AE Ratings Should Match Players To Arcs They Are Likely To Enjoy

I'll try to make this as clear and simple as possible.

The AE rating system should be designed with the sole goal of matching players to arcs they are likely to enjoy.

It should not be used to determine which arcs have merit on an absolute universal scale, nor to determine which authors deserve rewards and recognition, as these conflict with and detract from its proper purpose: to match players to arcs they are likely to enjoy.

I am not recommending any particular implementation, although I believe there are several that would do a much better job than the current one. I am only asserting that the sole, singular, entire design goal of the AE rating system should be: to match players to arcs they are likely to enjoy.

So how would you do this?
Like Netflix?

Quote:

You Five-starred this Arc! Players that enjoyed this arc also enjoyed these: <List>

It sounds like the system would have to retain the information about what you liked somewhere. I'm not sure how much storage this would take, but it sounds interesting.
I often hit the Random button and I have to tell you, the results are more often than not, pretty bad.

Netflix and Amazon are the models I had in mind. The system already stores which arcs you have played and rated, and IIRC even what rating you gave them, so that end of things would not be the implementation issue.

The real hog would be the recommendation data itself - in fact, generating recommendations efficiently is such a difficult problem that Netflix recently held an open programming contest to get a better solution.

Nevertheless, I really believe that development effort in this direction would make the AE a lot more useful and enjoyable for players. It would also help direct players toward niche authors who wouldn't otherwise get exposure.

In order for it to work, though, I also really believe that it can't be linked in any way to any sort of author reward system. Otherwise it will be manipulated, and made less useful as a result.