I wouldn't bother with a "fully fleshed out" GDD, but I advocate having at least a high level design mapped out. No need to go into details (the end result is ALWAYS completely different to the original formulas) but a high level design document (or scope document if you like) makes sure everyone has the same picture of what type of game you're making.

1. GDDs are really overrated ! They are a good way for unexperienced designers to get their mind right, or to write down the basic game play for larger productions, but most of the final game will be developed during ..well.. development time.

While I am just a novice game designer (working on my first title now) I have to disagree that GDDs are overrated. They are an essential part of developing a good game as they allow the team to all have an idea of what direction the game should be heading in. Of course things will change from the original document as you prototype, test, and iterate your games, but that is exactly what the version history section of a GDD is for. It helps to show what has changed and why for the betterment of the entire team, not just the designer who "needs to get their mind right". The designer has an important job in making certain that things are heading in the right direction to ensure the game turns out like the company (be it indie or big name) wants it to. Documentation of the teams goal and the endeavors they have made in producing games is not a thing to be taken lightly.

2. Journal: very similar to the first part, it helps to write down your mind.

Agreed on this point, as a personal journal is more for the individual to look on their own ideas for reflective purposes, or to ensure they don't forget a valuable piece of information or idea.

3. Prototyping: yes, this is a good idea

Once again, agreed. Prototyping is an absolute must in the cycle of iterative design. Without consistently testing your mechanics, levels, and even looking over graphical elements you are almost guaranteed to make a sub-standard game.

4. Cheats: they are necessary, but they are really dangereous. Using cheats is a way to avoid certain game play, the danger is, that you start testing the end-game only and negclet the start of a game. But the start of the game is the most critical part for new players.

This I have to disagree with to an extent. While some cheats can be fun to play with if you just want a relaxing playthrough where you mercilessly slaughter everything in your path (God-mode) or you type in a few keystokes to earn unlimited money to build your dream city in a simulation, cheats that are put in to obviously cover up bad design should be frowned upon.