Disclaimer: warning, technical ranting below!
In opposition to what one might think, I'm not ready with tactics. Not yet.
For my level I'm a pretty good tactician. But if you compare it with masterlevel, then you can see that there still is a way to go. The circles are a good way to improve. I have reached the limits of what I can achieve with this kind of training. For now I focus on my other weak areas, but I'm sure there will come a moment in the future I will have to bring my tactics to a higher level.
If I look at it now, doing the circles is a rather blunt method to ingrain only the lowest level patterns into the brain. Although it brought me to 1750, I doubt if it can bring me any further without modifications to the method.
In what directions should I look for improvement of the method? The post of Blue Devil about guided pattern recognition as well as my own experience with Zent Larsen seem to suggest that there are higher structures needed in the brain to guide the process of pattern recognition.
I guess that the inclination to produce such higher structures from lower ones differs from person to person. There are persons who are inclined to draw conclusions based on only a few facts, accepting or unaware of the resulting high errorrate of such conclusions. And there are persons who need an awfull amount of facts before they dare to base a conclusion upon it.
I guess that the former derive the most results from low level pattern recognition training, because they build the required higher level structures which are required to guide the lower level recognition with more ease. Do I still have to say that I reackon my self to the category that needs an awfull amount of facts before I come to a conclusion?
In the past I studied the thematic bishop sacrifice at f7 (based on 100 problems from Papa Polgar's brick). Since I have done that, I'm much more aware of such sacrifices. Plus it is much easier to see when it is going to work and when not. This is all due to the fact that I consciously builded the higher structures that help me to analyze positions with possible sacs. Wouldn't it be awesome if we could make use of the efforts of other persons to build these structures?
If I will continue in the future with tactics, I will start with experiments to test these guesses. In the mean time I am busy with annotated mastergames from where ever I can find them. So long the annotation is directed to clarify the themes and not just a variation dump. These themes are the higher level structures for both tactics and positional play. Are you for instance familiar with the thematic knight sacrifice at f5 in the Ruy Lopez?
So the study of themes with the aid of well annotated mastergames will be my way to go the coming time. Finally a method where I don't have to invent everything on my own!