I recently skimmed Mark Suster's Embrace Opposing Views at Your Startup and his message is simple:

There is something you can learn from points of view you don't like.

Suster specifically says not to write off potential "fad startups" like Foursquare, Turntable.fm, or Pinterest; rather, ask yourself what your startup can learn from why people are interested in them.

Suster then references Albert Wegner's request for an opposing view reader:

In general, we seem to be building too many positive re-enforcement systems on the web. How about “agree” and “disagree” buttons? If all you can do is “like” or “favorite” items, it becomes very hard to express that you care about something but have a different opinion.”

I agree that it's difficult to have a conversation on Facebook or Twitter with people of opposing viewpoints because the rhetoric devolves so quickly.

Still, I do believe the Web needs better discourse.

That belief drives the software we're writing for 106miles.net and PandaWhale that you are using to read these very words.

In the meantime, keep thinking.

I've suggested for HN and similar sites replacing the unidimensional upvote/downvote with a two-dimensional vote. One dimension would remain up/down, rating whether this is a contribution that's valuable, or should be discouraged. The other dimension would be left/right, for agree/disagree. (It lends itself to a small polar-rose presentation, though it might also make sense to have the two axes more distant from each other to avoid mis-clicks.)

Instead of a net total for agree/disagree, the agree/disagree subtotals would somehow be indicated. (A comment with +101 agrees and -100 disagrees is more interesting than one just at the default starting '1' net.) I think a little red/green sparkline could work pretty well, perhaps only visible on curiousity-hover.

The problem with a unidimensional feedback is that too many signals are being crammed into one crude channel. Disagrees become downvotes, which in HN and similar systems, actually suppress the prominence of viewpoints, and cause a 'sting' that makes the forum more prickly. I don't believe any amount of training can cause a lazy, hurried reader to refrain from using the easy click to signify disagreement... but providing a more precise outlet might attract the click instead.

I've felt this too. On my blog I post studies that can be downers or are quite controversial. Facebook's choice to call it the "Like" button instead of, say, the "Interesting" button directly affects me. People may read something fascinating about injustice, genocide or rape but, understandably, very few want to hit a button beneath it that says "I LIKE THIS."

Half a decade ago my colleague and I came up with a new idea for a social network: antisocial.net. It started out as just a way to +Hate something, but eventually we came to the same conclusion. It would have allowed a million shades of grey for why someone liked or hated a product, person, place, or thing. It was modeled after another site about "the world's worst bosses" that got sued into oblivion. We thought we'd give badges for being the worst in each category. Laughingly one of the badges we considered was a Devil badge. There's an old saying, "the he must have been very handsome and charismatic to be able to fool so many people". We figured we'd give it to the person who had the most likes and hates at the same time.

I've been thinking about ways to get people to indicate insightful/quality/"adds to the discussion" rather than like/agree for a long time and I really like @gojomo's idea of two dimensional voting. I wonder if people would understand and use it though?

If I built it, I'd probably put what people currently tend to do (agree/disagree) on the axis they are used to (up/down) and use the new axis for the new feature.

I assume 106miles and Facebook have a single voting direction like/null or props/null for the same reason, to emphasize the positive and to keep people from being able to suppress things they don't agree with. The "Say something nice or don't say anything at all" philosophy.

For sites like HN/Digg/Reddit that have up/down voting I suspect they could approximate the two dimensional data by taking posts with significant numbers of both up and down votes as being more interesting than simply the sum of the two numbers. A post with 100 ups and 200 downs may well be an interesting but challenging idea. A post with 0 up and 100 downs is presumably complete junk even though they have the same raw score.

The Reddit and HN ranking code is public and they don't do that but Reddit do have a controversial list separate to their popular list. Digg are more secretive about their ranking, so who knows what they consider.

I've been thinking about ways to get people to indicate insightful/quality/"adds to the discussion" rather than like/agree for a long time and I really like @gojomo's idea of two dimensional voting. I wonder if people would understand and use it though?

If I built it, I'd probably put what people currently tend to do (agree/disagree) on the axis they are used to (up/down) and use the new axis for the new feature.

I assume 106miles and Facebook have a single voting direction like/null or props/null for the same reason, to emphasize the positive and to keep people from being able to suppress things they don't agree with. The "Say something nice or don't say anything at all" philosophy.

For sites like HN/Digg/Reddit that have up/down voting I suspect they could approximate the two dimensional data by taking posts with significant numbers of both up and down votes as being more interesting than simply the sum of the two numbers. A post with 100 ups and 200 downs may well be an interesting but challenging idea. A post with 0 up and 100 downs is presumably complete junk even though they have the same raw score.

The Reddit and HN ranking code is public and they don't do that but Reddit do have a controversial list separate to their popular list. Digg are more secretive about their ranking, so who knows what they consider.