It was actually worse than that - Google+ used Reader as a backend. A lot of the stream-merging and ingestion features of Google+ were built on top of the Reader codebase. That meant that when Google+ product needs changed and they needed to alter the backend, they couldn't really do so without branching & disentangling or shutting down the external-facing Reader product. The latter was a much easier course for a time-constrained engineering organization, particularly since Reader as a project had been de-staffed and the engineers (that didn't leave the company) transferred to the Google+ org.