And this is a rather low-cost workaround that any survey organizer may practice now. Even within the same workspace (older model) the supervisors were sometimes issued interviewers accounts if in need to make corrections without resending the data back to the remote interviewer.
Yet, where our involvement is critical, is where no workarounds exists, such as if the interviewer needs to be part of more than one workspace. Given that the registration is immutable after initialization, the interviewer can’t submit to a different workspace or different server (without uninstalling the application and loosing the collected data). Switching that on the fly would be good next step.
Occupying different roles in the survey is also problematic from the training standpoint, it’s hard to imagine taking an interviewer, perhaps even very talented one, and putting him at the HQ role to work with statistical packages or controlling the API. The roles are sufficiently different in skills requirements to be interchangeable.