response to a protin on date conventions

[cbmvax!snark.thyrsus.com!lojbab]

?
Message  7:
Date:     Mon, 6 Aug 90 16:59:59 EDT
From: "Arthur W. Protin Jr." (GC-ACCURATE) <uunet!PICA.ARMY.MIL!protin>
To: lojban-list@snark.thyrsus.com
Cc: protin@PICA.ARMY.MIL
Subject:  times, dates, images, and S-W

Folks,
    (I am  a little slow in my reading and am just now responding
to things in the May issue of "ju'i lobypli".)

    On the subject of rationalizing the conventions for time and date
representations, the proposals (discussions between John Hodges & Bob)
were unimpressive, to put it mildly.  I hope you have names for the
months because I will strongly resist all those silly colons.  There
are two ways that I currently use for representing dates:
    (1) as number of day within month followed by name of month
        folowed by year as in
            06 August 90
or
    (2) as year followed by two digit number of month followed by
        two digit day of the month as in
            900806  or 19900806
The first of this is more (human) user friendly and requires no extra
punctuation or spacing: "06Aug90" can be unambiguously parsed.  The
second form is equally unambiguous, has not extraneous punctuation,
and has the fine feature that it can be sorted correctly by dumb
programs.
    (I will avoid a long discussion of the nonsense of mm/dd/yy
mentioning only that it is frequently indistinguishible from dd/mm/yy
and both forms have very large followings.)
    Without a major reform of the calander, I see very little reason
to do much with the representation.  The same goes even more so for the
clocks and time representation.  The abbreviations "AM." and "PM." have
served fine in the dual role of selecting which cycle of the clock
together with identifying the context for interpreting the numbers
(the numbers preceding are hour or hour:minutes or hours:minutes:seconds).
I again have found for dealing with dumb programs that it is usually
just fine to to extend the second form of date with a dot, ".", and
a twenty four hour clock without the colons, eg.
            900806.161632576
being
    4:16:32.576 PM 08 Aug 90

Now if you want to redefine all our clocks to fill a day with 100,000
units approximately equal to 0.864 seconds ...

    As the mind heals itself by forgeting painful things so now
the memory of "image languages" fades.

    I would like to share with you all my views of how language shapes
society's thoughts (Sapir-Whorf).  Individuals, in order to think
symbolically, must have symbols with which to model the thing they
are thinking about.  Society requires individuals to internalize,
at least a certain level of proficiency, one representation scheme,
that of the society's language.  Some of us use this representation
scheme very heavily (myself included) while others have personal
representation schemes such that they must translate from their
personal thoughts into their "mother" tongue to communicate their
ideas to others (my best friend is like this).  Clearly only those
thoughts that can be represented in some scheme can ever be thought
about and only those that can be translated into a shared language
can be past on (those that can't will die with the individual).
These are the hard limits but in practice most thoughts can be
--More--
represented in most languages, (but exceptions exist like "time"
and the Hopi Indians).
    The "soft" limits are like smoking and cancer.  You can get
cancer of the lungs without smoking and you can smoke without getting
cancer,  but statistics say that smoking increases the likelihood
of getting cancer.  So it is with thought and language.  Ideas
that are easier to represent will be more likely to be thought about
and shared.  Things that are more similar in their representations
will be more likely to be involved in analogies.  Operations on
thoughts that are easier with a given representation will be more
likely to be performed than the difficult operations.
    I have seen these limits first hand!  I have had problems in
mathematics where single statements expanded to fill a page.  Working
on those problems required dedication and meticulousness because
each manipulation step involved transcribing a whole page of representation
and a single mistake ruined the whole thing.  It should be no wonder
that scientists working on complex problems in physics (relativity,
quantum mechanics, and particle physics) developed very concise
representations for both the data and the operations to be performed
on the data.  Nor should come as a surprise that discoveries in the
substance of the field seem to occur very close temporally to discoveries
in the methods of representation in the field.
--More--


    thank you all for this forum,
    Arthur Protin


Arthur Protin <protin@pica.army.mil>
These are my personal views and do not reflect those of my boss
or this installation.

?

On Arthur Protin's message of 6 Aug 90 16:59:59 EDT
Subject:  times, dates, images, and S-W

> I hope you have names for the
> months because I will strongly resist all those silly colons.

1. The 'colons' correspond to the slashes in 08/30/1990.  As Dave Matuszek
mentions in another message, some separator is mandatory between the date
components, especially in speech, if we want to preserve unambiguity in
syntax.  There are, after all, three separate numbers there.  If you don't
believe this, try subtracting two dates to determine the number of days
between them.  Each of the separate numbers is in a different base.  The
number 'colon' "pi'e" is really a decimal point that says that the digits
on each side of the point are in different number bases.

2. The months have names - the number of the month + "-mast" from "masti"
e.g. pamast. remast. cimast.  So the abbreviation to digit representation
is natural.  Grammatically, though, if you want to express a date as
numbers plus month name, you have to turn ALL of the numbers into names
which is cumbersome to write (although some conventions have been used for
this such as:  "la 6d. 8m. 90n.")  There is NO special grammar set up for
dates; you must use the grammar for any other MEX or pseudo-MEX expression
to validly express dates (or times) in Lojban.

3. I used the colon because we DO you colons for time-of-day.  Why use a
different mark for dates when we use the same word for separators.  Of course,
just to be disconcerting, I sometimes use semicolons for both; but only if
I'm feeling ornery (and not when I'm trying to establish a convention).

> (I will avoid a long discussion of the nonsense of mm/dd/yy
> mentioning only that it is frequently indistinguishible from dd/mm/yy
> and both forms have very large followings.)

4. From what I've determined, mm/dd/yy is found ONLY in America, and not
universally here - the military uses Arthur's 1st version - dd MON yr where
the month is in text.  We also don't have gismu for the American (formerly
'English') measurement units, where the U.S. is out of step with the world.

5.  In addition to worldwide convention, we use dd;mm;yy for another reason
that is much more linguistic.  This order gives the most important and less
likely to be known information FIRST, allowing ellipsis of the rest.  If I say
I wrote this on the 3rd, you infer the month and the year the current ones.
If I say I wrote in on 3 Sept, you can guess the year is 1990.  But if year is
first, it can never be left out (except possibly with a null followed by the
'colon'), likewise year and month.
        Ease of writing sorting programs, or in fact any computer advantage
at the expense of human pragmatics takes the lowest priority in Lojban design.
We also try to avoid any argument based on aesthetics, since we have good
reason to suspect that American aesthetics in language is quite different from
other languages, and Lojban must preserve its cultural neutrality, especially
against American biases, if it is to be accepted.
        On the other hand Zipf's law is a valid linguistic argument.  It says
that most frequently used forms should be shorter, (and that if not, people
will tend to MAKE them shorter - which would destroy Lojban's unambiguity
if done haphazardly).  Thus we need a convention, and one that allows leaving
out unneeded information when this will shorten the form.

> Without a major reform of the calander, I see very little reason
> to do much with the representation.  The same goes even more so for the
> clocks and time representation.  The abbreviations "AM." and "PM." have
> served fine in the dual role of selecting which cycle of the clock
> together with identifying the context for interpreting the numbers
> (the numbers preceding are hour or hour:minutes or hours:minutes:seconds).
> I again have found for dealing with dumb programs that it is usually
> just fine to to extend the second form of date with a dot, ".", and
> a twenty four hour clock without the colons, eg.
>            900806.161632576
> being
>    4:16:32.576 PM 08 Aug 90

6. We are NOT trying for calendar reform.  Lojban is novel enough that we
don't need to reform the unreformable.  However, we do need a single convention
in a language to be used for communicating between cultures that do not share
a common convention.  The 12 vs. 24 hour clock is such an inconsistency.
Apparently in non-U.S. environments (as well as the U.S. military), 24-hour
clocks are more common than 12 hour clocks.  Most people in the U.S. use
digital clocks and watches as well.  So 'which cycle of the clock' is
inherently a biased statement towards preserving an obsolescent and declining
system.  We have to recognize that it exists, but we do not have to favor it.
We also need to be reasonably clear as to what our convention is, though.
If you see "900806.161632576" you have no way to even tell it is a date, much
less what the convention is for interpreting it.  We want to use the shortest
form that makes it clear that a convention is being used and whic possibly
suggests what that convention is to someone not used to using it.

7. In Arthur's final example, he adds a dot - this again is the Lojban
"pi'e" colon.  You wouldn't want someone to think it was a decimal point.
So don't use one.  On the other hand, he does use a decimal point in the
seconds of his time in the last line, using a colon for the other separators.
So

On Arthur Protin's message of 6 Aug 90 16:59:59 EDT
Subject:  times, dates, images, and S-W

> I hope you have names for the
> months because I will strongly resist all those silly colons.

1. The 'colons' correspond to the slashes in 08/30/1990.  As Dave Matuszek
mentions in another message, some separator is mandatory between the date
components, especially in speech, if we want to preserve unambiguity in
syntax.  There are, after all, three separate numbers there.  If you don't
believe this, try subtracting two dates to determine the number of days
between them.  Each of the separate numbers is in a different base.  The
number 'colon' "pi'e" is really a decimal point that says that the digits
on each side of the point are in different number bases.

2. The months have names - the number of the month + "-mast" from "masti"
e.g. pamast. remast. cimast.  So the abbreviation to digit representation
is natural.  Grammatically, though, if you want to express a date as
numbers plus month name, you have to turn ALL of the numbers into names
which is cumbersome to write (although some conventions have been used for
this such as:  "la 6d. 8m. 90n.")  There is NO special grammar set up for
dates; you must use the grammar for any other MEX or pseudo-MEX expression
to validly express dates (or times) in Lojban.

3. I used the colon because we DO you colons for time-of-day.  Why use a
different mark for dates when we use the same word for separators.  Of course,
just to be disconcerting, I sometimes use semicolons for both; but only if
I'm feeling ornery (and not when I'm trying to establish a convention).

> (I will avoid a long discussion of the nonsense of mm/dd/yy
> mentioning only that it is frequently indistinguishible from dd/mm/yy
> and both forms have very large followings.)

4. From what I've determined, mm/dd/yy is found ONLY in America, and not
universally here - the military uses Arthur's 1st version - dd MON yr where
the month is in text.  We also don't have gismu for the American (formerly
'English') measurement units, where the U.S. is out of step with the world.

5.  In addition to worldwide convention, we use dd;mm;yy for another reason
that is much more linguistic.  This order gives the most important and less
likely to be known information FIRST, allowing ellipsis of the rest.  If I say
I wrote this on the 3rd, you infer the month and the year the current ones.
If I say I wrote in on 3 Sept, you can guess the year is 1990.  But if year is
first, it can never be left out (except possibly with a null followed by the
'colon'), likewise year and month.
        Ease of writing sorting programs, or in fact any computer advantage
at the expense of human pragmatics takes the lowest priority in Lojban design.
We also try to avoid any argument based on aesthetics, since we have good
reason to suspect that American aesthetics in language is quite different from
other languages, and Lojban must preserve its cultural neutrality, especially
against American biases, if it is to be accepted.
        On the other hand Zipf's law is a valid linguistic argument.  It says
that most frequently used forms should be shorter, (and that if not, people
will tend to MAKE them shorter - which would destroy Lojban's unambiguity
if done haphazardly).  Thus we need a convention, and one that allows leaving
out unneeded information when this will shorten the form.

> Without a major reform of the calander, I see very little reason
> to do much with the representation.  The same goes even more so for the
> clocks and time representation.  The abbreviations "AM." and "PM." have
> served fine in the dual role of selecting which cycle of the clock
> together with identifying the context for interpreting the numbers
> (the numbers preceding are hour or hour:minutes or hours:minutes:seconds).
> I again have found for dealing with dumb programs that it is usually
> just fine to to extend the second form of date with a dot, ".", and
> a twenty four hour clock without the colons, eg.
>            900806.161632576
> being
>    4:16:32.576 PM 08 Aug 90

6. We are NOT trying for calendar reform.  Lojban is novel enough that we
don't need to reform the unreformable.  However, we do need a single convention
in a language to be used for communicating between cultures that do not share
a common convention.  The 12 vs. 24 hour clock is such an inconsistency.
Apparently in non-U.S. environments (as well as the U.S. military), 24-hour
clocks are more common than 12 hour clocks.  Most people in the U.S. use
digital clocks and watches as well.  So 'which cycle of the clock' is
inherently a biased statement towards preserving an obsolescent and declining
system.  We have to recognize that it exists, but we do not have to favor it.
We also need to be reasonably clear as to what our convention is, though.
If you see "900806.161632576" you have no way to even tell it is a date, much
less what the convention is for interpreting it.  We want to use the shortest
form that makes it clear that a convention is being used and whic possibly
suggests what that convention is to someone not used to using it.

7. In Arthur's final example, he adds a dot - this again is the Lojban
"pi'e" colon.  You wouldn't want someone to think it was a decimal point.
So don't use one.  On the other hand, he does use a decimal point in the
seconds of his time in the last line, using a colon for the other separators.
So
/XH'b#
C





lojbab