[0] https://www.amazon.com/Equations-Life-Physics-Shapes-Evoluti....
[0] https://www.amazon.com/Equations-Life-Physics-Shapes-Evoluti....
The Milky Way is a pretty average galaxy and it's 100,000 light years in diameter. In humans, if inter-hemispheric signals takes 3ms you can achieve roughly 40hz "synchronic activity". Let's say (and this is a massive simplification) that the inverse of the maximum inter-brain delay is ten times the "clock speed" of the brain. That would give a human that lives 80 years over 100 billion clock cycles in a life, and a galactic brain that communicates at light speed (so 3e-13 hz "clock speed") only 9467 clock cycles in 1 billion years.
So one single human life would have ten million times more "clock cycles" than a billion years of this galactic life-form existing. Or, the galactic life form would have about 3 minutes and 55 seconds worth of "normal" human thought, from the moment it's born until one billion years later. I know human babies certainly don't do a lot in their first 3 minutes and 55 seconds.
Or, for a simpler example: say a massive star goes supernova at one end of the "brain", damaging it and requiring resources from the other side. It'll be a minimum of 200,000 years before that section gets repaired.
When we search for life, we're probably looking for something very similar to us, something we can interact with and befriend.
I'll grant that there are probably galactic-scale patterns of light and gravity. My guess is that the amount of noise would make it impossible for any "life" to form. The unspecified type of wave would have to have an ability to store information in such a way that it would change the behavior of the waves.
As a tangent, these requirements of life is what makes the RNA world hypothesis so compelling. RNA is able to both store information _and_ act as an enzyme. 2-fer!
The answer is: carbon rich, watery life. Coming up with biosignatures for ourselves is hard enough, coming up with a biosignature for a galactic nebula is unbelievably harder because we have no examples. Extrapolating from one example is much easier than from nothing at all.