Research->Music Software Development->Developing MIDI-Related Application->Rondom Numbers & Music Theory
This section is my thinking toward rondom numbers and music theory.
Automated Composing That Uses Random Numbers
There is "Random numbers" in the basis when composing with the computer automatically. A different action on each occasion is taken for man even in case of being in the same situation, and, every time, if given instruction and data are the same for the computer, the same result is output. If I paraphrasing it, for instance, every time when the same instruction "Compose the tune like rock music" is repeatedly given to the computer, the same tune will be composed. Since this is not interesting, it is necessary to devise it by using random numbers so that an every time different tune is composed.
Random numbers can be used when composing by the hand work without using the computer. For instance, if you shake dice twice and you will set some rules such as 1 equals to "Do" sound and 2 equals to "Re" sound and so on. Similarly, as for second dice, you set rules such as 1 equals to quarter note and 2 equals to eighth note and if you keep on shaking the dice, you will compose a tune. (However, the tune will not be musical.)
Automated Composing By Random Numbers Without Music Theory = Chaos
I will talk about creating a melody of single note for the moment. (I will not talk about harmony for a while.)
Actually, I have created a program that is described as following.

Algorithm - Chaos

1 The following procedure is repeated until 20 sounds are generated.
2 Random numbers from 0 to 127 are generated, and consider them as sound level.
3 Random numbers from 0 to 127 are generated, and consider them as note length.

In MIDI, the sound level is shown by the note number from 0 to 127. The value from 0 to 127 is generated by random numbers, and it is allocated in each note number as it is.
The length of the sound such as quauter note and and eighth note are called sound value in musical terms. Moreover, there are quarter notes that have short gate time such as staccato and have long gate time with slur. Here, the length of only 0 to 127 tick (One tick is milli second) is allocated in the sound without using the concept like the sound value and the gate time, etc. Resolution is assumed to be 96, and strength of the tone and the sound is assumed to be constant.

Listen You can listen to the midi file created by the above algorithm.
You will agree that above file sounds like modern music with inconsistent tune. I will call this kind of music "Chaos" in this section. It is impossible to expect the next sound from the previous sound.
Automated Composing Without Random Numbers = Order
Let's think about an automated composing that adopts the music theory of "you must use only note number 60 and the sound length of 96 ticks".
Rule
All of note numbers are 60 and note length is 96 ticks.

Of course, this music theory is neither the theory of classics, nor that of popular music, only applies to the explanation here.

Algorithm - Order

1 The following procedure is repeated until 20 sounds are generated.
2 Sound level is 60.
3 Note length is 96.

Conditions other than the above-mentioned are the same as previous example.

Listen You can listen to the midi file created by the above algorithm.
This is rather tedious because the same sound will continue. You can expect 100% what the next sound will be from the previous sound. I will call the tune in this state "Order".