Not enough people check the new page. It's impossible to get an article on the frontpage unless like four or five people upvote it in the first few minutes after submission. This is extremely unlikely, unless you have a popular title or popular username.
Once something gets on the frontpage, it gets a LOT of upvotes, and the algorithm decays its weight so slowly that it is essentially a rich get richer phenomenon.
Possible solutions:
* Front page upvotes don't count as much, since there are more views. In particular, the "weight" of an article should be the probability that someone who views the title clicks on it, combined with the probability that someone who clicks the link decides to upvote it.
* Frontpage is stochastic. Instead of being fixed, you sample articles based upon their probability (or score). Each person gets a new frontpage, in order to actually explore which articles are good. (Exploration vs. exploitation)
Not enough people check the new page. It's impossible to get an article on the frontpage unless like four or five people upvote it in the first few minutes after submission. This is extremely unlikely, unless you have a popular title or popular username.
Once something gets on the frontpage, it gets a LOT of upvotes, and the algorithm decays its weight so slowly that it is essentially a rich get richer phenomenon.
Possible solutions:
* Front page upvotes don't count as much, since there are more views. In particular, the "weight" of an article should be the probability that someone who views the title clicks on it, combined with the probability that someone who clicks the link decides to upvote it.
* Frontpage is stochastic. Instead of being fixed, you sample articles based upon their probability (or score). Each person gets a new frontpage, in order to actually explore which articles are good. (Exploration vs. exploitation)