Maybe if you ran some kind of candy A/B Test.
Send two different kinds of envelopes to people, and keep track of who got what. Then, send a survey to these people to have them rate the candy's they received. After a few months, you should start to have a general idea on what types of candies are liked more and what types are liked less. Even though you'll never have a taste-consensus, you can get a general idea, and modify what kinds of candy you send and the amounts of the kinds of candies you do send.
Or, another way would be to send all the same candy to everyone and send out a survey. Then the next two weeks send another set of candy and run another survey.
Whatever you do / don't do, getting stats and info is really nice.
You can't, unfortunately, please everyone :( so might as well please a group of people _a lot_ and get a group of really evangelical fans, I think.
If I totally abandon the community aspect though, then I could start designing the ultimate sequence of candy that maximizes retention. Not sure which is better.
I'm not sure how to only send the good ones, and that would depend on the recipient's tastes as well. Part of the idea is to try new things, so it seems unavoidable that as I go through all the varieties out there, people will not like some of them.
Some people don't like gummy candy, some really do. Some people are really into "construction kit" type candy, some hate the effort involved and so on.
Perhaps there could be a way to make the subscribers feel like they didn't lose out even if they got something they really hated, but not sure how that would work exactly.