(Updated 20041011. Editing pass: clarified wording.)

There seem to be two points of view in this debate. The first is that the front page ought to be reserved for only the best content. The second is that the front page should emphasize the freshest content.

Why can't we say that the front page is reserved for the most interesting nodes, where a node's "interest" is determined as a function of both its "bestness" and its freshness? In other words, why can't newer nodes have an easier time getting on the front page? And, as front-page nodes age, why can't we require them to earn their continued placement in the prime real estate? If a node doesn't pull in enough reputation to say on the front page, let's take if off.

More specifically:

1. Let's analyze the database and determine the distribution of node reputations w.r.t. age for ages in the range of new (0 hours hold) through semi-recent (say, 96 hours old).
2. Based on our analysis, let's create a node-interest-index function f. Given a node n, let f(n) be the number of standard deviations above or below the mean that n's reputation is w.r.t. nodes of similar age.
3. Let's set the front-page threshold at X and populate the front page with nodes n for which f(n) > X. The idea is that eligible nodes (e.g., those nominated by human front paging) will be promoted to the front page only if/after their interest index exceeds X. Similarly, nodes on the front page will be removed when their index falls below X minus a hysteresis constant H.

The key characteristic of this approach is that its easier for newer nodes to hit the front page, but they only stay if they earn their place. At any given time, the front page will be populated with a smoothly blended and gently stirred mix of the newest nodes and older, higher-quality nodes – but everything should be of high interest. Newer nodes will tend to cycle through quicker, but the better ones will earn extended stays the front page.

This seems like a solution that satisfies both points of view. What do you think about this approach?