What do you think of this new Social Network website idea

[GrigoriySidelnikov]

A website I have been developing for several years is called Authentic Society.

Idea: I want to make a "Pandora for Everything"

The idea appeals to me personally, and that is the only reason I continue working on it. I like it because it will allow people to find new things they will like, but didn't know existed.

How will it work? Tell the site about your favorite movie (or any other interest) and the site will instantly know what books, music and/or a multitude of other seemingly unrelated items, that you are likely to enjoy. Suggestions would be based on common properties of each item.

I have spent literally years of research on developing this idea. So I already have intricate details worked out and it's just a matter of building a proper user interface to establish its function.

One day I imagine logging into the site myself and specifying my own interests and discovering something I really like but didn't know about.

The site has integrated social network controls that resemble the basic function of modern social network websites.

The main purpose is to provide the means for "discovery" of new things people will love.

[High Priest]

All you are doing is just integrating the existing service into a single website. Amazon suggests books related to the ones you bought. There are hell a lots of music websites which suggest the kind of music you might like based on your favorite bands and singers. I believe there are other websites too which do the same for other things.

Are you providing some kind of service that is completely new???

[Neobian]

Be specific - you want it a suggestion site or social site? I do like the idea of getting suggestions from a single site. Do you have the link?

Dont spend too much time on research, do a quick prototype and keep validating. It serves two purpose. Practical-ability of the idea and market penetration with brand image building. You may come up with best product but people already invested in inferior substitute and it is lose-lose for both you and customers.

Do a market research to find your competitors. Usually same ideas appear to many people .. some ignore it, some do it wrong way, some do it late, only few get it right on time.

[ssandra]

I like it!!

Especially since for example Amazon might suggest that I like other books based on those I've brought, but they don't suggest music. Or websites. Or things...

I'd go with a suggestion website coupled with a forum for the social aspect.

Let me know when it is up, because I'd love to be part of it!

[Brutha]

I think the important part will be about important data from other sources. As far as I remember Last.fm has an API that allows data transfer.

Are you providing some kind of service that is completely new???

That's a bad question. Business isn't about providing something that's completely new but about providing value to people.

[SatvikBeri]

It is extremely useful, but there are a lot of practical challenges.

A major one is just the amount of data you need. Figuring out what movies someone likes without having attributes like category (Horror, Comedy, etc.) and a huge amount of ratings already is next-to-impossible.

I don't mean to be negative-I would love to see this website work-but I know enough about this problem to know how hard it is.

[Brutha]

A major one is just the amount of data you need. Figuring out what movies someone likes without having attributes like category (Horror, Comedy, etc.) and a huge amount of ratings already is next-to-impossible.

I don't think having existing categories is that important.
Users can tag different items and provide him with information.

Having a way to manage ownership of items would probably be great. If I could scan all the books that I own over their barcode with my Android phone I would probably do it.
I could share the list of books with my friend and loan them books that I own that they want to read.

If I have the list of my books in the system I could star them quickly and you could give me recommendations.

As a sidenote, if I wanted to build something with social network functionality I would use Pinax.
It also worth to think about whether you want to be a facebook application.

[GrigoriySidelnikov]

Thank you all for your valuable feedback.

Brutha, what is your experience with business? You make valid observations - it seems like you are involved in some way. Perhaps you have built online applications before?

Yes, entering and categorizing data in a way that makes sense is the key problem that needs to be solved. One would be surprised on how difficult it is to categorize something, even though the word "categorize" sounds orderly. The process of categorization information is not as pretty as it seems it would be until you attempt to actually do it and learn about the obvious problems. The main problem lies in the difference between properties between different things.

I'm not referring to any person in particular in what I say below... But this thread has sparked the conversation on...

On being negative...

When I was doing my first oil painting without prior experience working with oil, I wasn't thinking about how impossible it would be to do this. When I was working on my first social network site that people liked better than myspace, I didn't think someone would willingly offer and ultimately buy the domain name from me for $1,300. When I wrote my first puzzle game for the PC called Briefcase, I didn't know someone would ever refer to it as "Great game, played it as a kid...". I find this especially humorous because I created it as a kid.

The bottom line is: There is no excuse for thinking "You can't do it" or "It's impossible", when you attempt to create a great online service, or anything else for that matter. If you don't believe you can accomplish something, then you are just not the right person for doing it. BTW-Steve Pavlina knows this, too. And you are posting on his forum.

When I painted my first oil painting, it gave me confidence I could do other things without prior education as well. I proved it to myself. Whenever you solve a problem, you can use that solving skill in many other areas as well, it sticks with you psychologically.

Most of the things you will attempt to build to completion are extremely difficult to do. I don't care if it's a video game, a painting or a suggestion site. Or whether it's Carmack porting Mario to the PC when everyone thought the PC didn't have the firepower to do it. It is essentially the same thing in terms of your ability to get it done.

Do you think Tim Westergren thought: It's impossible to create a personalized radio station every step of the way while he was trying to create it? He had people work part time for free to make it happen, after he ran out of first round of investment in a year (1 mill I believe).

People don't believe in their abilities every day. But it is up to them to make that decision. Because it is... a decision. You either go all in, or you don't do it at all -- BUT -- if you chose not to do anything at all then you should not also tell other people who ARE motivated that they can't do it. Because essentially those who have the motivation and the ability to do it are doing it for all those who are lazy and are not willing to do it.

[SatvikBeri]

Originally Posted by Brutha

I don't think having existing categories is that important.
Users can tag different items and provide him with information.

Of course-it's just a matter of how *much* data is needed before the predictions start becoming useful. With Netflix most users need to enter ratings for 200 movies before the recommendations start becoming really useful, and Netflix has one of the best recommendation systems out there.

The more you have pre-existing data/categories, the less data you need-and that makes it easier to achieve a tipping point in terms of numbers of users and recommendations needed for the site to be useful.

Having a way to manage ownership of items would probably be great. If I could scan all the books that I own over their barcode with my Android phone I would probably do it.
I could share the list of books with my friend and loan them books that I own that they want to read.

I agree 100%-the easier it is for users to enter their preferences, the better. It would also be useful if the website could look at the books they have on a kindle, or the reviews they've posted on Amazon.com, etc.

Another example would be integrating with 4square checkins to see what kinds of places the user likes.

[GrigoriySidelnikov]

Of course-it's just a matter of how *much* data is needed before the predictions start becoming useful.

It will be so-designed that you will only need one item for the site to know quite a number of interests. I don't want to sound mysterious about this but like I had mentioned, I have spent years on developing this idea theoretically and I have over 100 sheets of printer paper with ideas written down on them.

But nonetheless, it is true that in either case there will need to be a lot of data in order for the experience to be meaningful (but not solely for the reason of making a perfect match, which will happen even with a small data-set as long as it contains similar items). The match is completely reliant on the data structure itself, not the amount of data, the system is designed to match lots of things with a very small number of initial items. The more data the better, of course.

[Agota]

I think that is an amazing idea!

I would certainly use a site like that!

[seanner689]

Interesting idea. It seems like this sort of thing would be perfect for a Facebook app, since users typically enter their fav. movies, books, activities, etc there already ... and upon granting you access to their info, you'd have a pretty complete profile of all their interests to work with.

I'd be interested in seeing how strong the correlation is between somebody's book interests and their music interests, for example. I always assumed that stuff is fairly independent (aside from certain themes like religion).

From your example, it seems like you get a list of recommendations from entering one song or book that you like. Is that correct? This seems weird to me, since presumably everyone who entered that song/book would end up getting exactly the same list of stuff. It seems like the power of a site like this would involve looking at a full profile of many users across all interests to get some correlation data (ie their google searches, FB interests/likes, etc), and then getting as much "interests" data from each user as you can (ie, access to their FB info) in order to make the best recommendations.

There is a massive torrent out there, from what I remember, with a huge number of people's Facebook profiles (whoever had them shown as public). That would be a great resource for looking at people's interests and how they're linked (if you're still in the developing stage). Alternately, it would be and a great resource to test your own engine ... you could write a script that picks a user, enters some of their interests into your recommendations engine, and then looks to see if it picks out any of their other interests.

Sounds like a useful app if you can do it right ... good luck.

[GrigoriySidelnikov]

Originally Posted by seanner689

From your example, it seems like you get a list of recommendations from entering one song or book that you like. Is that correct?

One would be enough to trigger the chain reaction. But it's in best interests of the site to learn as much as possible. New suggestions will pop up, people will like them, too... these new items will link to yet a number of new suggestions, etc.

This seems weird to me, since presumably everyone who entered that song/book would end up getting exactly the same list of stuff.

No, the site will be aware not only of what you like, but how you like it. That is because there are ways for people to like the same thing for different reasons, and I think the discovery engine should take this in consideration.

Getting the data from other sites is not going to work, because there isn't a site (perhaps other than Pandora) that exists right now that understands exact preferences of an item, rather than simply linking them together for some unknown reasons.

[Brutha]

A question about your website:
"Say I enter my information and get recommendations once, why should I come back in the future?"

Brutha, what is your experience with business? You make valid observations - it seems like you are involved in some way. Perhaps you have built online applications before?

I'm sort of building an online application on my own.

In general most of my knowledge is more theoretical. This week I will visit the UXCamp Europe for the third year. The spirit of the event is that someone in the room holds a speech and says that the UX documention of Android is crap. We should do something about it.
Another day, there a discussion about the website of the biggest German newspaper.
The guy responsible for the website is in the room but not everyone knows his identity.
Someone says: "Looks this website is ugly, but it's supposed to be ugly. The target audience doesn't expect a quality website. It designed by some agenture who knows what they are doing."
Without identifying himself the guy responsible for the website says: "No, it's no agency, it's inhouse".

Being around those people somehow facilitates learning in a way that similar to learning a foreign language in the home country instead of learning it through your own work

[GrigoriySidelnikov]

That's funny.

A question about your website:
"Say I enter my information and get recommendations once, why should I come back in the future?"

Because you may want to continue discovering all other items you like (based on similarity).

For example, let's say you like the game Starcraft II. Your suggestions will be other Blizzard games, and also other real time strategy games, not necessarily made by blizzard. But one of the first suggestions will be Warcraft, because it is an RTS and it's made by Blizzard. The site will feed suggestions to you one at a time, or in small quantities.

There are obviously more than 1 thing that are similar to any other thing. I am thinking that this process should be gradual... without overwhelming the user. So, today it suggested you an RTS, tomorrow it will suggest you a fantasy book, because there are orcs in Warcraft, or something like that. The site will actually tell you why it suggested you something. Like: "Hey look there are orcs in this other thing, too". But then the user has the choice to say "Hey I didnt really like orcs ... I like the undead race..." so the site takes a note of this and says: "Ok, here are some books about zombies". Or some stuff like that.

[Anthoneee]

I would like it if it would introduce me to like-minded people in my area.

[GrigoriySidelnikov]

Thanks for everyone's input. It's nice to see that people like the idea. All of the friends I have talked about it also liked it, so this may be a good sign.

I have been improving the social network site I mentioned in this thread. It's not quite there yet, but there are significant improvements. I added a link to my social network site to my new signature.