Photo by anne.oeldorf on Flickr used under Creative Commons
The debate about whether FeedBurner’s inclusion of FriendFeed subscribers is a good or bad thing has me thinking how to determine a person’s reputation and influence.
As I keep saying, trust is the key component to New Web. Without trust, it’s difficult to build a community around anything.
Reputation and influence are the next big things in New Web. We’ve been noodling how to establish reputation for a while now and have some ideas for internal use on Connect.
What’s the big deal?
Beyond people you actually know, i.e. met in meat space, worked with on a project, how can you tell if you want to pay attention to someone? Sure, there’s a profile that might help you make the call, but how do you decide to read a person’s blog, follow her/him on Twitter/FriendFeed, friend her/him on Facebook, etc?
There are rudimentary methods that I’m sure everyone uses, since consumer apps don’t provide any reputation scoring (and centralized reputation is a pipe dream for now).
I’m sure you know them already.
Blogs
- FeedBurner subscribers count
- Google Reader Subscribers Greasemonkey script
- Technorati authority
- Post frequency
- Comments per post
These are highly subjective and don’t really help you vet the content, but I guarantee that a blog with more subscribers is frequently viewed as more influential, which is why the recent FeedBurner change to add FriendFeed subscribers borks up the system a bit. Of course, the system was already borked.
I usually add new blogs based on the recommendations of friends (ahem, trust), or I’ll test drive them for a week or so to see how the content and commentary is.
A lot of bloggers now show you how many Twitter followers and FriendFeed subscribers they have, which only makes things more messy. Does having a lot of followers on Twitter make you more influential?
Cue the segue.
Twitter, FriendFeed
- Number of followers
- Ratio of followers to following
- Overall activity
- @ reply or commenting frequency
- Cursory review of the person’s blog, applying the blog methods to determine reputation and influence
I lump these two services together because they both employ the asynchronous model for networking.
As with blogs, I also employ the recommendation of friends model both direct and implied, i.e. by looking for overlap with my network.
As Twitter has grown by leaps and bounds, it’s become more difficult to know who’s worth following and who’s not.
- Number of friends
- Number of shared friends
- Overall activity and commenting
- Profile information
- Cursory review of the person’s blog
I’d argue that because of the trust built into the synchronous model employed by traditional social networks, they produce the best and most consistent reputation. The addition of friends in common was a smart move to help people vet potential friends that can easily become part of a reputation system.
It’s only a matter of time before Facebook adds some kind of reputation scoring to their network and adds it to their Facebook Connect payload. This will be the first shot in the war for reputation/influence, which will be part of the war for identity.
So what?
Reputation and influence will be very hard to establish, due to their inherently personal nature. If tomorrow Twitter or Facebook announced a reputation score, people would go nuts comparing themselves, bragging, complaining and dissecting and questioning the algorithm.
Speaking of algorithms, Google is an interesting player in reputation because they have so much data about you and what you do online.
For a long time, forums have calculated scores for members based on contributions, activity, etc. This model works, but it hasn’t been applied outside forums or across networks. This is a good starting point, but it needs major tweaking.
There are solid patterns that can be followed to create reputation and influence scores, and the score should be a combination of algorithm plus user’s scoring of each other, similar to the forum model. The algorithm needs to be smart enough to know when people are gaming though, which is complicated.
Overall, determining a score on the consumer web is a very tough row to hoe.
Not so much inside an enterprise where core values create a basis, e.g. use the company handbook as a starting place for corporate values on which to base reputation. With a baseline, it becomes easier to model reputation and influence based on social activity.
What do you think? Is this a pipe dream for the consumer web? Can reputation be scored inside a company? Did I miss something big?
I don’t know everything, natch, so find the comments and enlighten me.
Why does there have to be one rating? Why not let the users manipulate their own data and create their own formulas (and, of course, share them) to see which metrics are most representative and productive?
I'm pushing that route for Galapag.us: http://www.facebook.com/pages/Galapagus/1003399…
“…met in meat space”
I like it.
I'm all for a feedback-based component, but leaving it all to users creates inconsistency, doesn't it? Plus, generating reputation for people makes them care about it, whereas if it's all on the users, it's harder to get started.
Look at social networks for an example. Did you one day decide to join Friendster, MySpace, Facebook, LinkedIn, Twitter or were you invited? The vast majority of people were invited to join.
So, I prefer an algorithm that factors in user-to-user kudos style points, but also uses other criteria with equal or higher weighting.
Wish I could take credit for that. Alas, someone said it to me last week.
I consistently find it curious that people feel more comfortably judging people they meet in the physical than they do getting a feel for people online whom they most likely know far more about (unless they're in the habit of having background checks done on people).
Con artists in person are far more difficult to detect offline than on because there is a record and inconsistencies are made more obvious.
Interesting point. I suspect con artists online will evolve over time as the low-hanging fruit disappears, e.g. 419 scammers, and social networks with trust built in will be a natural place to start.
Hi! Very interesting! I found your blog post from galapag.us 😉
I'm totally for local trust metrics. See http://gnuband.org/papers
Paolo http://gnuband.org
Thanks Paolo. Ben from Galapag also commented. I definitely think reputation and trust will be areas where enterprises can fully leverage all the social layers being added. Maybe this is the true differentiator for E 2.0.
On the topic of measuring influence on Fbook, anyone see anything / tools / apps trying to do this? There's a ton for Twitter (makes sense with the liberal API and all), but have you seen anything re: measuring influence, reach and reputation on Facebook?
I think the project Ben Turner, who commented above, is working on is for FB, but I don't know if it's specifically meant for reputation.
I haven't seen much reputation applied to FB, and due to its origins, not sure reputation makes much sense. FB has never been a casual meet-new-people place, focusing on existing relationships instead. Therefore, reputation means virtually nil on FB.
There are some apps on FB that measure networks by depth, strong/weak ties, etc. but they are focused on individuals. So, unless you display your network analysis, no one can see it.
I think an outside source has to use FB data as part of a reputation engine. FB wouldn't see this as a core of their product, unless they move to a discovery model more like Twitters.
Agree re: reputation from inside of Facebook. But having stats i.e. when I share a piece of content = this much engagement is powerful. This is more influence than rep. based.
Thoughts?
I guess I'm not sure if/how FB tracks that type of metric. I suppose they could surface pageviews or clicks/impressions for items you share, but based on the gen pop of FB, would they invest in a feature that isn't very germane to their base users?
Somehow I doubt the average FB user cares about engagement as a metric. Probably just wants that old college buddy to see the funny drunk girl picture. I also think engagement doesn't apply well to FB b/c the network is closed by design, whereas Twitter and blogs have open networks. Tough to measure network vs. network.
I guess I'm not sure if/how FB tracks that type of metric. I suppose they could surface pageviews or clicks/impressions for items you share, but based on the gen pop of FB, would they invest in a feature that isn't very germane to their base users?
Somehow I doubt the average FB user cares about engagement as a metric. Probably just wants that old college buddy to see the funny drunk girl picture. I also think engagement doesn't apply well to FB b/c the network is closed by design, whereas Twitter and blogs have open networks. Tough to measure network vs. network.