As also discussed in this thread, creating a scatterer event with each voice panned randomly is not possible without using a complicated workaround. Adding a pan parameter to the Scatterer Instrument would be a much cleaner solution. Besides, the pan parameter fits perfectly to the concept of a scatterer instrument, and it feels like an omission not having pan included in it (or possibly a technical limitation?).
You’re not the first person to observe that the collection of tools provided by the scatterer instrument is arbitrary, and not well-suited to all use-cases. You’ll be glad to hear we’re planning a solution: We’re going to overhaul the scatterer instrument to make it more modular (or possibly deprecate it in favour of newer, more modular instruments that cover the same functionality).
I’ll add you to the list of people who’ve requested this kind of improvement.