I've tried this before and I find it difficult to trust this "other self". What works better for me is to treat "other self" as an information source and then have "real me" make a decision with this information in mind.

The "real me" is sort of like a benevolent dictator. It has to have the final say, but ultimately it's capable of deferring to "other self" when appropriate. Maybe this is just a thing with me ("real me" always has to have the final say, I can't just trust someone else).

Of course, this is just one data point. Other people very well may be different.