See, this is a good case for a modernized-and-improved remake.
Originally envisioned with as many as 7 shipwrecks, the layout of each apparently to be procedurally generated à la Beyond Zork ,
Procedurally generate the layouts and the puzzles.
You’d have to set up each puzzle with multiple solutions, using different resources. Allow for partial success on some paths. Really you’d want lots of treasure in each wreck. You’re trying to maximize the take and return alive, not find the single victory condition. (Although I guess you’d want a “boss treasure” in each wreck just because that’s more satisfying.)
In modern terms, it’d be “Captain Verdeterre’s Roguelike”.
With this setup, the supply-buying phase becomes a strategic challenge rather than a pick-the-right-choice puzzle. The shop offerings would be randomly generated too – with slightly randomized prices. (Otherwise there’d be an optimal strategy for that phase, which is boring.)
You could also randomize the date so that the tide table is interesting. :)
(But skip the “what day is it” puzzle.) Tide level can be a meaningful variable for your dive time – some areas are accessible only at high tide, or low tide, or with the tide moving a certain way.
There are also some signs in the code that there was supposed to be more to do on the journey to the wreck. Instead of just going to sleep and waking up when you’re there, you were apparently supposed to take shifts and look out for obstacles?
Expand this or leave it out. Honestly I’d say leave it out.
Returning to port to buy more supplies is a viable strategy. Your final score is profit divided by how many days you spent, or something like that. So going back for that harpoon gun is a big penalty, but maybe it’s worth it if the shark is blocking a really important area. However, sailing back and forth isn’t the fun part of the game – compress that into a simple “what time to launch?” decision.