really like the idea of being able to rank friends so that you don't miss out on stuff from less 'popular' users - Frederic

Interesting thought Thomas. So maybe the ranking you apply to a user could be a multiplier against the FriendFeed algorithm that determines the "best of" sequence. Higher rankings will drive these users higher on your "best of" pages. - Hutch Carpenter

that's it exactly Hutch. It would be pretty simple to do (I think) and I'd be surprised if the FF folks haven't already thought of this. Ranking users on a scale from 1 to 100 would be your own subjective filter into their "best of" system that would result in a far more engaging and engrossing "best of" discovery page. - Thomas Hawk

similar attention profiles? http://friendfeed.com/e... - Erhan Erdogan

When you apply it to search it becomes the most powerful of all. It's going to take FF a few years to really index the web through their social filter in a meaningful way, but in the end FF could be far more powerful as a search engine than a social network or social media aggregator. - Thomas Hawk

Erhan, APML focuses on topics, not people. What Thomas is suggesting is that a person, regardless of their content, could be ranked higher than another person. It's up to you, as a subscriber, to determine how interesting that person is. - Mark Trapp

I read a comment on a previous post here on FriendFeed mentioning that more than a third of discussions on FriendFeed probrably are _about_ FriendFeed... Most of the items where there are a lot of replies have something to do with FF. I wonder if this service is going to become way less interesting when people finally get bored of talking about the service. What happens when you run out of "nothings" to say? - Jonathan Sterling

Yes, right! IMHO APML is by far more interesting than "ranking" people with notions like karma or such. I'm on the Internet because I want/like/hate/need stuff and it's about these stuff, it's about me first. Then may come other things like friendship and sociability. I'm not against teh "Social" word, of course. I'm for the "me" word that tends to disapear a little bit... :) - directeur

Jonathan, you're following the wrong people (as curt as that sounds). If you expand who you follow to people outside the social media echochamber, you find a lot of posts about a wide range of topics. - Mark Trapp

directeur, we all have relationships with people, not topics. Dehumanizing social media doesn't solve the issue. - Mark Trapp

@Thomas: Interesting. Sounds a bit like a souped up StumbleUpon or Del.icio.us. Different ways at attacking the question of how to bubble up the best of a topic. The time element is the bit that always makes it difficult IMO. - AJ Kohn

Mark: I'm with you, I really do. But what defines "friendship" is our common interests, don't you think? I'm not about dehumanizing the web :) - directeur

Jonathan that's the thing though. Those are the most popular discussions based on likes comments etc. but not the most relevant. Allowing people to rank users would filter more of the most popular out and bring more of the most relevant in. I might not care what 99% of my contacts had for lunch today. But I might care about what my wife did even though it's not popular or more broadly generally interesting. - Thomas Hawk

Directeur, Interests are what might lead to a friendship, but friendship is not completely defined by one's interests. I might meet a friend at a great italian deli, and we hit it off. A few years down the road, he tells me, "Man, I'm tired of italian food." He doesn't suddenly stop being my friend. There's a link between us, after the friendship has been established, that transcends any list of interests. - Mark Trapp

Do you propose this simply for the "best of" or for your actual feed as well? - Frankie Warren

@Mark,@Thomas I don't think like you. I said similar content followers/like'rs/commenters are owners of your best points in Thomas's idea. This will be possible with creating real attention profiles with all your interests in FF. - Erhan Erdogan

Frankie, it would be the algorithm's job to sort out ranking and relevance. But your subjective rank for a person would be a very significant component that went into the algorithm. Think about it this way. Your brother twits, "I just bought a new car." Most people are not interested in this but you probably are. Even more than the 18 popular posts about twitter, google, apple and FF. By letting you rank your brother 100, you ensure that you see that ahead of the other stuff. - Thomas Hawk

Mark, yes that's true, but it's only because you have other "interests" that keep your friendship alive. The day where you don't have no, and I mean NO interest that connects you, that day will be the end of your friendship. Moreover, online friendship and real-life one are a little bit different. We may define those interests, maybe we just can't express them, but they still are here - directeur

Erhan, APML does not solve a very real problem people have: they have a bunch of friends and people they find interesting they want to follow, and are inundated by an inability to say "I like this person more" or "this person is more important to me." If you look at MySpace or Facebook: who are most people following? Their friends. That's what people care about. I think Thomas's suggestion happens to be a good one to help solve that problem on Friendfeed. - Mark Trapp

I have a kind of "manual" friend ranking system. I subscribe to the ones whose posts I most often click through, like or comment. Other users I find interesting but who I don't want in my main feed get their feeds subscribed to with Google Reader. On top of that, I have a few folders on my bookmarks toolbar, each containing a list of links to friendfeeders' feeds pertaining to a field of interest or set of relationships. I call it efficient stream absorbtion - Slippy "Threadsbane" Lane

directeur, I don't know what to tell you. A friendship on that superficial level isn't a real friendship. I'm not interested in dehumanizing people into a list of interests: people are ends in and of themselves, not means for me to extract information. - Mark Trapp

Some good ideas in there. It would be nice to have a hand in "what is relevant to me". - Joel Gray

Mark: Maybe ULML (http://userlabor.org) may do this, but for interests -the things that mostly gather us- nothing equals APML I think. - directeur

APML does not solve the filtering problem: people don't reduce everything into a list of interests. It's a square peg trying to fit in a round hole. - Mark Trapp

Mark: i'm speaking about the programs, the software we use... I think you got me wrong. In real life I don't rank my friends, I even have friends that don't share any interests with me. I'm talking about programs, about apps like FF or other social web apps. these apps need a formal way to know who you are, and who I am, because, for that app, I should be what I like/hate - directeur

@Mark : ) I really don't care about contents are my friends' or not. And don't want to share with my all other followers. For this focus should be content. I think you need a niche FriendFeed like BestFriendFeed.com ;-) - Erhan Erdogan

Mark: me I do, If my best friend *here* talks about Britney Spears, I won't like to comment, click, or "Like" it. I just don't care. Period. But if an unknown person talks about Coltrane, or Dizzy, I'm here to say loudly that I LOVE them :) - directeur

When you are online or offline, you're dealing with human beings. That fact doesn't simply just become irrelevant when you use software. When I hang out with my friends at a bar or whatever, they are the same people there as they are when they message me on Facebook or Twitter or Friendfeed. I want to know what they are up to, moreso than what you're doing or what Robert Scoble's doing or anyone else, even if they are talking about the same thing your'e talking about. What they say has more value to me. - Mark Trapp

Mark: we're a little bit different. That's not to say that I don't have friends or that I don't care about them... No, but I'm a coder (just like you I guess) and whatever we say, we always need formal ways to (re)define things that we borrow to the real life. - directeur

directeur, you could still rank non friend people of interest highly. It would be entirely customizeable. If a stranger consistently blogged about things of interest, you could always rank them 100 too. It would be up to you to apply ratings as your relevance filter. I'd be very interested, for instance, in a pro photographer if they posted lots about photography even though they were not my best friend. - Thomas Hawk

Thomas think this is a great idea although "technically: I can't contribute to the conversation. Blindly hoping algorithms are my friend :D - Mark Forman

Thomas: Exactly, that's a big advantage of attention profiling. I'm a jazz lover, So I'm very likely to be a friend for other jazz lovers, and filters like these will help me find such "interesting" people. Not that others are not worth my consideration, but we may exchange a lot more when we have the same interests - directeur

APML is like discovery for interesting information: like Thomas, he's got an APML profile which weighs pro photography very high. Feeds it into a service and it tells him a bunch of people who he might match (like Toluu, for people). He says "man, John Doe is really cool. I'm going to subscribe to him." That's where APML leaves off. John Doe isn't merely defined by pro photography, but that's what got Thomas interested in him. But now that he's interested in him, he wants to know more about John Doe. - Mark Trapp

This is a brilliant idea -- personally weighted "best of" summaries. - Dewald Pretorius

So APML's solution is to say "forget everything about all my friends other than my interests:" It's not that simple. I hate car talk, so I'd rate the keyword car really low. My friend just got a new car: I want to know about that, but I won't because my interest in cars is low. It doesn't solve my original filtering problem. - Mark Trapp

Yes Mark, we agree on this, and that's what I just said too... BUT we discover future online-friends through the _content_ they share/publish or are interested in. - directeur

If you agree with me, I don't understand why you and your feedego team try to push APML every time filtering comes up. It doesn't help filtering subscriptions _I already have._ - Mark Trapp

Wouldn't it be possible to get around all this and create a personalized list of items that are personalized according to a ranking system. But at the same time keep a river of news view. That way we keep our subscriptions while allowing a second view of the same information through the prisim of our ranks. FF already does some of this through their best of sections and their rooms. So we effectively have two lists - friends that we subscribe to and freinds that we rank acording to our individual interpetation of interestingness. The issue is not how we do this - interest is subjective not objective. Hence my rank of,say, Scoble's stream will be different to yours. The issue is how FF will allow us to do this. - Roberto Bonini

My interest in new cars are very low as well. In fact I could care less if someone got a new car. But if my *brother* bought a new car I'd be interested, even if mildly. But in general we are more interested in some people than others for a myriad of reasons, family, friend, common photography passion, romantic involvement, whatever. The point is by ranking them you could better assist an algorithm in discovering and searching that content. - Thomas Hawk

Fun! I will create a new google docs file :p We must go on there - because we want to draw something and write a little longer :-) - Erhan Erdogan

Mark: Feedego is NOT a social thing. It's really about ego. There's no friendship in feedego. I agree with you on the fact that APML helps you *discover* interesting future friends. I advocate APML in general because I do belive in it, and filtering IMO is a concrete example on what APML may serve. Don't you think? - directeur

Roberto, exactly, "best of" by relevance would not *replace* the current "best of" river by sheer popularity. It would just be another optional view. Same goes with search. But i think it would likely be more interesting and relevant than the current best of. - Thomas Hawk

Directeur: No, it's not! You yourself agreed: APML solves filtering in the discovery phase, it does NOTHING for the content I'm already getting. Take the car example: how does APML not fail, spectacularly, when handling that? - Mark Trapp

P.S. I'm the only feedego guy on Friendfeed btw :) People talking about APML aren't members of feedego, they just like APML, like I do :) - directeur

I'm in directeur :( So bad - hearing such a thng like this from you :-) - Erhan Erdogan

Mark: I've never said " it does NOTHING for the content I'm already getting" ! I told you That ME "directeur" am not like you "Mark" and I won't care about anything about "Britney Spears" be it a content submitted, loved or whatever by my best friend. We're different on this Mark - directeur

At this point, I'd love to have an APML profile just to block every time someone mentions APML. - Mark Trapp

Erhan: I meant you aren't in the feedego staff :) - directeur

Mark: Ah! see? - directeur

@All :-)) Thanks for discussion. But this is end of our comment limits ;-) @Mark all of the things would be together ;-) - Erhan Erdogan

Mark: sorry to annoy you, I really didn't meant that. But a last example for the road: See? Me and you have had many misunderstandings in others discussions... That's it... wer'e human... But the next time I'll see you talking about APML I'll be interested, because you have an objective point of view, maybe different than mine, but I'll be interested in what you'll have to say. - directeur

You could achieve the same by collecting and analyzing a person's clickstream data without asking that persons's personal ratings for his/her topics of interest. My guess is that that is what Google is trying to do with the clickstream data of its users who opt for the option of complete Google "personal history" record. It is a scary thought to let Google know about every click one make on a web page but it very useful data. Like I do not have to look into the browser history. I can simply go back to it and search it for what ever I was looking for at certain web site. This type of aggregation when shared with others becomes even more interesting and some web sites are trying to do just that. In fact an interesting research topic would be to see the differences between the clickstream data based recoommendation system and what a person provides as a personal rating system. - Javed Alam

Big point Javed and yes APML supports this already, that's called "implicit concepts", the others are "explicit" ones. And yes, an application may smartly create a profile from your behavior, but implicit concepts can't be really reliable without something explicitely given by the user. - directeur

@Javed That's all good for conjunction between APML and Thomas's point alg. ;-) But as you said we should see datas after researches ;-) - Erhan Erdogan

I like this notion of relevance. I imagine the "best of" stream could also be re-ranked based on your personal stats of who you find interesting, negating the work of manually ranking your relationships. - Andrew Smith

Mark: take it easy man! I didn't started the thing with APML, Erhan did :) And I'm not shoving APML down everyone's throats. You still say "we", "everyone". You're free to do this of course, but I personally never use "we" online. That's maybe the main difference between us, and that's indeed what separates our points of views, but still, I never see you as a woe... You're a guy who has his own opinions, I have mine and it's okay. (Even when you delete the comment to which this one is a reply) - directeur

@directeur Okayyy. I'm boomer ;-) Heheh! You're funny ; ) I subscribed to Mark - see you in his new comments - we will always punish you ;-) - Erhan Erdogan

Great idea Thomas. It reminds me a bit of how I started to organize my RSS feeds into categories based on levels of interest rather than topics. That way, I could do a better job of keeping up on the blogs that were important to me. - Jeff Smith

Great idea. I study within artificial intelligence and I can definitely say scoring and quantifying this data is the way to a more effective and efficient network. Another question though, do you find it depressing that you're blog has no comments - but FF has over 53? - CannonGod

Great post. Also see an earlier thread I started discussing more or less the same concept - http://friendfeed.com/e... - Aviv