Mate in 6 [own composition]

The solution is : 1.Qc1! Ka2 (Qxc1? 2.Ra4+ Kb2 3.Rb8#)

   2. Rb8! Qxc1 (other black moves lose quicker) (Bxc2?! fails to ...Rg8+ intermezzo)

   3. Ra4+ Qa3 4. Bxb1+ Ka1 5. Rxa3+ Ra2 6. Rxa2#

@arisktotle Pretty happy with this one. Quite a few thematic tries, and nice key moves. The final mate is also rather pretty. 

I heard someone say longer mates than mate in 3 were usually easier to solve, and now I can see why! there's bound to be a lot of restrictions you have to place if you want just one solution.

It's hard for me to spot how to improve this one further at this point but I'm always open to hear what you think. (the pawn on c6 is the only piece that doesn't strictly contribute to the solution but it removes some ugly lines/options for black to check the white king)

Well, I looked at it for a while and I very much liked the queen sacrifice! There is however an issue with the move order. On the 3rd move white can pick 2 moves which in the end turn out to be interchangeable and lead to the same mating position. To a game player that doesn't matter but in a direct mate problem (which #6 is) it should be avoided. The management of move orders is one of the most important and difficult tasks for a problem composer. And that's true for all problem types, also for endgame studies.

Unfortunately, it is not always easy to fix a dual move order. I looked for a while and I couldn't find an obvious fix. There may be one though. Some work to do.

Note that it is harder to keep your solution dualfree with long solution sequences. For checkmate problems, mate in 6 is pretty long and therefore harder to control than say a mate in 3.

Thanks to you saying that I found a very nice idea almost immediately. The "obvious way" is to make them move one way is clearly to put one in front of the others way. Not only does this solve the dual problem, it also solves the Bxc2 cooked solution which I added the ugly rook on g5 to deal with, but has now become obsolete. Cleaned up some more pieces and added some tweaks to make sure Qxe4 doesn't come with check and Rxe1 wasn't a quicker mate. This should be totally dualfree, and it has less unnecessary pieces

Thanks for the help, since I was kinda stuck on this position before you pointed out the dual.

Ah, that is a clever fix! You are thinking like a composer!

In general, I am very happy when somebody finds an error in one of my problems. Errors are an indication of missed opportunities. When you repair them you often end up with a better problem than you had in the first place (even without the error). So you did exactly the right thing!