I’ve noticed a few interesting responses to my post on simplifying XFN. While my intended audiences were primarily fellow microformat enthusiasts and “lower case semantic web” types, there seems to be a larger conversation underway that I’d missed — one that both Adam “Everyware” Greenfield and Tim Berners-Lee have commented on.
In a treatise against XFN (and similarly reductive expressions of human relationships) from December of last year, Greenfield said a number of profound things:
…one of my primary concerns has always been that we not accede to the heedless restructuring of everyday human relations on inappropriate and clumsy models derived from technical systems – and yet, that’s a precise definition of social networking as currently instantiated.
All social-networking systems constrain, by design and intention, any expression of the full band of human relationship types to a very few crude options — and those static!
…it’s impossible to use XFN to model anything that even remotely resembles an organic human community. I passionately believe that this reductive stance is not merely wrong, but profoundly wrong, in that it deliberately aims to bleed away all the nuance, complication and complexity that makes any real relationship what it is.
I believe that technically-mediated social networking at any level beyond very simple, local applications is fundamentally, and probably persistently, a bad idea. From where I stand, the only sane response is to keep our conceptions of friendship and affinity from being polluted by technical metaphors and constraints to begin with.
Whew! Strong stuff, but useful, challenging and insightful.
Meanwhile, TBL defended a semi-autistic perspective in describing the future of the Semantic Web (yes, the uppercase version):
At the moment, people are very excited about all these connections being made between people — for obvious reasons, because people are important — but I think after a while people will realise that there are many other things you can connect to via the web.
While my sympathies actually lie with Greenfield (especially after a weekend getting my mom setup on Facebook so she could send me photos without clogging my inbox with 80MB emails… a deficiency in the design of the technology, not my mother mind you!), I also see the promise of a more self-aware, self-descriptive web. But, realistically, that web is a long way off, and more likely, that web is still going to need human intervention to make it work — at least for humans to benefit from it (oh sure, just get rid of the humans and the network will be just perfect — like planes without passengers, right?).
But in the meantime, there is a social web that needs to be improved, and that can be improved, in fairly simple and straight-forward ways, that will make it easier for regular folks who don’t (and shouldn’t have to) care about “data portability” and “password anti-patterns” and “portable contact lists” to benefit from the fact that the family and friends they care about are increasingly accessible online, and actually want to hear from them!
Even though Justin Smith takes another reductive look at the features Facebook is implementing, claiming that it wants to “own communications with your friends“, the reality is, people actually want to communicate with each other online! Therefore it follows that, if you’re a place where people connect and re-connect with one another, it’s not all that surprising that a site like Facebook would invest in and make improvements to facilitate interaction and communication between their members!
But let’s back up a minute.
If we take for granted that people do want to connect and to communicate on social networks (they seem to do it a lot, so much to that one might could even argue that people enjoy doing it!), what role should so-called “portable contact lists” play in this situation? I buy Greenfield’s assertion that attempts by technologists to reduce human relationships to a predefined schema (based on prior behavior or not) is a failing proposition, but that seems to ignore the opportunity presented by the fact that people are having to maintain many several lists of their friends in many different places, for no other reason than an omission from the design of the social internetwork.
Put another way, it’s not good enough to simply dismiss the trend of social networking because our primitive technological expressions don’t reflect the complexity of real human relationships, or because humans are just one of kind of “object” to be “semantified” in TBL’s “Giant Global Graph“… instead, people are connecting today, and they’re wanting to connect to people outside of their chosen “home” network and frankly the experience sucks and it’s confusing. It’s not good enough to get all prissy about it; the reality is that there are solutions out there today, and there are people working on these things, and we need smart people like Greenfield and Berners-Lee to see that solutions that enable the humanist web (however semantic it needs to be) are being prioritized and built… and that we [need] not accede to the heedless restructuring of everyday human relations on inappropriate and clumsy models derived from technical systems.
I can say that, from what I’ve observed so far, these are things that computers can do for us, to make the social computing experience more humane, should we establish simple and straightforward means to express a basic list of contacts between contexts:
- help us find and connect to people that we’ve already indicated that we know
- introduce us to people who we might know, or based on social proximity, should know (with no obligation to make friends, of course!)
- help us from accidently bumping into people we’d rather not interact with (see block-list portability)
- helping us to segment our friendships in ways that make sense to us (rather than the semi-arbitrary ways that social networks define)
- helping us to confidently share things with just the people with whom we intend to share
There may be others here, but off the top of my head, I think satisfying these basic tasks is a good start for any social network that thinks allowing you to connect and interact with people who you might know, but who may not have already signed up for the service, is useful.
I should make one last point: when thinking about importing contacts from one context to another, I do not think that it should be an unthinking act. That is, just because it’s merely data being copied between servers, the reality is that those bits represent things much more sacred and complicated than any computer might ever be programmed to imagine. Therefore, just because we can facilitate and lower the friction of “bringing your friends with you” from one place to another doesn’t mean that it should be an automatic process and that all your friends in one place should be made to be your friends in the new place.
And regardless of how often good ol’ Mark Zuckerberg claims that the end game is to make communications more efficient, when it comes to relationships, every connection transposed from one context to another should have to be reconsidered (hmm, a great argument for tagging of contacts…)! We can and should not make assumptions about the nature of people’s relationships, no matter what kind of semantics they’ve used to describe them in a given context. Human relationships are simply far too complicated to be left up to assumptions and inferences made by technologists whose affinity oftentimes lies closer to the data than to the makers of the data.
Factory Joe 
At it’s core the Semantic relationship is about what a computer algorithm could infer from data present to it lacking all other knowledge. And there’s the mistake.
My contact list is very simple. It’s actionable. I list people because I want to interact with them. And I put them in buckets that fit those interactions.
Invited to my next party? You go in the “socialize with” bucket (aka friends, but not really). Share the TPS report with you? You must be in the “NDA sharing” bucket (aka co-worker, but not always). Plans for the holiday? How about the “you don’t get to choose” bucket (aka family).
Want to know how I relate to people in different buckets? Watch my actions. If you watch my actions, you won’t have to infer anything from the bucket name (or risk guessing wrong), you’ll have all the information you need.
If you want to assume that you know me based on some blueprint, but don’t want to watch and pay attention to what you do, then you probably care more about the bucket name. Like you said, that would be the semi-autistic worldview.
It’s not that the relationships are complex, fluids and hard to pin down, but the fact that there’s not relationship information conveyed in the bucket name. It’s an illusion created by someone who gave me a checkbox, and the closest thing to “socialize with” that wasn’t family or co-workers was friends. So I conformed but in doing so lied. What I really meant was “people I’ll mass invite to the next party”.
So think of it in terms of me person, and buckets that make it easier to group actions/people together.
By analogy, the Semantic way is another attempt to create a super-smart keyword-assisted search engine to organize all the world’s knowledge.
But the me person way of looking at things, is the equivalent of PageRank. And we know which one worked better. Just let the data be and observe what happens.
Hey, I’m the guy who asked that “discretization of human relationships” question at SXSW. Great panel, BTW. ( =
I think Assaf is correct (mostly). Don’t ask me what my relationship is because humans are bad at exactly specifying things; in fact, I’d say that perfect accuracy is an enemy to a healthy human relationship, especially once you try and plumb the fractal depths of human experience.
I think that relationships are fluid and complex and hard to pin down, though. If I haven’t contacted someone for more than six months on your social network, then the network might think we’re not friends… but it wasn’t watching me in RL talking to them every day. Or if it *was* watching me talk to someone in RL every day (via a GPS-enabled phone, say) then it might assume we’re best pals when in reality we’re forced to work together despite our differences.
I think trying to ferret out the deeper meaning behind the connection isn’t that valuable. What can your network do for me when it knows my close friends vs. acquaintances vs. unfriendlies? Actually, heck, that’s a good question: What will the network do for me once it knows the qualities of my relationships?
Excellent point: because we don’t yet have the ability to mirror the complexity of human relationships online does not imply that it could never happen. Instead of complaining about reductivism, the real question should be: what are the abstract qualities of our human relationships that any online system *should* mirror. I never trust anti-reductivists who rely on some mystical qualitative character of an experience.
One might ask also what kinds of *new* interactions social networks enable. Twitter/Status updates are a good example: it doesn’t make sense for me to snail mail or even e-mail all of my friends when I do something, but Twittering gives me a channel to spread news to those who care.
Chris, another great article. I think it’s amazing to see that a great tech dude like yourself keeps challenging technology for the sake of its users!
The problem is pretty complicated, I agree with you(and Sir Berners-lee) on that. What strikes me about this is that it isn’t very complicated for one factor in the equation. Humans are quite capable of handling complex relationships. We might “know” or be linked to thousands of people on-line but we usually instantly know the value of our relation to those people.
Another way to think about this problem is not to think about the relationships themselves, but maybe look at the ways I want to communicate or interact with them. There are a lot of different ways to communicate now, and often thee ways say something about the relationship too. For example. I SMS a lot (Europe right), but the people I SMS, I know quite well (family, friends, co-workers). When I Twitter, I simply join in on an on-line conversation with many people. Some of these I know, others I barely ever interact with. When I respond to your blog, I do this because I believe you write excellent stuff which triggers my thoughts again. In each of these cases the relationship I have with others is different. I’m not your best friend, and even though we have never met (yet) I take the time to follow the things you write.
I never, ever, fill out those dumb question forms from Facebook, or any other data hogger trying to get his hands on “my social graph”. Its for two reasons, first I don’t see the need for them to know that stuff, but secondly I do not see any added value in registering that info. It doesn’t remotely reflect the relationship I have with these people.
I think that the basic stuff you mentioned are probably worthwhile implementing. Personally I am becoming quite fond of the way Friendfeed lets me meet new people. I don’t use the serve very often yet (for other reasons), but the neat thing there is that if I follow Chris Messina, I also get to see things from the people you care to follow. A natural way of meeting new people.
Finally, this is almost as long as an average blog post of mine ;-), I wonder if we can turn this problem upside down. Why not let me have all my friends with me on a place that belongs to me, and me only. I could then visit services, and if they are willing to service me, bring along some of my friends.Technically, this would end the destination problem every service currently has (who has the biggest). I could visit Facebook just once, and send a poke to someone I know fro the sake of it, even if both him and I aren’t registered users on Facebook. Wishful thinking I guess.
I think there’s a simple approach that would work very well for my 300+ FB contacts, 600+ LI contacts, and 3500+ MS contacts: a “relationship strength” slider. Leave the interpretation of strength relatively ambiguous so that I can decide what it really means *for me*, but a slider that goes from ‘weak connection’ to ‘strong connection’ would be great. Then, elsewhere in the social app, let me filter my contacts / connections based on strength.
This works both with social and professional connections and is quick, easy and quite manageable.
I hate the Facebook “categorize your relationship with this person” grid, but hate even more that I can’t have any qualitative indicator on my connections on LinkedIn, MySpace, FriendFeed, etc etc etc.
Trying to perfectly model all human relationships in a grid or series of values, however, is crazy. We’ll never have any improvement because we’ll all just disagree and never get beyond stage 1…
Good stuff.
I’m just glad you folks are beginning to talk about these issues and have quoted Adam Greefield. I remember when Tantek decorated the offices of Technorati with an XFN poster like wasn’t it just the coolest thing ever and I thought, “Oh no he didn’t!” But he did. I figured maybe there was something I just didn’t understand, or wasn’t willing to… because I certainly didn’t feel comfortable reducing my fellow human relationships to a list of checkboxes. Especially in public.
+1 on the manual import (not unthinking) for humans. I would personally prefer the magic auto-import, but definitely see why one might not want to.
very interesting read, and some insightful comments. two comments in particular – Assaf & Dave Taylor focus on a method/tool which would work for them.
Focusing on the main article, especially related to the bulleted list about making things humane – the reason soc nets make arbitrary buckets is because we are using them for many reasons and not for a clearly defined one. Using a network for talking to friends and maintaining a relationship with all your clients on the same network as the same person is where this arbitrariness and confusion and misrepresentation sets in. Facebook’s new permissioning is a small work-around this but still not powerful enough and it is hard to get right. And these are just work arounds and not really a complete and intelligent solution. Maybe thats 3.0 and we are trying to get to atleast 2.5 now.