TechEcho

Over the past couple of months, I've been considering the problem of the decline in comments on HN, against sites where I've seen the quality hold over time. As a result of this, I've come to three conclusions:1. Community size (that a user can see & engage with) must be artificially limited (although total community size doesn't) 2. A weighted voting process needs to be in place 3. To scale, the controls must be algorithm-based, not moderator-basedTherefore I've come up with the following system that I'm going to be building and trialling on a site in the near future:1. When a user joins, they're initially put into the city population. Here they are shown the firehose - an unfiltered news stream, with all users present2. Each time a user interacts, either by adding a comment, posting an item or clicking a link, the action is recorded3. Over time, the accrued weight of this will ensure that what they're shown is limited by the types of content they've historically interacted with, the users who've submitted that content, and the users they've engaged with in commenting4. To ensure freshness/adaptability, a certain amount of the hose will always be shown, and the selection will have a time component, so that they can view different content and thus the selections can change as the user changes5. Users will only be exposed to other users who loosely fit the profile that they've selected in to, with a maximum of around 150. Users will be able to follow other users, thus ensuring that these people are never swapped out of their engagement circle, and all content or comments that they submit are presented6. To solve the obvious issue of thousands of people following single users, a power user will see content from their own followed list first, then other data belowLooking at the content side:1. Each item would be tagged according by key terms in the content & the user who submitted the item2. Each submission & comment would allow voting, thus allowing for segregation between casual browsing and active engagement, to allow for more refined user-level filteringThoughts/suggestions appreciated.

2 comments

Mzabout 14 years ago

1. Community size must be artificially limitedThis might end up being a good way to kill the community. If something stops growing, it typically starts dying. A better goal is to try to find ways to problem-solve such that it supports healthy growth rather than willy-nilly growth. Supporting healthy growth may also have a side effect of slowing growth, which can be good, but artificially limiting size is likely to be a very bad idea.2. A weighted voting process needs to be in placeWhy?3. To scale, the controls must be algorithm-based, not moderator-basedRules-based crap that isn't adequately supported by the wisdom of good human judgment tends to lead to the sort of bureaucratic nightmare that people whine about concerning the government (et al). So I doubt it.

评论 #2441692 未加载

mooism2about 14 years ago

What you refer to as the firehose --- what people are exposed to in your scheme when they first register --- is also what is shown to people who have not registered. They are people who we would either like to register and join in (if they would improve the site) or not register (if they would detract from it). We want them to either be attracted or repelled by the mix of links and discussion on offer.If all that we see is personalised to us, we lose the ability to shape what they see, and we lose the ability to attract people who will contribute to the community we want to see.

Show HN: My thoughts on solving Eternal September

2 comments

Show HN: My thoughts on solving Eternal September

2 comments