NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
From: John Clements
Date: 2021 Feb 20, 23:31 -0800
Yep, sounds like fun!
A few thoughts:
1) In the "related work" category, you may already be familiar with Leonard Gray's "100 problems in Celestial Navigation," an entertaining sequence of problems that model specific trips, and that allow you to work sights given data (by no means always correct) and compare your answers to those in the book.
Naturally, this isn't quite what you're proposing; I think the key difference is the idea that your track might depend on the decisions you make about course, and that you might run aground--virtually, of course--as a result of bad navigation.
2) As a software developer, I know that projects like this live and die on the motivation of the developer and the interest of the users, and so I like to focus on the "MVP", or "minimum viable product." What would be the absolute minimum you would need in order to make this run, and be amusing?
I think it might look something like this: model a trip between two initially fixed islands in the pacific (modeled as circles of radius, say, 20 miles). Build a near-trivial wind model (assume a distribution of possible wind directions and speeds, pick one every hour). You get to interact with the model every hour. It tells you approximately how fast you've been going, you specify the course you want to sail and whether you want to shoot any bodies. You specify the bodies and the approximate alt/az pairs, and you get readings (with a suitable amount of noise). After some number of days, if you haven't yet reached the destination, you're assumed to have run out of food and died. Basically, the "oregon trail" of celestial navigation.
There are about a million things to add to the model to make it more interesting. One of the major ones would be interesting and realistic coastlines. Weather conditions would also be important. But for the MVP, you might be able to get it running in ... a week? Note: no graphics required :), but you would *absolutely* have to have a pretty good model for the actual positions of the specified celestial bodies.