Midi file construction

I'm going way back to June with this one.

[Joe Monzo, TD 213.23:]

> One thing that really infurates me about MIDI sequences that
> many others make (as downloadable from Classical MIDI Archives,
> for example) is that they play them in on a keyboard with
> lots of rubato, but their sequencer has no way to keep track
> of where the beats fall, so when you look at the MIDI sequence,
> the rhythmic notation looks nothing like the original score.
>
> This lessens the value of their future use and modification
> for me.
>
> I use lots of rubato in the Tempo window to do all that,
> retaining the original meters and rhythms.

Joe, I hate to "infuriate" you, and I understand your point, but

. I'm lazy; keeping track of that extra info is a pain. I play in
real-time and measure in thousanths of a second.

. I don't usually need a score, so who cares?

. Using the midi scheme can distort the timing of the piece.

This last point is the one that most concerns me. Rubato can be
achieved using tempo meta-events, true, but what about the slight
rolling of a chord, or other not-quite-simultaneous striking of keys
which (as long as it's not excessive) adds to the performance?
Squeezing many notes into the same time does make the sound more
precise, but music isn't always about precision.

One can of course subdivide the midi beat and achieve the same thing,
but doesn't that defeat the point of using midi timing? Or are
score-making programs clever about interpreting such tricks without
adding 1/128th notes, etc.?

Now a confession: those retunings of your midi sequences that I e-mailed
you today, heh heh, have LOST the midi timing information; SORRY! I
like your Springtime Rag, by the way...

JdL