Brad Feld, an investor in AdaptiveBlue, has a written a good post titled Friend Hierarchy.
Brad writes about his desire “to categorize [his] “friends” into a hierarchy that [he] gets to create. One example set might be (lover, buddy, friend, acquaintances, met once, jackass, enemy). This “friend set” should be user customizable and private.”
It reminded me of this discussion I had with Leigh about a hierarchy around friendships built on trust across verticals and subject matters: I trust this person on subject matter X, and this person on subject matter Y even though the network might trust Z.
In life we organize our relationships into a large number of hierarchies, with individuals spanning different sets and lists. Trust, friendship, expertise (both domain and absolute), knowledge across verticals… the list goes on.
And yet online we have “friend.”
I see two interesting things to discuss and debate here:
1) Which hierarchy is most important to surface online first: Is it deepness of friendship? Level of knowledge? Amount of expertise? Or something totally different?
2) Is the best way to surface these relationships through explicit means or via implicit methods?
To dive into the second point a bit: would you like to have your friendships surfaced, automatically, via existing trust relationships and have the level of the friendship defined for you, automatically, by how often you interact with your network?
Or would you like to explicitly add a friend and then define the level and nature of the friendship?
I’d like to hear from a number of people on this topic. Alex/Andy/Karen/Rion/Jeff, what say you? Micah and Stan? Leigh, you sort of started this whole thing. What about Anne, Becky, and Leona? Anyone? Bueller? Bueller. Bueller.
{ 1 trackback }
{ 26 comments… read them below or add one }
Fraser,
My thinking is that as the number of networks increase, there is an explosion of content, both unique and cross-posted, the data is being accumulated that allows for a better understanding of relationships (both in the negative and in the positive).
So, the idea would be that there has to be answers to three questions:
1) content discovery: Where are my friends? Where is the content they are generating? What related or semantic content exists that I dont know about? Basically, tell me what I dont know and what I should know.
2) relationship recommendations: Based on my current networks and content development (both in terms of frequency and amount), tell me whom I should know, and more importantly, who should know me. Take these two people: One) Big in facebook and twitter, light in last.fm and flickr. Two) the opposite. But these two folks are friends, and post about similar topics on their blogs. Therefore, there is likely people they should know, even if their social graphs are different.
3) social search. Now that I have discovered content and relationships, I need a way to manage the amount of content, and have a quick and easy way to recall and be notified about new content from trusted sources.
Without content discovery, you cant recommend relationships. Without recommended relationships, there is no content. And without each of those the need for social search declines. Trust become systemic, and friend hierarchies would emerge organically.
At the end of the day, I would have a single application that had a list of all of my relationships, broken into a hierarchy/segmentation, with the ability to discover content and search through all the relationships. Then my network become a living organism that grows based on the members of the network.
And then we take over the world! (Ooops, too much Pinky and the Brain.)
Fraser, we need to sync up soon, and thanks for the link.
I’d like to think of these things not as hierarchies but as networks. Each one of us participates in multiple networks - friends, school, family, work to say the least.
In each network we are connected to people via relationships of various strength. As we interact or not, the strength increases or decreases.
Facebook categories are way too much and do not map onto strength anyways. The point is that strength evolves with time.
To me, there needs to be a tool which figures out the strength automatically. Xobni I think is on its way to do it via email.
I’m guessing I’m somewhat older than most of your other readers, Fraser, so I’m at a huge disadvantage here. I read kids books and haven’t had an adult, abstract conversation in more than five years and some of this is hard for me to follow. I also have never used MySpace or FaceBook, but have derived much pleasure from LinkedIn, which has reconnected me to college friends and former colleagues from 20+ years ago.
The social network sites truly don’t appeal to me, only partly because of the difference in generations. I don’t have time to learn to use them, and I’m kinda over worrying about what the hive thinks is hot. To put a fascinating discussion into concrete terms, if you’re in my email address book, you’re there. Forever. If you find me through LinkedIn, and I’m happy to be reunited, you get into my email address book. Repeat the “Forever” part.
Having online “friends” fade in MySpace or elsewhere might make such sites more manageable as people enter the workforce, marry and raise families and get on with life. But it’s not terribly helpful if you someday need expertise on a very narrow subject and vaguely remember a guy named Dan who might know a thing or two, but your sunset clause kicked in and Dan has faded into the electronic ether. (Hence my preference for LinkedIn, I suppose).
As for organizing and rating people by their expertise; again, in theory, this sounds like a far more efficient use of social networks, especially if it can be customized. But it doesn’t take into account that life is a long, circuitous journey. Again, to ground this in concrete terms; what you know today may be obsolete tomorrow (Ask my husband about DOS sometime). You will likely change jobs; the odds are you’ll change careers too. You’ll abandon some interests or avocations and cultivate others; ones you cannot predict today or next year or next decade.
If you’re being ranked–and ranking others–that’s only a snapshot of what you know and where you are in life at this one moment.
Yes, I’m no help at all, but thanks for letting me philosophize. :-)
@Micah, we most definitely need to sync up soon. I was chatting with Eric O recently and he said you were a great guy. That was good enough for me… but then I came across your post (and now this comment) and it’s a must-do in my books :)
In NYC anytime soon?
Let me see if I can summarize - at the 30,000 ft level - what you’re saying: you’d like a network to do analysis and present friendships (or connections) that need to be made?
@Alex, they’re all networks but as you say yourself each network has relationships of various strengths.
Facebook seems to automate some of the evolution of strength - the strength of the relationship is impacted by how often you interact with them. They have categories that cross networks, without introducing hierarchy; both limitations in my opinion.
Does Xobni surface strength across various networks, or a single network?
Anne, your comments and view points are always welcome and appreciated. The benefit is that you come with a different - but equally important - perspective.
You’re part of a community online without being explicitly part of a community. You know what I’m saying? You have a thriving group around the Cybils, but you haven’t had to join MySpace or Facebook to create that. It exists therefore it is :)
The question is, how does your group strengthen the community that is there, while not being there?
Ah, I see what you’re driving at. Cybils as a sort of hub outside the sphere of MySpace or FaceBook. And how do we maintain and strengthen that community? Who rises and falls in terms of authority and/or popularity? How do members communicate?
Let’s see if I can grapple with that, being unfamiliar with those two networks. We heavily, heavily use listservs, both to find volunteers and to do our deliberations. Many judges also use IM and Skype for their deliberations. We publicize ourselves via LiveJournal and MySpace and of course other blogs.
Mostly, we assign people to become organizers or judges based on our reading of their blogs, and assessing their qualifications–a highly subjective approach, one that prizes authority over popularity, mostly, but not always so. Ours is a small niche, and even with several hundred kidlit and YA blogs out there, we eventually figure out who’s out there and what they’re saying. But it’s hugely time consuming, and my google reader is filled with hundreds of posts I don’t have time to read.
I’m not sure this answers your question, or simply circles around it looking for a place to land. But basically, we wing it. :-)
Simple……..cross referencing of rss subscriptions.
Just get everyone to submit an opml file or feeds individually, have that user add structure to those feeds (top level category subject matter as well as simple relationship definition…freind,family etc).
Then, have the users prioritize the feeds/individuals within each category and you can then cross reference to what others have done and present new relationships from that.
That would actually be a “social network with value”. Chicken and the egg thing of course, but it has to start somewhere. Someone correct me if this exists right now.
Example…I have about 10 categories of feeds in google reader and have dragged the ones to the top in order of must reads to occasional reads. The problem with that is, I have no clue if what I have in each category is “Comprehensive”….thats what I want, to discover more experts in a subject matter. As everyone goes niche, a simple staring system of individual feed items wont cut it.
@Anne, these type of questions often circle back :) The nice thing about blogs is that they surface people with similar interests and create hubs of like-minded individuals.
Which brings us to Marc…
The OPML file could be considered a representation of a network of like-minded people. Originally I thought the idea felt complex, wondering how many networks would have OPML files and then I reread Anne’s post and saw the magic two words: “Google Reader”.
Interesting thought.
Sorry I’m late… traffic….
The unspoken thing here is that only a percentage or two of the internet’s users can even understand these terms. What’s needed is an interface which builds from our implicit networks and information but allows us each explicit, “final word”, control over the specific ranking of individuals and groups within those networks.
The key is in combining implicit with explicit information and permissions. Google is perfectly placed to do this. The concept could be extended into targeted marketing, allowing users to have some explicit control over the ads presented. Seems to me there would be alot of revenue there…
jon
@Anne & Fraser
Fraser, when I was 25 I seeked optimal state and understood nothing but efficiency. Fast forward 10 years and I see what Anne is saying - lets just connect in inefficient human ways that we do connect.
Broader, I do not think there is single audience or solution here, people want variety based on their age, habits, experiences and background. And as long as there demand there will be supply - LinkedIn, MySpace, Facebook, etc…
@fraser so there are a couple of things here:
1) is terms of a friend hierarchy its difficult to determine what “level” a friend can be placed automatically. Lijit does it by link levels. So a direct link means I trust you, you link back, and the trust level has increased, and we are probably better friends. If I had a friend on three networks, are they a better friend that if they only exist on two?
This is why I think a separate app makes sense. It pulls in all your friends on all the networks you are involved in, and then (sort of like Fuser) applies some logic to place them in a hierarchy, which I can manually adjust. This heirarchy might shift over time (sort of what Xobni does with email).
2) What I want is an app that does the above PLUS: 1) discovers content; 2) social search (Lijit kinda does #1 and #2); and 3) relationship recommendations.
For #3, this app would tell me:
1) What networks are my friends on that I dont know about (kinda content discovery, but really friend finding).
2) Who I should be friends with, but am currently not.
3) Who should be friends with me, but is currently not.
Then, I should be able to do basic updates to all social networks I am a part of from the application.
This doesnt exist yet. Imagine taking that application, and interweaving it with Adaptive Blue. Now, there is a ton of content that would allow the ability to answer the question “Where should I go on vacation for $3k.” Right?
I get to NYC every couple of months, but you and Eric should come out this way…see some grass, maybe breathe real air?
I realize that I wasn’t really invited to the party, but I thought maybe I could offer up my thoughts on the subject of organizing friends.
Connections with individuals are incredibly complex, that I’m sure we’re all aware of. Our connections are naturally categorized in an amazing variety of ways when you think about it, and simply you’ll never be able to design ‘the perfect tool’ for organizing them. The better option is to give a flexible tool that offers an easy-to-use experience, IMO.
Look at the ways we connect to people, I mean.. they can be connected to us through a friend, through a hobby, through an organization, through an event, etc. Even more than that we prioritize these connections. We talk to A more than B, we trust C more than A on certain topics, but B is the go-to for information on another topic.
Every time we meet a new person, we have forged a connection with them. It can be one-sided, or not, but that connection exists. From that moment forward we build on our connection with that person, or choose not to build on it and it slowly fades away.
An appropriate tool would remember those connections, and the people we conversed with, and let us start filling in the blanks for it. Once the connection is forged, the machine won’t be able to learn nearly as much in a useful way as we can teach it. The tool would allow us to create ’smart groups’ of connections based on rules that we apply.
For instance, let’s say that I have a conversation via Skype with Alex (which I have done several times) this tool would remember that connection and some basic information about Alex (name, date of first contact, means of first contact, a record of what transpired if applicable) and from that I could start sorting data as it relates to Alex. I would perhaps would to teach the tool that Alex would be a go-to for information about the semantic web, or web 2.0.
This could be accomplished using keywords, tags, ratings, or a variety of other methods. The system shouldn’t be trying to be ‘perfect’ it should be trying to be flexible.
That’s my (worthless) opinion, it makes me feel good to think I’m contributing.
Fraser,
The problem with explicitly indicating the types of relationship people have with you is that it is similar to tagging, few people do it to the extent that it becomes terribly useful. And the relationships change. So I think in order for something like this to truly work it needs to be automatic.
As Alex said I don’t see it as a single solution. I see it as a combination of a number of different factors that can be heavily influenced by the field of social network analysis (http://en.wikipedia.org/wiki/Social_network). For me having the following data could go a long way to understanding my relationships in SocNets:
-the degree others in my community act on information I put out
-Cohesion (degree to which your friends are connected to others)
-flow (who passes on information about me)
-nodes (who are the other people with significant networks in my network)
-common interests (maybe just the number of keywords we have in common)
These automatic methods could be augmented by manual indication of trust and level of knowledge.
As for having it defined for me, I would want it visually displayed in a manner that was helpful to me in organizing what I want to do with those friends. Maybe it would be nice to have suggestions that “you may wish to send this message to …” But I get afraid of that because I don’t want an automatic system dictating how I relate to my friends and colleagues.
@jon
Correct, not many understand these terms but they can be flattened into something else. Look how far the dig icon has come. It’s on some big media sites now. A system can eaisly allow for an icon that says anything really to represent the adding of a sites main rss to ones friends in the context of what I spoke about without that person ever knowing what rss is.
The network that Fraser has in mind I think really needs to not try and ride the 2.0 bandwagon as far branding and presentation is concerned. Keep it plain, simple, fast, and above all the fewest clicks as possible (google has perected this). These rules will never change as long as the web is served up via a browser and a mouse.
More specifically, I should be able to click on an icon on ReadWriteWeb.com for example and have it take me to a page on OnlyRelevantStuffMatters.com and have it immediatly, without social network entry barries (account creation, friend invities) show me a list of URL’s that others have grouped in a similar bucket. Then let me register once I’m hooked. This process will weed out the scammers as the number of users come on board.
Thats it for 1.0 release, just that, no friends, no buddies or pals, no special rankings/ratings comments or opionions. Just me adding the stuff I know I like, and show me some more stuff thats relevant that others like. Like I said, chicken/egg thing of course, and geeks like me wouldnt mind being early adopters. Especially when my interests are everything from SEO, Affliate Marketing, Web Dev, Investing, Cosmology, Flamenco…..blah blah blah.
Wow, I spent a lot of time writing this, I need a blog. And, after re-reading this 3 times……….its essentially this…..
(future press release for Jan 2009)
“Human Rank by [CompanyNameHere] trumps Googles Page Rank for web search.”
Wow do I feel like a tard. Looks like del.icio.us is essentially what Im asking for and really a first step in what everyone is talking about and something I need to use much more. However, their SERP and homepage needs some serious work and every taxonomist knows that user submitted tags ends up in in mostly garbage, they should clean that up.
LOL, sorry for my long post about something that already exists.
As with Alex/Jonathan I do not see a single solution here. Adam says it well when he states that “connections with individuals are incredibly complex.”
If I was a betting man, and I’m not, I’d say that in the long run social networks do not exist as we know them because, returning to Anne’s statement, “lets just connect in inefficient human ways that we do connect.”
The lasting contribution of today’s social networks will be the social network constructs that everyone now understands. The idea of “friending” someone, the life stream, low friction communication forms such as the poke, the super poke and the super duper poke.
@Marc, you do need a blog - your comments are always content rich.
@Micah, I was there a few times in the last quarter of 07. Is such a beautiful place.
@Jon and Jonathan,
You two share my point of view on one particular aspect -
Jon: “The key is in combining implicit with explicit information and permissions”
Jonathan: “So I think in order for something like this to truly work it needs to be automatic”
But automatic presents challenges because of the various ways that we connect with people (work, hobbies, interests, organizations).
Which leads my brain back to social + meaning. No?
A couple things:
To the comments:
- people don’t know these terms
- not many people tag
The medium isn’t all that old and how quickly things are getting adopted is accelerating at an enormous rate. People who are 24 today don’t see technology (i.e. I still go on my computer or go online - they don’t - they are the network) and that will have a impact. People won’t need to know what a tag is, they will just understand what the task of tagging enables them to do.
So a second comment:
I love the comment about trust needing to become semantic. And I would also agree with the notion of a model that is both explicit and implicit. I do want a “social tagging system” where I can associate my network with specific things. Maybe it’s simply ‘friend’ or maybe its ‘friend’ ‘technology’ ‘writer’.
Of course this is all getting my brain working this morning (which is much appreciated Fraser as I got delayed in Chicago and didn’t get to sleep until 2:30am) so I think i’ll probably write a post of my own.
:) Leigh
If I were to give an example of how data might be formated, and even automated to some extent, it would be “address book 2.0″ (if you will).
To elaborate:
A vCard can already contain a LOT of very specific information about you, so let’s update that a bit to include a basic tagging system, trust rating (perhaps even set by topic) and more information on the actual connection than the individual. More importantly, the card would probably contain lifestream information (this person writes XYZblog, and has a facebook page, and shares items on GoogleReader)
The client side solution would be more intelligent, it would interpret lifestream information and add it to this person’s data in a usable manner. Over time you’d build a usable customized database of all of your contacts.
To make a quick and dirty comparison: Think iTunes for People.
I actually kind of (in a way) touched on this topic in a post on blogger that I wrote while completely sleep deprived. If you’ve got a few minutes to read my ramblings they’re here:
http://web2-oh.blogspot.com/2008/01/my-life-suite.html
Adam, great post. I still am not convinced that “iTunes for People” is it. I want iTunes to listen to music. I want Netflix for movie rentals. I want Amazon for purchasing books.
I’m not convinced that I want a separate thing for networking. To me that’s analogous to a real-world club that met to network around networking. Maybe such a thing exists but more often than not people get together around sports, sports teams, music, boy scouts, etc.
How do you see your idea handling that?
Leigh has posted her thoughts about a social network tagging system, which sort of overlaps with Adam’s thought.
Maybe it’s an extension for how Adam’s thought could be implemented to provide value to verticals? i.e. Leigh, tagging Dave “tech” in Adam’s “iTunes for People” would slot him into some filter for tech searches?
Dave? Leigh? Jon? Beuler?
http://leighhimel.blogspot.com/2008/01/wanted-social-network-tagging-system-to.html
@Fraser
Spot on with taging value to verticals, but also the ability to give weight to things within those verticals, be it people or books.
Oh, I think you misunderstood me. What I meant is that your address book would kind of act like iTunes for people.
In that regard I meant that you could sort, add data, download data, create ’smart groups’ and so forth based on certain criteria and rules. The address book shouldn’t be a social-networking application on it’s own though, it should be integrated with several things including IM and Email clients.
I’m sorry if I wasn’t clear, I typed that post very late.
I think Adam’s post is great - but I”m thinking less social networking than social search. How do we utilize our network to make us smarter.
I do think taging value to verticals could work. With this concept, the success aboslutely will be in the execution. If there isn’t the right balance between implicit or explicit value then it will get too complicated too fast.
But as with most things, the value will be in the network - the individuals feeding their ascribed value into the borg and having that borg get smarter because of it.
(oh and I am also sleep deprived so if this didn’t make sense, feel free to earnestly dismiss :)