Richard's Rules of Order for Tables

In much published material I've seen, tables aren't drawn up intelligently. Ignoring the fixed width issue, many seem to maximize whitespace and thus make following the records difficult. The problem of whitespace-following was quite pronounced (although not specifically mentioned as a problem) in the butterfly ballot of the much-publicized USA 2000 Presidential election in Florida.

The rules I suggest are simple:

For example, from MTA, which of these two tables has easier matching?
Travel Information Center 1-718-330-1234
Travel Information Center for non-English-speaking people 1-718-330-4847
Travel information for people with disabilities 1-718-596-8585
(TTY) 1-718-596-8273

1-718-330-1234 Travel Information Center
1-718-330-4847 Travel Information Center for non-English-speaking people
1-718-596-8585
(TTY) 1-718-596-8273
Travel information for people with disabilities

Many "tables of contents" get this wrong: they put page numbers on the right and compensate with connecting periods or alternating color or gridlines.

Of course, this assumes that the data is non-overlapping (space isn't at a premium).

On Endianness

Here I use a looser idea of big-endian and little-endian:
big-endian
The most groupable/significant portion is at the beginning, and successively less significant portions follow.
All numbers are written and pronounced this way (e.g. one thousand one hundred eleven). Phone numbers (country code, city/area code, exchange, line number) are also this way. Bibliographic citations tend to have page number/line number last.
little-endian
The least groupable/significant portion is at the beginning, and successively more significant portions follow.
Internet e-mail addresses are written this way (accountID, subdomain, domain, TLD), and postal addresses on envelopes are written this way. In many nations, people respond to "what's your name?" with given name followed by surname.
Many discussions of endianness seem to find little advantage of one over the other. However, a case can be made for big-endianness over little-endianness:

Over a slow communications link, big-endian allows more processing to occur before the full message is received.

For example, if someone gives you their postal address, would you rather hear the nation first or last? Or, within the USA, would you rather hear the state first or last?

The same reasoning applies to dates (which ought to be year-month-date) but dates are often excepted because they are so short (only a few seconds to deliver). On the other hand, how often have you heard "half past six" instead of "six thirty"?

goldfndr@mindspring.com.