I guess 30 years is some breathing room, but 10 years, being just 2.5 halving cycles away, would be close enough I’d say it’s time to test candidate algorithms. Especially if there’s a significant chance 10 years is pessimistic.