Understanding implies communication. When something is 'said', something is understanding what is being 'said.' If we look at how computers understand, or understanding in a generally computational way, we should talk about communication.[1] Additionally, when dealing with communication we are dealing with at least 3 things: something that is sending (the sender), something that is being sent (the data), and something that receives (the receiver). In the case of the computer, all three things are abstract and malleable. Also, the mechanical parts of understanding can be performed at various levels. For example, in the case of sound, if you want your computer to play (understand) a sound file on your disk so that you can listen to it (understand it), three things must happen.

1. The receiver (either the sound hardware itself or the software that operates it) must understand the format of the data and receive that data in the proper form and in the proper order.
2. The sender (a software method that reads the data) must understand the format of the data and send it to the receiver in the proper form and in the proper order.
3. the data must be in the proper format (a digital form of sound such as 44100 short integer sets of 2 samples for stereo CD quality audio).

In this line of communication, all three things (sender, data and receiver) can be changed or morphed into something else[2]. Thus, the name software. If one of them where to change slightly, 'misunderstandings' may happen. Also, the same sender and receiver methods on the same data will look and sound differently on a different hardware platform. The hardware itself must understand the format of the 1's and 0's - if the 1's or the 0's come first and how many of them to read at a time - 16bit, 32 bit or 64 bit.

DataDada simply reads and writes data, not caring if a particular data is informative or meaningful from the perspective of the receiving software methods or not. It makes the assumption that all data on disk is sound in the form of interleaved stereo samples at 44100 Hz and video in the form of 480 pixels width and 240 pixels tall, where one pixel is composed of 32 bit values - 8 bits each for red, green, blue and alpha values. I chose both of these formats because they are generally speaking the usual formats for digital sound and video. A change in either of the receiver or sender methods from 44100/stereo audio to 480/240/32bit video will result in a significant (although perhaps not meaningful) change in look and sound.

It has been argued before that in a world of continuous informative data, noise is simply that which lacks form and meaning. In this sense, noise is essentially the only possible breeding ground for new forms and content[3]. In the discrete world, meaning can be created out of anything, and noise is sometimes the result of a simple shift in the place of the read/write method on the data or a shift in the method or data itself.

footnotes
---------------
[1] Even when attempting to understand yourself (a difficult thing to do) your consciousness is having an intimate internal meta-dialog.
[2] Also, translation methods could be used to translate the data to fit the format of the sender or receiver. For example, translating a 16bit audio sample to a 32 bit floating-point audio sample.
[3] I say content here lightly. It can almost be assumed that form, format, design, style and genre are the receiver methods that create the illusion of content.