[r-t] Exhausted search spaces

[Don Morrison]

On Wed, Feb 3, 2010 at 3:06 PM, Mark Davies <mark at snowtiger.net> wrote:
> Eddie, I think you've missed the point of Don's composition numbering
> system. He *has* published all of them. He numbers his compositions (as far
> as I understand it, no doubt he will confirm) in the order he produces them,
> regardless of method or length.

Roughly. Because experience taught me I had to change an earlier
numbering scheme the lower numbers in particular are not necessarily
chronological. Similarly, I periodically rediscover something I
produced in the past but never put on the site, and thus they end up
not being chronological. Or, similar, usually slight, chronological
aberrations also happen when I produce compositions of as yet unnamed
methods that remain unadded to the site until the band that
commissioned them rings them.

As you've undoubtedly noticed I tend to produce a fistful of
compositions for the same method, altogether. Within such a fistful
added to the site at the same time I generally try to order them such
that the later (higher numbered, sorted to appear later on the page)
compositions are the "harder" ones, and the earlier ones are "easier".
But that only applies to a single batch added at the same time, and in
any case is horribly subjective.

> No doubt he had hundreds of millions of compositions of Lake Garda Delight
> to choose from, but has only in fact selected and published three of them.

No, not really, at least in practice. Unlike the procedure I
understand Mark uses, I rarely produce an inordinate number of
compositions, and then use some sort of selection procedure to pick a
few out of it. I generally iterate repeatedly, cycling between a
little bit of computer search and a little bit of twiddling of what
I'm asking the software to do, possibly even fitting in some bits by
hand. While it's usually just a small number of such iterations, for
some compositions I may go around that loop fifty or a hundred times I
expect. Such an iterative procedure usually, though by no means
always, produces a much smaller number of compositions from which I
pick. Often fewer than a dozen. Though, of course, I'm performing some
sort of implicit selection earlier, driving the restriction later on.

This correspondance has made me think that when next I'm working on
something, sometime in the next few days undoubtedly, I should keep a
bit more statistical information about what I've done and so on. I
suspect such will illuminate this a bit more. In particular, it rarely
makes sense within my style of working to say I found 132,579,302
compositions in four hours, and picked the one that maximised this
particular musical score, or whatever.

> Don also gives pretty detailed musical analysis for each composition he
> publishes, which pretty much tells you what he was looking for. Admittedly
> his composition descriptions are not quite as fulsome as mine. ;-)

There, see, I am not *that* long-winded, am I?*




* That question is rhetorical only; please don't answer it.


-- 
Don Morrison <dfm at ringing.org>
"Diversity may be the hardest thing for a society to live with,
and perhaps the most dangerous thing for a society to be
without."     -- William Sloane Coffin, _Essays on Public Morality_