Self policing for "delete me" posts

There were some mistake submissions which got deleted from the front page. They were immediately obvious as mistakes, but what if a dupe was submitted a week later? a month later? ten years later? A dupe may leave and re-enter relevance given the circumstances. In a different context, a dupe may not be a dupe, even if it references the same information.

For the last two cases I saw, "delete me" works, but I find this self-policing idea suboptimal. There will inevitably be users in the future who make mistake submissions. If they are completely irrelevant, or carbon copies, fine. But if they are submissions like "AMAZING PHOTO!", what if the user finds it personally significant? As the size of the crowd grows, no matter how focused your initial crowd is, or how you intially positioned your product, there will be a scattering of interest. Unless you follow an editor and moderator model, slashdot for example, you will be unable to filter your content to a specific standard. In the case of an open-participation community, enforcing focus may be excessive: after all, the community is defined by itself, not the application it uses. If you gather more and more forgetful users, they may appreciate dupes.

After you hit some sort of critical mass of users, many new users will arrive by friend recommendations, or mere interest in the community size, rather than independent discovery. Independent users join after searching for and spotting something tailored to their specific interests, but there is no guarantee that referred users fall under the same profile.

In interest of monetization, it should be hard for popular sites to escape changing to accomodate a more diverse userbase. I believe this has hit both digg and reddit. Sooner or later, with the rapid growth seen here, news.yc also (which seems to evolve faster than the other news sites). In this regard, plus 1 for slashdot model (editors plus moderators). python_kiss also observed this on his project (interested in your thoughts). This also makes me question the strategy of pre-defining and then trying to self-enforce a niche in an open and user-driven website. Of course you can use subgroupings by topic, or do something like Ning, but lurkers that never bother to customize their preferences, whose numbers are probably significant, will be hard to please. You might even have better luck with them by showing random stories than by showing the most popular or most active.

Maybe the popularity measure has shown its limit in usefulness. Users come for content quality, be that quality of information or emotion (flamewars give belligerent folks an adrenaline boost); the quality is driven by a few. I would come back here just to read the comments from a couple amazingly good commenters. Those comments could spawn complete discussions by themselves, like the questions from good panel moderators in conferences. More than popularity, then, the submission rating should be influenced more by the overall "quality" measure of the discussion. For example, in slashdot, you look at the number of comments, then the total number of, say, +5 comments. Usually it's something like, say, 10 percent. When it's significantly higher, I anticipate a good read.

That reminds me again. Topic Karma is an interesting attempt. python_kiss? :)