First, Eiron, thank you for raising this question. This is something that I have thought about at length. I appreciate your insight.

Second:
] the things I'm describing may not actually be the vision for
] this community. As it's not my project, I wouldn't presume to
] argue strongly for such a shift.

Do not let this dissuade you from engaging the conversation. A community cannot be owned. There is a delicate balance. I cannot simply take everyone's suggestions. Everyone always thinks their perspective is the right one. But I strongly encourage people to think about how this community should evolve. If I was as good as I wish I was this thing would be bigger then Google. I strongly encourage this kind of conversation, especially on the site.

] As it turns out, this stuff is all pretty hard, and completely
] different than the existing infrastructure. As a concept, only a bare
] germ at this stage,

Exactly. Thats why this dialog is useful. I don't know the answers here.

] i think there's merit, but I know there
] are other priorities right now, even if this was a direction
] for the future.

Thats true, but there is no reason this conversation cannot proceed while other priorities are pursued.

There are several perspectives on this, and each one presents problems.

1. There is a tremendous power involved in associating everything on this site by particular URLs, and one which has not really seen much application yet because of the scale of the site, and that is the discussion bookmarklet. MemeStreams associates a discussion with every url on the internet, and using the discuss bookmarklet you can jump right from any page to it's thread on memestreams. Once the site is sufficiently large this will be extremely powerful. Anytime you are reading an article and you want some more insight or critical commentary you'll be one click away from it. I think this is something the internet really needs that no one is doing right. I want to see it happen.

Having said all of that, you clearly need to have a way to put multiple URLs in a post and at least have them be clickable. Thats a basic, basic thing that I need to implement.

2. The topic system. The problem with topics is that there aren't enough of them and things need to be filed into multiple topics.

If we created enough topical granularity to be able to discuss something as specific as the tsunami the topic system would be unwieldy. Also, the topic system reflects a cultural bias. We have a 9/11 topic. We don't have a tsunami topic. If we lived in South East Asia our priorities would likely be reversed. We'd see 9/11 as a transient event and the tsunami as something that would impact our culture for years and require its own discussion space.

If we allow for things to be put in multiple topics this creates some significant user interface problems. Do you really want a way to file something under 12 different places? Would people use it. How would you indicate that on screen?

3. Taggle: I recently memed some interesting content from slashdot about generic tagging systems like the one used in del.icio.us. They have the advantage that you can come up with a tag on the fly like "tsunami" but the disadvantage that every version of that is its own tag "tsunami, asian tsunami, tsunami 04, etc..."

One of the things that del.icio.us does that is cool is associate topics with eachother. This hopefully mitigates some of the messiness. Maybe I should move to a system like this?

4. Bayesian Filtering: Google news does a really good job of associating related stories with eachother. I don't know if they are using Bayesian filtering, but it would serve such a purpose. I might be able to group memes this way.

A simpler approach which Jeremy suggested quite some time ago would be to have a "more like this" link on every meme which culled out keywords and searched the site for them.

Bayesian Filtering might do a very good job of associating related concepts together.

5. Communities. There needs to be a way to form communities within the site. Small groups of people with a shared interest. Communities are interesting because they might also be thought of as folders. A community with only one member is just a folder for filing Memes.

One of the ideas that occurred to me was the notion of tying communities and topics together. Instead of having a list of topics next to your memestream, you have a list of communities. You can make communities on the fly. Anyone can join them. In fact, they work a bit like Taggle. But, they are associated with topics. Multiple topics. If I create a Georgia EFF community I might associate it with politics and technology.

When you post memes you don't file them in a topic, you send them to a community. They get filed under the topics associated with that community. The topics they get associated with most frequently are the topics they are likely to appear under in the timeframes view on the topic pages. Collectively, we'll file them right even if individually we might not. Thats the theory anyway.

RE: Links, Memes, and Memestreams