Choosing and Justifying the Discount Factor

Your team is debating the discount factor for the notification agent. One engineer argues for \gamma = 0.5 ("we mostly care about the next few notifications"); another argues for \gamma = 0.99 ("churn is a long-term problem, and we should account for it fully").

1. Using the effective-horizon interpretation of \gamma, quantify roughly how many future notifications each choice weighs meaningfully.
2. Given a trajectory where sending now yields r_{t+1} = +1 but raises mute probability such that the expected reward 4 steps later is r_{t+5} = -15 (conditional on reaching that step), compute the discounted contribution of that -15 under both \gamma values and explain what this implies about each policy's behavior.
3. What downside is there to always choosing the largest possible \gamma close to 1?