Heralds

Aodhan Ite an Fhithich aodhan at dobharchu.org
Tue Oct 22 20:57:48 PDT 1996


Dia duit!

"I. Marc Carlson"  wrote in a message to All:

 "MC> Don't look at me, I can't *write* programs at all, although I
 "MC> *can* write a rhetorical absurdity, which, come to think of it,
 "MC> was what  that suggestion was.  Or would you seriously consider
 "MC> turning over  the entire acceptances and rejections process to a
 "MC> machine?

Why not.  The entire Launch, Ascent, and Orbital Insertion of a Space Shuttle
is entrusted to machines.  Certainly checking armory for conflict is not going
to result in someone's death if the program has a glitch.

 "MC> If so, then I'm sure the people who are here who DO program
 "MC> computers, could tell us (hypothetically and with no commitment,
 "MC> mind you) if it's  too hard to come up with one to run a simple
 "MC> comparison of characters  in a specific order, and que the results
 "MC> up in a variety of different categories (say, "the first five
 "MC> words are the same", "the first four", and so on).

That's not the way you do it (I *do* program computers for a living).  Blazon
is a language, just like English, French, FORTRAN, or C.  It has a grammar,
syntax, and vocabulary.  If one can describe the grammar in a sufficiently
rigorous manner, then one can write a program to parse a blazon into its
grammatical and syntactical components.  One can then compare two such parsed
blazons, using a scoring system based on your rules of conflict, to determine
if the blazons in question actually conflict.

There are a few flies in the ointment.  Blazon is a sparse language and
attempts to eliminate redundant information (primarily repeated tinctures) and
also provides some information only by inference from what is explicit (in "per
pale an X and a Y" and "per fess an X and a Y" the locations of the X and Y
must be infered from the "per pale" and "per fess".

It's by no means trivial, but it isn't nearly as hard as Natural Language
Translation (English to French, e.g.), which has made enough progress in recent
years that's it is feasible to buy software to do such translations.  

>Tadhg and I attempted several years ago to produce a grammar for blazon.  It
>only addressed our ideal for blazon, not the abominations that too
>frequently
>appear in the SCA Armorial, nor all the period variants, and even so was
>incomplete..

 "MC> Since the point I was trying to make was that unless you want to
 "MC> turn your rejections and acceptances proceedure over to a simple
 "MC> algorithm simple to check for conflicts, you are going to have to 
 "MC> have retain human beings in the process, 

I'll come back to this in a moment, but, your right, we're going to have to put
up with human heralds and their foibles for a long long time.

 "MC> (for example, you can tell there's no real
 "MC> difference between a "Pentagram" and a "mullet of five points, 
 "MC> voided and interlaced", while a computer might not be able to),

That's just a matter of sufficient look-up tables - an internal thesaurus.

 "MC> You have my sympathies.  I have recently come to learn that major
 "MC> reason that so many things never get done in the SCA is that the
 "MC> people who are motivated to do something or other are so quickly
 "MC> burned out by dragging the people who can't be bothered to hoist
 "MC> their fat asses into doing anything that either won't be handed to
 "MC> them completely, or won't win them another award; as well as those
 "MC> people who, for whatever reason, feel threatened enough by change
 "MC> to actively campaign against even open dialogue (never realizing
 "MC> that the torpor of their comrades will be more than enough to kill
 "MC> even the simplest of dialogues).

There's a set of sub-programs in the Shuttle Flight Software known as
"Auto-Land."  These programs are able to fly the Shuttle to a landing without
human intervention (note that the Shuttle is a powerless glider - you only get
one attempt to land it).  Outside of test labs and training sims, this software
is *NEVER* used.  Why?  The Astronaut-Pilots would be unable to justify their
inclusion in the crew.  The aerodynamic forces during ascent are too dynamic
for human reaction time to cope with, so computers control the ride to orbit. 
On-orbit maneuvering is quite counter-intuitive, so the computers handle
rendezvous.  My wife (who has worked with both trained and in-training
astronauts) assures me that any of them can be trained to fly the last few feet
of a docking (once the target is in sight).  If the Pilots didn't land the
Shuttle, there'd be nothing left that any other member of the crew couldn't do.

I could write an automatic conflict checker; I have over my career written code
like most of its pieces as parts of other projects.  I am quite certain that,
outside of some initial testing and "wow, gee-whizz this is great" use by a few
branch heralds, such a program would not see any real use in the College of
Arms for the same reasons the Auto-Land code goes unused - it would put
specialists out of work.  So, I've never been motivated to do more than
co-author a paper and make some mental design notes.

However, if anyone is willing to pay me consultant's rates...

Baron Aodhan Ite an Fhithich, ML
Dobharchu Herald



More information about the Ansteorra mailing list