FWIW I think the multi algo works like this.
You set your miner software up with either multiple pools or multiple calls to the miner software in a batch file where each call runs a different algo. All the pools/calls use the SAME multi algo port and id details.
The Multi-algo port detects which algo you are using and routes you to the correct pool. IF the algo is not the most profitable of the set then the Multia-algo port closes the connection, this then makes the miner move onto the next pool (needs failover-only set) or causes the miner command line to exit and run the next one in the batch file.
So you configure the miner software to loop round trying each algo in turn against the multialgo port, the port will reject/close the connection if that is not the most profitable algo of the set. your miner MUST support all of the possible algo's which is why there are different ports for different algo sets.
This is a good explanation, much better than on the FAQ page. I like MPH but the frontend is a mess in useability terms and it could definitely do with better explanations. It would save a lot of newbie questions and confusion here. Not being able to easily see at a glance what you are mining at any given moment desperately needs to be fixed. I understand however that big improvements are pending.
Definately better then the FAQ.
except its wrong about many things (djmuk's theory)
Are you saying the FAQ is wrong? or djmuk's theory? Cause from what i can see his theory is absolutely correct.
12xxx ports will only allow you to connect to whatever algorithm is most profitable for what you have set in the hub section of your account. All others will not accept you connection.
The batch file loops through miners until it finds the one you can use.
17xxx ports will let you connect whenever and will switch coins, but does nothing to switch algorithms.
The FAQ actually makes sense configuration wise, all we really need is a clear explanation of what and why its doing it.