Machine Learning Speculation – Network Compression Ratio?? Water parameters?

From GAN network models I wonder. Since the generator depends on noise. How little noise is necessary to make it work.

Could we get it to work with a little as a single random data value, like a bit. Varying with time randomly.

So from this I speculate that there is an effective way to compress information. But requires perhaps a fast computing power.

So physically, water is important to life and water is a network. Since life is also complex. Water might be complex to. The weight parameters of water to sustain all life on the planet might be so large that it needs to be compressed somehow.

So what do I mean by water is needed for life. I guess that ?computing ready weight parameters in the water acts like the surrounding network between life networks.

?All contact surfaces to water needs water to sustain life.

Basically water is important.