Information Technology Reference
In-Depth Information
The new algorithm is at least a hundred times faster.
Be clear about which base a number is in.
1
Quantities that are in the same unit should, for consistency, be represented to the
same precision. In physical experiments, it is usual to represent quantities to the
same relative precision, that is, the same number of digits. In computer science, in
which values are usually measured to the same absolute precision, it is more logical
to represent quantities to the same number of decimal places.
The sizes were 7.31 and 181 Kb, respectively.
The sizes were 7.3 and 181.4 Kb, respectively.
A paper gave the same figure in different places as “almost 200,000”, “about
170,000”, “173,000”, and “173,255”—an entirely unnecessary inconsistency.
Be realistic about accuracy and error. Your system may report that a process
required 13.271844 CPU seconds, but in all likelihood the last four or five digits
are meaningless. You should not imply accuracy by including spurious numbers.
For example, “0.5 s” is not equivalent to “half a second”, since the former implies
that careful measurements were taken. Guesses and approximations should be clearly
indicated as such, with words such as “roughly”, “nearly”, “approximately”, “about”,
“almost”, or “over”; but don't use wordy phrases such as “in the region of”.
Percentages
Use percentages with caution.
The error rate grew by 4%.
This example is ambiguous because an error rate is presumably a percentage. It is
better to be explicit, and to avoid mixing kinds of percentages.
The error rate grew by 4%, from 52% to 54%.
The error rate grew by 2%, from 52% to 54%.
When stating a percentage, ensure that the reader knows what is a percentage of
what. If you write that “the capacity decreased by 30%”, is this 30% of the old
figure or the new? The convention is to use 100% as the starting point, but in a series
of statements of percentages it is easy to get lost. Use percentages rather than odds
to express probabilities.
The likelihood of failure is 2:1.
The likelihood of failure is one in three.
The likelihood of failure is about 30%.
1
It is said that there are 10 kinds of people in the world, those that understand binary and those
that don't. (And how true it is. A reader suggested that “10” be changed to “ten”.)
