well what would you do?

Without planar rifts, the only way out is probably to cast a dank pit site somewhere off that island and hope for the best.

Yeah, I think the map generator still needs some work, either by tweaking the algorithm or doing a post-processing step to reject bad maps. In particular, from a competitive perspective it seems bad to allow:

- starting on an island (check that their exists a path from some space next to each fluxwell to some space next to each other flux well)

- buried in slow terrain (almost as bad as an island). I've had maps where it will literally take my creatures 6+ turns to make it out from my base to somewhere useful (the center was an ocean so I had to go to the corners. This is especially bad when they are all forest and your opponent is sylvan.... (check the total movement cost of the shortest path used for the above calculation and set some reasonable threshold)

- having both close to the center spaces (the ones 1 space towards the middle from the flux well be water (not sure if this can happen or not). The positional disadvantage of your opponent needing only two vision two buildings to make it to the center and you needing more is big against fast decks.

I've seen worse. There ways to get off that island with pure dark.

You can build in sight of the cat.
Two vision buildings can insta-build off the island.
And of course fliers and aquatics can make it off pretty quickly.

The flip side is that I love discovering islands early to mid game to grab. And when you opponent gets stuck on an island like that, it's not nearly as bad.

You should've build on another side

About map generator: I already changed it so it always check that you not completely screwed by map.

About starting on island or in deep area (many forests/mounts/etc) - you (and your deck) should be ready for such cases. Another way to avoid it - start on another side of you see lots of water or deep terrain.

I won most games when I was starting on small island with MF deck (that as you know don't have many terrain changing things).

kash; honestly, Altren is right. you and your deck should be ready for such cases (or hope not to get into those - if do so, I lost the game.. (btw)), as if you want fair chances for both players, you can play chess instead..
if the opponent gets creapy terrain you'll be for sure happy.. so you must calculate with the option that you will be the opponent here..

Altren; what changes have you made? as 'not completely screwed by the map' is pretty wide.. and because I really liked it the way it was.. even though I lost many games thanks to the bad terrain...

I don't think it is really a matter of being ready for such cases. I do look at the map and if one portion looks bad go for the other side. But in plenty of instances from what you can see originally things don't look bad or both sides look bad. I like the randomness in maps and that you need to be ready for a variety of terrain. It does make you take into consideration things like packing cloudbursts if you want to play ocean cards. But really, the only way for a deck to be ready for deep forest is to either be sylvan or elemental with a lot of flyers. Otherwise your early game is at a disadvantage.

I don't consider any of the cases I mentioned impossible, and a good player can beat a weaker player despite the disadvantage or with the right type of deck turn it into an advantage (sylvan treent decks love starting in deep forest, vapor mahal decks often don't mind starting on an island as long as they can play one in the right spot to get off). However, as a philosophical matter, I prefer games to be decided mostly by the skill of the player and less by randomness. So I do prefer things towards the chess end of the spectrum (although not all the way, pure determinism is boring). So I do find dropping my first creature, seeing the map, and thinking "well that sucks" fairly annoying.

I think you should extend the visible area at the beginning of the match by one towards the center.

I've experienced maps that sucked for both players, but easily came out on top because I was playing with a more agile deck that could handle some terrain issues, while they were 'stuck in the mud' as it were.

But in response to your question-- I would planar void, or cast a 2 sight building and use the area of control to jump across the pond.

Against a good opponent? Say "gg" and resign.