I was thinking about whether opportunities to do good that only exist for a limited amount of time be prioritized and I haven't seen any discussion on this particular issue.  I think there are reasons to prioritized limited-time opportunities and this has some implications for global priorities. The following are some thoughts on the question based on a simple toy model. 

Consider the following case: there are two distinct opportunities to do good (A and B), both costing $1 million. One can do more good with A than with B, but B is only available for one year while A will always be available. Are there cases where we should do B rather than A?

If we assume that the agent has a fixed budget of $1 million with no temporal restrictions on how the resources are spent, then the choice collapses into a simple choice between doing more good or less good. And the time-limitedness of one of the choices seems irrelevant. 

But real choices are not like this. Resources are often better modeled as cashflows rather than fixed budgets. If, instead of a fixed budget of $1 million, I have a cash flow of $1 million every year. Then I should clearly do B while I still can, and then do A next year. There is no cost to waiting one year to do A. (for our purposes, let us assume away temporal discounting) Similar to income, labor time is also better modeled as cashflows. If two opportunities both require one year of my time, but the less important one is time-limited, then we have reasons to favor the time-limited opportunity during the time at which it is available. Underlying this is a familiar point: decision-making must take opportunity cost into account. Neglecting time-limited opportunities has a higher opportunity cost than neglecting time-unconstrained opportunities. This means that knowing the monetary/labor costs and intrinsic benefits of relevant intervention opportunities is not sufficient to make an optimal decision. We also need to know whether these opportunities are time-limited and how our resources relate to time. 

This has some implications for global priorities. Perhaps some areas are particularly worth working on while they are in their infancy, such as AI safety/policy, outer space governance, pandemic response, and climate catastrophe prevention. Other problem areas will foreseeably persist for a long time, such as generic philosophy research, global poverty, animal suffering, nuclear security, and global pandemic preparedness. The upshot is that, even if spending on a limited-time problem area has less (ex-opportunity cost) expected value than spending on a non-limited-time problem area, it may still be rational to prioritize the former over the latter. 

I would be interested to hear if there has been discussion on this which I've missed and people's thoughts on this in general.

Some outstanding questions that I haven't addressed:

• What if we expect limited-time problem areas to keep popping up (if we do not live at the hinge of history)? Or what if there are more than two options? How do these change our evaluation?
• If we build more features into the model (cashflow is not fixed but growing), would the results change?
• How does our degree of confidence in a problem area being time-limited or in us not living at the hinge of history interact with whether time-limited opportunities should be prioritized?