Looking at skilled go players, I can come up with some different areas that they excel at:
* Great endgame
* Fast and deep reading
* Excellent positional judgment
* Light play

These are just brainstorms that come to mind from the top of my head. So how do I improve in these areas? Some are straightforward:
* Great endgame - Study endgame problems; practice
* Fast and deep reading - study various life and death problems; maybe some difficult, some easy
* Positional judgment - practice evaluating who is ahead and/or behind

But I get stuck on light play... The best I can think of is, "try to consciously think about playing lightly". How might one improve this skill?

_________________
be immersed

Hi Kirby,

Training your deep neutral nets with tons of examples from Go Seigen and AG may be one path.

2- or 3- stone games as W ?

I don't have an answer, but I do have an observation.

Your first three bullets are all objectively good things that are never at odds with our goal of playing well:
* Great endgame
* Fast and deep reading
* Excellent positional judgment

Your fourth bullet - playing lightly - is this *always* something for which one should strive? If the opposite of light is heavy, then playing lightly seems good. If the opposite of light is solid, then depending on context, I might prefer to play solid than to play light.

Question: Are stones light, or are strategies light? I don't know.

Good observation. I think that I view "playing lightly" as a strength in stronger players, because it's a skill that I seem to lack. It may not always be the best strategy, but this light and flexible way of playing is something that I'm not very good at.

It may not always be a bad thing, but there are certainly times when I should probably have the ability to play lighter.

_________________
be immersed

Reviewing pro games that you feel exhibit that quality seems like a good way to train that skill. I also think you could play games with the goal of sacrificing stones. When every stone you place is a potential sacrifice, there is no reason to make groups heavy.