Comment by cableshaft | Hacker News Reader

Comment by cableshaft | Hacker News Reader

Parent Full thread

Anyone who can take this on in a better or with a different approach to what already exists would be greatly appreciated to creators. I'm one, and I've been frustrated with the state of many discovery engines out there.

I've been toying around with the idea of some sort of P2P discoverability engine myself, but I don't really have the time or energy to really tackle it myself.

Just improvements on a 'recommendation algorithm' isn't enough, to me. You need to provide many, many vectors to get from one piece of content and discover something else. The more direct connections, the better, kind of like a wiki, but focused on community provided connections.

It's kind of like when I'm looking for new games for a console and I search "Top 10 games for blah" and check a bunch of list or "Underappreciated gems for blah" and check a bunch of lists, hoping for something I haven't seen before. Well what if there could be associations for just about any aspect of the content. Maybe the nodes could be auto-weighted based on how many people follow that connection, so anyone could add a connection, but the most travelled would rise to the top when a bunch of people click on it, possibly with some sort of extinction.

Anyway, it'll probably never materialize, it's just something that's been rolling around in my mind for awhile.

Doesn't this sort of already exist? Dribbble for instance, you can have followers and usually the top artist have tons of activity through comments on their content. You can easily find these top artists through their search filters.

No, not really. That's more of a standard web 2.0 approach to content, almost like Tumblr, from what I can see (likes and comments probably wouldn't be a thing in what I'm invisioning, for example, although following updates to the page probably would be). And I don't literally mean "Top 20 most popular whatever", as existing portals do that already, as it's easy to do a database query and order it based by views, or sales, or whatever.

I'm saying I currently watch like 15 different top 20 lists from a bunch of sources just trying to find the 4 or 5 unique items because they all tend to mention the same stuff except one or two things. I'm trying to find the more obscure or weird stuff that's still good, not just the most popular.

What I have in mind is where the focus is more on the connections between two specific pieces of content, that goes beyond a general recommendation engine and the connection can be based on many things. A tag doesn't fall under that category because it will just list anything that matches that tag (although a focus on tagging is at least better than what the Apple App Store has now, for example), and the tags are usually decided by the content creator, not other people who might see a connection that even the author might not.

I'm thinking more like on Wikipedia, where you read one subject, then on that subject you find a link to something else you're curious about, and go there, and then on that page you find something and go somewhere else, etc. The connections are community driven (each connection can be strengthened or weakened by the users), and the reason for the connection can be whatever is provided by the user that made the suggested connection.

Additionally, I'm thinking of using a metaphor more like exhibits in a virtual museum instead of 'hey, i made this thing, come look at it!'. The creator isn't necessarily the person who 'curates' the exhibit, or at least not the only person. Museum might not be a hip enough to attract people, so there might be presented differently, but the idea is a lasting place people can go to find out about that content and a bunch of links linking to it and from it from other content on the site for all sorts of reasons.

The problem is that often there is something you made that would be appreciated by, say, a million people worldwide.

These people are mostly unconnected with each other and are not aware of your existence. Even if a few would find you, there is no chance of the typical friends-tell-friends chain reaction of discoverability. It would die out, since most of the people in-between won't find your work interesting.

So, how would you make it easier for the random-nobody creator to reach his sparse-million audience?

Reply on news.ycombinator.com