I seldom find places that can be reached but cannot be routed.
Many beaches connect path or tracks and can be easily walked.
Issue is part of foot routing and was already raised about city squares: AFAIK most of them are actually routed along the perimeter, as there were no other chances for pedestrians.
Wherever possible I trace Strava popular routes that usually follows the natural ways.
But squares most of times don’t change so much the distance as beaches do.
How to map these situations?
What router(s) are you most interested in?
(edited to add) if it’s Garmin routing that you’re interested it, although I’ve never tried it I’d expect it’d be pretty easy to add “cross-beach navigation” (by routing around the edge).
Bin mir nicht sicher ob es hier nur darum geht wie man am besten über Plätze, Strände routet.
Am einfachsten ist es die Wege über den Strand mit einem highway=path informel=yes zu verbinden.
Organic Maps, Osmand, opensroutingservice, all of them don’t route beaches. I didn’t test others.
The first two are “all in one” apps that have built-in routing. I don’t know about Organic Maps, but most aspects of OsmAnd are ridiculously configurable; it wouldn’t surprise me if routing was too. You’d been to ask the app developers that question though; it’s not really an OSM data question. OpenRouteService (which I presume is what you mean) has a bunch of examples, but these aren’t just clicky pointy websites - you’d need to create a bunch of rules yourself.
Looking at this in a sort of “meta” way, any natural=beach polygon in OSM might be assumed to be “infinitely navigable” (by foot, anyway), similar to a (polygon-denoted) highway=pedestrian + area=yes where our wiki claims “pedestrians can travel freely in all directions.”
There might be cases of natural=beach where this isn’t true (maybe not easily-foot-traversable sand, but cobble-stone-like surfaces which are difficult or impossible to navigate by foot?) and therefore, “not” (to all this). But if you think of “beaches are where pedestrians can travel freely in all directions,” whatever routing you are doing “opens up to infinity” as soon as you cross the polygon “into” the beach.
I don’t know if this helps, but sometimes a “meta” way of thinking about how you might do something (like routing) can “snap to an answer quickly” with just a small gear adjustment in your thinking. And so something as simple as starting with a baseline assumption of “beaches are de facto foot-navigable for routing purposes” can “just work.”
Maybe you look at further surface=* tags and come to a “no, not foot-navigable” conclusion…to be determined.