I've been playing Scrolls, but I think I might have hit my end.  It's a decent game, but it's just too freaking random.

1) You can only have 3 of any given scrolls.  The standard that I hold deck based games to, Magic, has a max of 4.  The decks are larger in Magic than in Scrolls, but any given 4 of is still more likely in Magic.

2) Due to the paucity of cards (total and available to a player) in Scrolls it feels even more restrictive.  I might only have 3 one drop creatures in my deck in Scrolls, where in Magic I might be able to have 8 or 12 cards that fulfill that purpose.

3) No mulligans.  This is probably the worst of it.  If you get a starting hand of no creatures, you can be pretty screwed. Comebacks are possible, but are very difficult.

4) No choice on playing or drawing first.

All in all, it's a decent game, but still a bit disappointing.

The problem with Fez is that it goes from being a nice, low-stress exploration platformer to a tedious puzzle game.  Like, once you understand 2 or so of the cyphers in the game it becomes "Go to place A, write something down, Go to place B [or stay where you are], Correctly input long string of commands."

Holy shit but not have root access makes things so much harder sometimes.

What do you mean by rigidbody?  There are a lot of flavors of rigidbody in Unity.

So, here are my quick thoughts:

1) Depending on the collision geometry that makes sense for your game, use a character controller where radius > height/2. Then rotation doesn't matter (at least for the character controller, you'll obviously still need to handle it for your graphics, camera, etc.).  The only bad part of this is that you lose the use of the convenient collision flags/isGrounded().

2) A trigger rigidbody.  Handle all physics/collision restitution yourself.

Also, it doesn't really matter.  A* is old hat and it doesn't really matter what heuristic you choose (so long as it is admissible, obviously).  It's a far more tricky/interesting/taste-driven problem of converting the results of A* into something that looks good and believable.

The way I did it was to use the shape he shows as a template and draw your tiles on top of it (obviously scaled to whatever size you want your tiles to be).  Have an additional layer with each tile numbered for debugging purposes.

My workflow was:
1) Make tiles.
2) Test out tiles.  If something didn't look right, load in the debug numbered tiles to quickly pinpoint the problem.
3) Repeat 2 until it looks like what you want.

You don't have to have numbered sheet, but it can sometimes be hard to figure out exactly which of the tiles is being used.

I'm sure one could make an actual dedicated program or a plugin for gimp or photoshop that does what you want, but I don't know of one in existence.

Maybe one of the methods covered HERE.

That is a good article.

Even people who wing it probably have some sort of design goals they are working for.  I mean, they might not have a written list, but I sincerely doubt anyone gets too far without some idea of what kind of game they want to make.

Yeah, honestly, the best one is Succubus/Lilith.  If you went with more of that style, I think it would definitely be better.

The only minigame that's ever felt good to me is reloading in Gears Of War.

I can't imagine anyone doesn't set design goals before starting a project. 

so this is a bit for people who use OGMO level Editor. I added Levers into the game and made it so that each Lever has a Node. When the Lever is thrown an activator is created at that location and that activator allows it to interact with anything I want. In the new test the lever will drop a bridge for you.

The next step is to have a Wire object that you can place on an object and then place a node on another object to completely modularize it.

That is gorgeous.

Derek Yu.  Spelunky drew me to here, which trapped my soul forever.

In terms of people who do good work and inspire me:

Kenta Cho
Terry Cavanagh
Increpare
Smestorp
Brian Walker
Tarn Adams