That was my thought too so I have a demo under development. Unfortunately, it won't be ready by Oct.11, and of course, it's all about the right team anyway which is what I'm really interested in. The main reason for posting this idea is to find like-mided entrepreneurs interested in the idea so we can build a team and, as you said, "demonstrate the idea by showing...build an app".
"... The main reason for posting this idea is to find like-mided entrepreneurs interested in the idea ..."
Now I see. Even a mocked up static demo is better than nothing. The Oct 11 date is plenty of time. Now I get what "your going to have to work real hard" comment means especially when you come up against fixed dates.
"... Here's a very rough mockup that was put together with some brainstorming, just to give an idea of what the first user introduction might be like ..."
Better. Now make some more that demonstrate your key points and that will also help you start building. Wish you'd put this up first as it starts to put flesh to words (idea description).
Yeah, steffon and I are working together. Between the two of us, we can make a pretty great interface that is very focused and has defined boundaries. The problem of boundaries tends to be what most potential investors are worried about. We've definitely got that concern covered.
What we need is someone who is amazingly good at neural networks, algorithm design, and advanced mathematical engineering with programming languages. These people don't grow on trees; they're extremely few and far between. But the main asset for this idea isn't the idea itself, or even-- to some extent-- the user interface, but the way the math works behind the scenes. That's our innovative secret, and it's the reason I'm excited.
1.) One of the main problems I am trying to solve is in your second sentence: "Once you find a chan you like..." It's time consuming to find new content you like, especially in a setting as nebulous (for most people) as IRC. With the discovery engine, all a user needs to do is rate recommendations based on what they are interested in, and they are immediately connected with groups of like-minded people and what they know. Fast and easy.
2.) In addition to connecting to groups of like-minded people, the user gets recommendations from those who are most influential and most in the know. All content is not equally desirable, and users using the discovery engine will only get recommended that content which has empirically shown is wanted by people that think like them (relevant).
Of course. I only mentioned IRC because there are some similarities that seem to indicate your idea could work.
>only get recommended that content which has empirically shown is wanted by people that think like them
Seems like an OK idea, but I think the site might suffer if this is taken too far and the groups just become echo chambers, walled off from each other entirely. I just think a site could be more popular if it focused on finding common links in disparate groups instead of isolating them based on common differences. Think of it like radio: in my music directory, my biggest artists are The Crystal Method (41 songs), NIN (112 songs), and Sarah McLachlan (85 songs). No single radio station, AM, FM, or XM, that I am aware of covers a range like that. Wouldn't it be cool to find a place that let you keep close to your core group, but at the same time gave you tips on what other groups thought was cool, just in case you agreed?
"groups of content" are collections of content that users create themselves (the social bookmarking aspect of the site). Users that want to build a reputation for being in the know and influential are incentivized to make these collections relevant. The groups also become lists users can form to bookmark content they find around the web and want to save it one spot.
Based on what you have put into your collection of content, and collections others have made, the algorithm aggregates those similar people into the same network. People in the same network get content recommended to each other from their like-minded peers that they have not discovered on their own.
With regards to the feasability of such an algorithm, I've talked about it with many mathematicians and machine learning programmers and the wheel does not have to be reinvented for this application. The tools already exist, and just have to be customized and tweaked for this application.
In other words, I make a "group of content" that I call "movies" and it gets compared with other peoples' "group of content" that they called "movies".
Why isn't the netflix recommendation system useful for generating recommendations from "groups of content" labelled "movies"?
If it works for movies, why can't it combine "groups of content" labelled "science fiction"?
The point of having groups of content is to combine traditionally dissimilar types of content-- movies, music, housewares, etc...
For example I can make a group of content called "new living room" and add all of the things that go into my new living room. This includes the music and movies that I have stored there, the type of TV I bought, the type of couch, stereo receiver, speakers, or even the paint on the wall. When someone searches for something within that collection, the system knows that someone else, somewhere, has combined that "thing" with the other "things" in the collection, so they get rated higher as being compatible.
This page explains many of netflix's limitations well: http://harry.hchen1.com/2006/10/03/391. But more importantly, look at these limitations in light of how the discovery engine is organizing its preference data and how it's collecting preference data.
The critical difference with the discovery engine is the idea of a group of content that users fill themselves with content based on criteria they see as relevant. Yes, a users aggregate preference composition is important, but what is more important is their set of preferences regarding a specific collection of content. This way, a user can be really into classical music, horror movies, and modern furniture, and get relevant recommendations for each interest, connecting with people who are most in the know regarding each interest.
I'm not sure what you're asking here-- Netflix doesn't care about anything but movies, and it probably wouldn't be able to recommend movies any better if it knew your musical tastes, or even how your tastes compare to mine.
The idea is that if you search for "sony SSK70ED," on the "discovery engine," it will show you what other people have paired with those speakers, such as receivers, furniture, and televisions. In a way those things are "similar" to the speakers because they complement them. Of course, the system shows you similar speakers first, but the complementing items are interesting results to have when you're searching for a specific item.
> I'm not sure what you're asking here-- Netflix doesn't care about anything but movies, and it probably wouldn't be able to recommend movies any better if it knew your musical tastes, or even how your tastes compare to mine.
That's wrong. The only thing that is movie-specific about the netflix recommendation system is the preference data that it runs over. It doesn't know movies from eyebrows.
If netflix (the company) also collected preference information about music, the recommendation engine would predict music preferences. And, since it would have both music and movie info, it would use music prefs to recommend movies and the reverse, just as it uses movie prefs to predict movie prefs today.
Amazon's "users who bought {something} also bought" is an example of "doesn't know anything about the domain". (They have to tone it down to keep it from recommending "strange" things that are way out of category.)
Disclosure: I know the guy who implemented NetFlix first recommendation system and have written a collaborative filter myself. I know what I can do with the fact that we both like the Pogues and Chunky Monkey. I still don't see what I can do with how we group those preferences.
"I still don't see what I can do with how we group those preferences. "
The way in which this site will allow users to group their preferences seems like a slight organizational difference when compared with other recommendation sites that use collaborative filters, but it has huge implications. This post is meant to give people a taste of what I'm starting to try and find others who are interested. I'd love to talk about any and all specifics and their implications especially if you are programmer. If you want to chat my AIM is rocksld3.
I've got a few ideas, such as front-loading the site with content and preference data by using differnt API's, such as Flickr's API, Youtube's, del.icio.us, Last.fm's. This way when you showup, lots of your preferences and rating "work" comes along with you. Additionally, the preference dataset is jumpstarted.
What you would tell your little sister about the site is "find more things you like, and if you're good at finding new content, be recognized for it."
I agree: I rarely find that suggestions based off of anonymous, aggregrate consumer "social" data helps "pick" worthwhile suggestions for me. If a category of shopping is completely new to me, then what everyone else knows can make me more aware of the commonplace options (which is helpful). But if I know anything about the category of items I am looking at, I really don't care what the anonymous masses already know because I know it already too or it's a very "obvious" connection.
My thoughts exactly. At least it will give those outside of the US (or an aversion to AT&T service plans) a chance to enjoy part of the iPhone experience.
I completely agree with this article. And a startup that could efficiently tap into people's hunger for fame and recognition on a broad basis could become a destination for user generated content that far outdoes current web 2.0 offerings.
"We observed that users cite a variety of reasons for posting content online--chief among them, a hunger for fame, the urge to have fun, and a desire to share experiences with friends"
What's interesting about the notion of reputation and fame is that, for a user to feel like they belong or are "having fun" or a held in esteem by a group of people, that group needs to be a compatible reference group meaning, they need to have things in common (this idea was touched upon in the article "The Problem with Social News" http://news.ycombinator.com/item?id=50015 and the discussion of homophily, meaning loosely birds of a feather flock together). For examples, entrepreneurs are much more likely to be concerned about the opinions of other entrepreneurs than other people. I take more seriously the music suggestions of people that have the same tastes as me (obvious, right?).
So if everyone on a site could be plugged into a compatible reference group, the fame and reputation motivation could be leveraged across much larger populations of users, not just the ones that want to be, for example, on the top viewed videos on youtube. Each user would be compared against those users with which they have common interests, so the idea of having a reputation becomes more salient, so people spend more time contributing, and the content becomes better for everyone at an individual level. Lots of sites are able to harness the reputation motive on subject specific areas (like Y Hacker news), but no site has tried to harness the motivation for reputation across all interest groups. Definitely email me if this interests you because it fascinates me and I have plans to build such a site.
But would the Digg of music be enough? iJigg, which is something like that, launcher earlier this year. http://www.techcrunch.com/2007/01/18/jigg-that-music/. Although it doesn't seem to have moderaters with real music knowledge. But even if it did, that wouldn't be enough to make results relevant for everyone because the criteria for relevance is an individuals concept of "good". Sure moderators might know what is "good" for certain audiences, but if results are organized in a digg-esque fashion, it becomes a top-ten list, meaning that no matter how good your moderators are, they can't generate recommendations that will be "good" for individuals (and the strategy to appeal to the largerst number of individuals is to appeal to the lowest common denominator, which one could argue radio already does well.)
The key with these communities is seeding them with, OMG, domain-specific talent.
I can imagine a social platform where various DJ's, music supervisors, musicians, etc, could be the curators of long long long playlists [organized by mood, genre, style, time period, lyrics, whatever] that are tagged with a bunch of metadata [enabling them to be recombined easily].
Non-curators can suggest tracks and gain authority by having them added to curated uberplaylists.
Or something like that ;-)
The key is knowing your ass from a hole in the ground about music...really. Not just thinking you have good taste...that's how all the media execs (that all tech people hate) got into this mess in the first place.
Anyways, I'm already on this, if anyone is interested further you can contact me ;-)
I don't think that it is entirely accurate to equate news delivery with music delivery, nor do I think the lessons from journalists are entirely relevant when it comes to building a social platform for music playlists. Cultural objects (e.g. music,art,fashion,film,literature etc.) are substantively different than news articles.
The news is largely valued for its accuracy, timeliness, and topical relevance to a reader. Music and other cultural objects are valued for their enjoyment, which is contingent on individual personal preferences. I'm skeptical of a platform that would use domain-specific talent to create playlists for what comes down to individual personal preference. Domain specific talent makes much more sense regarding news where the relevance can be easily determined because it is largely topical, whereas the relevance for music is based on personal preference, which can't be determined through domain specific talent.
I'm also not convinced that more meta-tagging is the answer either. If I tag a song with a mood, genre, style, etc., it will help a user find that song based on those new categories. But in the end, finding new songs based on categories isn't that helpful because what matters is finding songs based on individual personal preferences. For example, I might know I want an energetic, intense, electro-rock ballad, but defining that domain doesn't guarantee I will like the results. People rarely like all the songs on an album, and as far as domains go, albums are very tightly defined domains. Likewise, using Uber-Dj's as domains don't seem to be the best answer. After all, their playlists are based off of their preferences, and, regardless of how much they "know" about music, unless their preferences are like mine, their list will not help much.
You're being pretty dogmatic about what music is...
You may be right that, in some respects, "cultural objects are substantively different than news articles", but not in ways that are germane to this argument:
"The news is largely valued for its accuracy, timeliness, and topical relevance to a reader. Music and other cultural objects are valued for their enjoyment, which is contingent on individual personal preferences."
I addressed this in an earlier response..."personal enjoyment" of music is inexorable from "timeliness" and "relevance."
I get more "personal enjoyment" from John Zorn than anyone else, but I don't wanna hear it when I'm making out with my girl, or at a club.
Think about the success of mood/circumstance specific mix CD's like "Ibiza Club" or "Ellington for Lovers"...
Musical preferences have a lot more to them than "people like you also like"...
Also, with all the choice available, it's easy to overestimate people's desire to even HAVE choice.
Finally, I am imagining a more sophisticated metadata model that is not just about creating categories for things. I don't really have the details on that, though ;-)
The examples you use as counter-arguments to my approach are not really counter-arguments, but deal with a different issue than what I was talking addressing in my last response. I was talking about finding music based on personal preferences. Your counter-examples of "making out with my girl" or dancing "at a club" are group situations where group preferences are most important.
You are right when you say that compilations like "Ibiza Club" or "Ellington for Lovers" make a lot of money: so does selling Muzak (the background music in supermarkets and most commercial spaces with music). Background music is the ultimate "genre" that is compatible with group preferences: no one is offended... but at the same time, nobody really cares.
I agree with you that musical preferences have more to them than "people like you also like"... this would disregard the reality that people sometimes do categorize music by situation/mood, categories like listening to music to dance, exercise, study, make-out, host cocktail parties etc.
But as I suggested before, domain specific categories of music do not have to be mutually exclusive to personal preferences (currently they are). Let's say your goal is to have romantic music. Why not use a "people like you also like" function bounded by the category of romantic music? This way you could get romantic music, i.e. music everyone thinks is romantic, and romantic music that you like as an individual. Even better would be to use a "people like you and your girlfriend also like" function within the specific category of romantic music :)
"Also, with all the choice available, it's easy to overestimate people's desire to even HAVE choice." This is pretty fatalistic don't you think? I think the explosion of choice online frustrates people because they KNOW something is out there that they will really love, but they can't FIND it. This screams opportunity for a website to act as a choice agent to direct people to the music, video, merchandise etc. that they want but can't find themselves among the infinite choices. Infinite choice results in infinite search costs without a decision agent.
Okay, I think I have identified a central reason why we disagree on some of this stuff, and it basically stems from views on art criticism.
The phrase "personal preference" to me makes a case for relativism:
"If I think something is good, then it is good."
WRONG.
As Bruce Sterling and others have argued, just because you like something [or just because you can make something, like a mash-up] doesn't mean it's good.
I don't have time to parse this much more right now, but the fact is that it behooves technologists to develop a robust art criticism within these applications.
Another reason I am high on the idea of people making decisions about what to listen to vs. algorithms [if there has to be a choice] is that, while there is a LOT of choice in the music world, it's not even CLOSE to infinite.
I believe that among the 10,000-50,000 (a pretty random number of people i picked), "legit" (whatever that means) music aficionados could parse out the vast majority of great music in all genre/mood/styles, from Britney to Bach, in a fairly short amount of time.
Also, this paragraph:
"You are right when you say that compilations like "Ibiza Club" or "Ellington for Lovers" make a lot of money: so does selling Muzak (the background music in supermarkets and most commercial spaces with music). Background music is the ultimate "genre" that is compatible with group preferences: no one is offended... but at the same time, nobody really cares."
Is utter bullshit. Do you honestly think I'm thinking of it that way?
Finally, let the record show that in all my hundreds of hours listening to last.fm and Pandora [PURELY for research reasons] I have NEVER EVER heard anything that I wasn't already aware of that was half decent. EVER.
So, my perspective on this is skewed. Like everything else, 95% of people don't know shit about music, so these incremental algorithmic solutions are a panacea to them.
Your idea of weighing "influential" people heavier in an algorithm has a lot of merit. I just think that pure human-powered peer production would get to the solution of better music for every occasion a lot faster.
This makes sense to me for publishers and writers that aren't in the top 25% of books being sold. Marketing for many consumables can be thought of as a two-step diffusion process: Diffusion step 1. Utilize the mass-media to send your product message toward a target audience. Diffusion step 2. Hope that the early adopters and opinion leaders within those audiences tell their friends and exercise their networks of influence so that "word-of-mouth" perpetuates the message first sent by the mass-media.
For many cultural product industries, like publishing, they release many more products then they can afford to buy mass-media advertising campaigns for. For those products without mass-media coverage, it makes a lot of sense to give away free copies of the book. The innovators and early adopters that crave and promote new products have the material they need to jumpstart the second step of the diffusion process (word-of-mouth), and a publisher and writer could see higher sales numbers than they otherwise would have: they still benefit from second stage diffusion without the mass-media costs.
And if there is a concern that the "message would get out" that the book is free online, the publisher could only give away the first 12,000 free online. It seems like putting a cap on free copies could maximize the potential of "word-of-mouth" diffusion and limit the risk of lost sales.
