My mind often wanders over a 'laundry list' of possible interface innovations to improve social news site interactions. A few of the items:<p>(1) <i>2-dimensional comment rating</i>. The 1-dimensional upvote/downvote conflates 'advances the discussion' and 'agree': there's no simple way to register, "good comment but I disagree" (other than another written comment). This makes some simple disagreements earn downvotes, and because downvotes have a connotation of censure (and even censorship as the comment descends in visibility), increases negative/adversarial feelings. Hiding the net comment scores has helped a bit (with other costs), but adding a 2nd-dimension -- a little compass-rose -- to ratings would let up/down be "worthwhile/unworthy" while left-right could be "agree/disagree". And, the separate agree/disagree counts could be shown, the give a sense of magnitudes rather than just net differences. My theory is that by providing an outlet for non-suppressive disagreement, fewer full replies would be necessary. Everyone could 'weigh in' without so much of a corrosive sense of status-retaliation/tribalism.<p>(2) <i>separate parallel stream per-item (or per-comment) for 'carping'/correction comments</i>. Often a headline sucks. Or an article or comment has blatant logical, factual, or grammatical errors. The community can't resist racing to point these out... and to some extent that's necessary, and can result in an headline/article/comment improvement for clarity/correctness. But it's also somewhat distracting and low-value, especially if the carps persist after a correction makes them redundant, or the carps become more prominent that substantive points. So give them their own tab. By convention, corrective/meta comments should go there, and moderators (or community votes) could also move misplaced comments there. My theory is this would retain their corrective value (especially for those most interested in that sort of precision) without clogging the 'main' thread with their bulk and hypercritical mindset.<p>(3) <i>community rewritten/ranked headlines</i>. Abusive headlines often waste readers time, create unnecessary discussion tangents based on misconceptions, and bias talk in a more adversarial direction. But corrections by moderators are inconsistent (and themselves often controversial). So maybe let the community propose alternate headlines, and vote on which is best. Only the top headline, presumably improved by group action, appears on the summary list views... the alternative proposals on the detail page.<p>(4) <i>subheads (aka 'dek' or blurb/tease)</i>. Allow a second-line in submissions for display in list/summary views, as is common in many journalistic presentations. More context can help save time/attention. Because writing these can be as challenging (and subject to the same abuses as headlines), if the headlines can be community-corrected per the above, the subheads should as well.<p>(5) <i>event/topic clustering (a la Techmeme)</i>. Interleaving 10 stories about some attention-grabbing new release or controversy with 20 stories about other things heightens a false sense of novelty/urgency but doesn't lead to more coherent evaluation/discussion. All the stores about "Company X releases Y" (and similar) could be grouped as a contiguous block. The block might rise/fall on total votes to all stories; the relative prominence of the stories within the block could be based on their individual votes. They'd have one comment thread, avoiding redundant comments (and comments like, "as I mentioned in the other thread"). The groupings could be moderator-controlled, algorithmically-controlled, or even community-influenced.<p>(6) <i>ignore 'sunk time' in ratings decay</i>. Consider story A, submitted at an inopportune time, say time 0. It starts with one vote. Over the next 6 hours, 4 more votes/resubmissions trickle in. It never hits the front-page and quickly leaves the 'new. Now, at hour 6, elsewhere on the net, story A gets more attention. It's wrapped/aggregated as story B, which is also submitted. In the next 15 minutes, say that A gets another 25 upvotes, while B gets 20 upvotes. I have the strong impression that B will shoot up to the front page -- 20 votes in 15 minutes! - while A will still languish in obscurity -- merely 30 votes over 6 hours. And yet, for the overlapping, recent 15 minute period, story A got more votes. I suggest it ought not be penalized for the prior hours it spent in the doldrums, before it really 'broke' more widely. The right decay function (looking back from now rather than forward from submission time) could make this work. My theory is that removing this extreme recency bias would mean some slow-building stories, from more foundational sources, would get better treatment than well-timed later sensationalizing/oversimplifying rewrites.