After looking into it extensively and much internal debate we have decided that multi-algo as it currently stands is flawed from a security perspective. While it sounds good on the surface there are some rather complex implementation issues in reality which have become apparent while looking at the code, that we are not satisfied can be solved in a reasonable way, or at least I am not confident I can solve them and I feel it is good to know your limits
I don't want to go into too much technical details but most of the flaws revolve around the fact that 'difficulty' is a somewhat arbitrary measurement, while it can be used to meaningfully compare two blocks from the same algorithm to one another, there is no real relation between the difficulties of two different algorithms. i.e. It is not really meaningful to say that a 500 difficulty Scrypt block is worth more or less than a 500 difficulty Groestl block. While it can be determined with some analysis for a specific moment in time with some work it requires a level of 'perfect information' that is not compatible with a p2p crypto currency type system, and even I as a human cannot tell you for a fact that the answer I give you now will still be the same in say six months time.
I have looked at various proposals to try and solves these issues, and what several coins have done and have found theoretical problems and flaws with all of them.