But I see no reason, either in functional integrity or in social practices, that topical circles would diminish G+. Those not interested would simply not use them -- this is true of FB groups and lists.
It is incredibly difficult to achieve the network density and connectedness required to make a social network stick. Attention is in short supply -- witness Ning's failure to become the social network of interest groups. Perhaps it's not feasible for vertical and niche networks to match the success of the generic social nets. For this reason strikes me as in the ballpark to enable sustained topical conversations where the audience already exists, and where shared topical interests could augment individual profiles and deepen engagement.
The model itself /is/ the Twitter model (circles as lists, reshare as retweet, comment as reply), with only one fundamental addition: "limit visibility of this tweet to these lists", which drastically helps specific use cases where otherwise people currently end up with multiple Twitter accounts (and then "awkward stuff" starts to happen due to how Twitter publicizes the resulting interaction with that tweet).
In fact, the only similarity between Google+ and this feature is that both involve groups of people. That's about it. I don't think implementing this would improve Google+ at all; it may be a good idea for another site, and the groups thing may be a good way to go about it, but not Google+.