The General Manager of Reddit told me that he thinks Subreddits are successful, in part, because the service displays so many statistics (# of members, uniques, etc). As a result, redditors try their best to optimize them and the communities grow.

Similarly, we all try to optimize the number of Followers we have on Twitter. And the number of comments you had on your Myspace page was a BIG DEAL, way back when.

How important do you think it is for a social platform to give its users a number to optimize? Even pageviews on WordPress and LinkedIn are super addicting.

We purposefully left one-click interactions out of the Branch UX – "so that you're not influenced by popular opinion" – but I wonder if that's a mistake...

I think numbers are gimmicks.... they might work for a while / help you bootstrap sometimes, but ultimately their power runs out and frequently they end up distorting the primary value of systems.

E.G. I actually wonder what twitter would have looked like without a number early on -- there is an argument that they would have less of a pollution problem to pay down (if they actually do have one at scale / i feel that way as an individual user, but obviously others on this branch know better)

Numbers/scores/feedback can be a valuable method for helping create "liquidity" on a platform. I do think it's important to think about the long term implications of the numbers you choose. If you can pick metrics that leads to behavior that are consistent with the long term dynamics you want to create, then they can be really powerful. I don't think that Twitter follower counts are a good model for that, although they seem to have been valuable early on.

Sam and Satya –- you both seem to think that Twitter's Follower metric, though helpful early on, has resulted in a lower quality system. But I would argue that metrics like "Retweets" and "Favorites" deserve more blame for that, no? If you agree, isn't Facebook equally guilty with their "Like" button?

And I'd argue that, because my Follower count is so prominently displayed, I tweet "better" and more popular content because I don't want to lose followers... Though that doesn't necessarily mean what I tweet is higher quality...

It's definitely a useful tool and can help drive good behaviors if designed well.

Follower count was obviously useful for Twitter, as it got those who cared about it competing in a way that drove growth for the whole system. (Turning on the TV in April 2009 and seeing Anderson Cooper telling viewers to go follow @CNN on Twitter, so they could beat Ashton Kutcher to 1M followers is the kind of advertising you can't buy.)

However, we purposely didn't built an actual leader board on site, because we didn't want to imply that's what it was all about. Ultimately follower count is a poor proxy for quality and a pretty indirect feedback mechanism.

Josh, your take on follower count vs. retweet and fav count is interesting—and opposite my intuition. I think it'd probably be better if what was highlighted was people's actual reaction to the tweets rather than their overall popularity (which is often more a factor of their offline fame than their tweet skills).

I guess I don't buy that "tweet skills" produce the type of value that I [personally] look for. Sure, optimizing for retweets and favorites will lead to more "popular" content, but I'd much prefer to just hear what's on your mind than what you think will be popular.

David Galbraith has 1,500 followers. That guy deserves a lot more, based on his experience and knowledge, but I guess he's just not good at tweeting...

I think two things are important to your initial question. Short content like tweets, reddit posts, tumblr posts, etc. is quickly evolves on a platform and a point based feedback system can be helpful as a feedback loop to the authors about what content is good. This is why you will often see memes cycle through reddit, which has the benefit of keeping content both familiar and fresh at the same time. For something like Branch I think it might be less applicable. What would you be giving feedback to? A specific post? An entire branch? I don't think it would be very helpful to either the person giving the feedback or the person receiving it because it would be overly vague. Would "like" buttons improve wikipedia.

Secondly as Ev said, I think "points" definitely add a competitive and endorphin releasing aspect to your platform. Which helps people talk about your product rather than just use it (or try it out). But it is important to not get caught up in any single number as a metric which represents value. As Ev said making a leaderboard would have implied that "followers" is what Twitter is about, making the follower count one of the most important metrics on the site, which does not align with the product. Instead you have to look at many different metrics (and maybe "points" isn't one of the important signals) when trying to surface quality content on a platform.

Also this is a daunting participant list to be added too O_O

"Would 'like' buttons improve Wikipedia?" That's awesome. Spoken like a veteran tweeter.

To your second point, you're right, Favs and RTs are like crack but that doesn't mean they're good for you (or the content they lead you to produce).

More interestingly, I wonder what Twitter would have (/would) look like if there was a (topical) leader board? Or a Billboard Charts-equivalent for tweets?

I think 'numbers' are an abstraction layer for essentially what is "how does this tap into the ego of the user'. They are important from the perspective that they give the users a reason to come back for. However, at the same time, they need to be very very carefully crafted since they influence the kind of community you will grow over time. In a not so great implementation, I believe, they look like karma points on forums - a mad race for those in front and an intimidating and often irrelevant number for those far behind in the race.

I like to think of online communities as a barter system. So the simpler and often necessary question is when the user invests their time and often ideas (in the form of content) or actions (in the form of upvotes, favs etc), what does the system give back to them besides non quantifiable intellectual or emotional upliftment. This 'number' is just that.

Also regarding the whole follower count vs good tweet discussion: I feel this depends a lot on the network. Likes are way more important IMHO on Facebook than friend count or even subscriber counts but not so much on Twitter.

The other interesting question is how do these numbers affect the behavior of those who are not the 'leaders' in this virtual leader-board in the long run.

People are looking for an indicator of validation on a network. Whether numeric or an icon, it reinforces the sense of progress. A number is easy, though FourSquare was different by expressing status in categorised badges.

Before centralised social networks people invented expressions of status for themselves, independently: published graphs of their Apache logs; top incoming referring domains on blogs; and then FeedBurner's widget *was* a numeric expression of status/achievement.

But the targets are transient: Referrers became a spam vector, follower count loses its personal connection. I think each is a short term guide, but what if networks aggressively disposed of them. What if the follower count just disappeared when you reached 500?

Or instead of disappearing, what if it had incremental as well as decremental aspects. Would that lead to a self correcting system or just add complexity?

Today, I noticed that a sports blog embedded a branch: sportsclt.com I'd never heard of it but quickly deduced that it wasn't a big "win" for us because there were zero comments and zero tweets.

Tuhin's 1.6k followers helped me feel comfortable accepting his request.

The number of Likes a Facebook post has helps their system decide how high on your Newsfeed a post should appear.

The reason I don't like Favs and RTs is because they don't convey value (i.e. they don't change ordering of tweets). They are merely dopamine hits, and don't help outsiders gauge value like the number of Followers do.

Do you ever open a tweet before you read it to see how many Favs and RTs it has, and then read it based on the number? No.

Ben and Tuhin – your posts do make me wonder if there are diminishing marginal returns with these metrics? If a post has 12 Likes or 27 Likes is that really that different? They're both clearly "good." Same with 2k Followers vs. 4k.

I think the short answer to this Q is yes. In growing any platform you need to give people reasons to want to participate. For some folks who will most actively spread the platform, a number is a good motivator, and you need these folks. The other great motivator is in-person contact (I saw your post on FB) that helps you know you have an audience. Numbers worked for LinkedIn (# connections), Facebook (# friends), as well as twitter, etc. There are diminishing returns here though -- on LinkedIn even around 250,000 users we capped the "# connections" to 500+ so people didn't continue pushing past that since > 500 was actually less useful for the network since it instigated a lot of weaker ties.

Mafia Wars on Zynga was the first game to really exploit this behavior and actually incentivize you based on how many friends you had in the game. We can debate on another branch whether that was successful.

But I think Satya's point is most important -- you have to motivate the features you want - the ones I have seen work best are generally around userbase and audience growth.

In terms of Fav/RT metrics and content metrics -- I find these are much harder to compare since it depends on audience and often devolve to cute cats. But for personal use it can be really interesting. Worth noting Twitter didn't show total # favs or retweets beyond 15+ until just recently.

Twitter Favorites are important because they're dopamine hits that are *given* from one user to another. How else do I know if people like my tweets? Tweeting for favorites is like telling jokes to your friends for laughs – sure, there's some influence by popular opinion, but it's also just normal social behavior.

Right now, my options for giving feedback on Branch are limited. I can a) write a response, which is a lot of work for others to read or b) tweet an @-mention, which is a disorienting context switch.

I think platforms benefit from a one-click Like/Favorite/+1 feature, but that's a separate question from whether those appreciative gestures should be aggregated as a number, or even displayed to participants/readers of the Branch.

Here's my general rule of thumb (some part of hunterwalk.com)

Any communication service which publicly displays a metric serving as a proxy for popularity will cause users to take steps to increase that number. To think otherwise is kidding yourself, so be clear about: public, private, unavailable. And make sure to apply correctly based on behavior you're trying to create.

For example, I find it strange that Path shows a public "read" count given they focus on strong ties. Read count viewable by my friends makes me want high counts too & thus max my # connections to 150. This data may be relevant to the poster (to show engagement and encourage more posting) but given that emoticons provide public feedback, why a "read" count too?

I don't think it's necessary in all cases. I agree with Hunter that if you are going to implement any sort of metrics, make sure you're implementing the correct ones in the right places. The story of the Digg Leaderboard springs to mind as an example of a metric-tracker that didn't scale with the site's popularity (and was taken down as a result).

The important metrics to me, as a creator, revolve around knowing that people are getting value out of the content I'm producing. This may mean passive stats like number of views or more active stats - likes, comments, retweets, etc. It helps me shape what I share going forward.

It appears we are gravitating towards two slightly different metrics:

A) Direct metric or micro feedback: One that gives you (creator) feedback and joy for the 'content' you create - to help you improve and also for others to see which content of yours or otherwise is good. e.g. Likes, RT, Fav.

B) Indirect metric or cumulative feedback: One that helps others see which user is 'good' instead of having to go through their individual content to figure it out. I would say this is a further abstraction of 'A'. e.g. Karma points, Follower count, Subscribers etc

So between A and B is their a logic as to which one should be made more important or focused upon? Does it depend on the strength of ties in the network and if so then how?

I don't think it's that simple (though I wish it was). For example, if you post something to Facebook you get Likes (direct) which dictate how high it shows up in other peoples' Newsfeeds (cumulative).

Pinterest's Pins is an example of a feedback system which I like. I "pin" something because I want to save it for future use/reference, and that in turn helps other users discover cool stuff.

Agree on the oversimplification in my last post.

Understand what you are saying but I feel a lot of that depends on the specific network and is very hard to generalize.

I feel the Pinterest example falls broadly under the second category where they have micro feedbacks like Repin and Like on the content but also a cumulative score for the user (followers).

When someone repins or likes my pins I get happy and that's about it - maybe it leads to more people following me. But at the end of the day, I still look for the Follower count because that eventually affects the repins and likes I get.

Circling back to the original question:
What is that 'single' number meant for optimization on Facebook, Twitter, Pinterest or Tumblr?

I posted something relating to this on Medium: medium.com