Fun game.

First key = go into middle cell, and look right (you can’t use telekinesis until you find your wizards hat (so E when close enough).

Now, regarding the mouse tile selection. My technique is similar to this (http://answers.unity3d.com/questions/134412/selecting-and-unselecting-objects-with-raycast.html), but I’m sure there are other ways to manage it. Each tile has a MeshCollider attached to it, so I use a Physics.RayCast from the camera through the mouse cursor and if the object I clicked on is a Hex Tile (by looking at a component on the raycast hit object), then I’ve selected a tile!

Both of the code samples should give you the same result. The difference is whether or not the tiles are parented to another game object or not. In my case, I am setting each tile as a child of a single grid object (so I could for example shift the grid object around and all tiles move with it), so I represent the tile’s coordinates in terms of local coordinates instead of world coordinates.

In your code, by passing the position vector directly into the Instantiate function, you are declaring the world coordinates for the tile, not relative to any parent.

Here are some links that might help:
http://unity3d.com/support/documentation/ScriptReference/Transform-localPosition.html
http://answers.unity3d.com/questions/40439/whats-the-difference-between-transformlocalpositio.html

Keep at it! And send me links when you have something to share.

I have

GameObject tile = (GameObject)Instantiate(HexTilePrefab, new Vector3(x * spaceBetweenTilesHorizontal + (y & 1) * inradius, 0, y * spaceBetweenTilesVertical);, Quaternion.identity);

which gives me the same effect. Is there a reason to use these transform and localPosition code?
Sorry for my ignorance, I’m new to Unity.