly, the response "TCP is connection-oriented," does not define
"connection-oriented," it just gives an example of a connection-oriented
protocol.

In general, the difference between a definition and an example can be tricky,
but in this class, and on the tests, the difference is straightforward: it's
the words in the textbook you see around the bold-face appearance of the word
or phrase being defined.

There's a simple test for determining the quality of a definition: use the
definition to determine whether or not objects (concepts, whatever) are
correctly included in the definition.

For example, does the definition "HTTP is port 80" for "well-known port number"
determine whether or not port 25 is a well-known port number?  It's hard to
say, but because 25 != 80, the answer may be "no."  There are two problems with
"HTTP is port 80" as a definition of "well-known port number:" first, it's hard
to understand what it means, and second, if you think you understand it, you're
liable to get the wrong answer.

In contrast, consider "a mapping between a port number and a service that's
unique throughout the Internet" as a definition of "well-known port number."
With this definition it's clear the question "Is 25 a well-known port number?"
turns into "Is port 25 mapped to a service in a way that's unique throughout
the Intenet?" and the answer is "yes: 25 maps to SMTP."

Similarly, the response "TCP is connection-oriented" to the question "What does
connection-oriented mean?" is unhelpful in answering the question "Is UDP
connection-oriented?" (Note that, in this case, answering "no" because TCP !=
UDP gives the correct answer, which is why it's important to test the
definition with positive and negative cases).  On the other hand, the response
"a protocol is connection-oriented when the protocol at each end of the
connection keeps track of the state of the data transfer" also gives the right
answer, "UDP is not connection-oriented" for the right reason ("UDP does not
keep track of the data-transfer state").
Received on Mon Feb 24 2014 - 11:26:46 EST

This archive was generated by hypermail 2.2.0 : Mon Feb 24 2014 - 22:46:34 EST