You can read my whole spiel at http://www.circleofconversation.com/#tab=theory, but basically the idea behind the app is that people like talking to strangers online(like we're doing right now) as long as there is a sufficiently precise topical specifier(like we have here). The app attempts to accomplish this by letting users give their chat rooms tags which fade out over time and then disappear if they aren't refreshed/spoken aloud. They're constantly replaced by now-accurate tags, so ideally every conversation has a very precise set of topical specifiers at all times, making them worth joining.
I'd love to get some feedback on the general concept as well as execution specifics, though the UI is rough and the feature set is basic. Thanks, HN.
P.S. I'm intentionally posting this in the evening to try to avoid any significant traffic, and to give myself time to hot-fox any glaring bugs, since I do client work during the day. I bet this sounds familiar to at least half of HN's readers.
Also, after I signed up with a username (and a blank password!) I was not able to see any text that I typed in the chat. Then when I came back later and tried to login as that same user, it just redirected me back to the front page of the site. When I navigated back to the chat, I wasn't logged in. This is all on Firefox 3.0.1/Linux.
Anyway, very promising. Keep on hacking!
I didn't completely understand it until reading your comment.
Seems very difficult to get off the ground without a large, extremely active userbase.
I don't use twitter or friendfeed but it looks very similar to it.
Still not sure if I see a real use or appeal for it - I could be wrong. I'd like to see how it plays out after some time.
Domain name is wayyy to long..
Nice work so far though.
I'm finding it hard to sum the site up in a phrase, but I know this is important.
>Seems very difficult to get off the ground without a large, extremely active userbase.
Yep, same old problem. I don't have a marketing budget, so I'll just have to try doing it the hard way.
>Domain name is wayyy to long..
At this point in the game it's either made-up words, non-descriptive domains, or descriptive, long domains. I like conversational speech, so I went with the third option. I think it's very easy to remember, which may offset the length. I think it's funny that we consider three words so long when it comes to the web, but I basically agree with what you're saying.
I read the whole Theory page and it struck me as blah blah blah blah blah, except for these two points:
1. Conversations on topics you want, going on right now; 2. Karma keeps the quality high.
To me those are the core, and they're a pretty cool core. If I were you I'd strip everything else out. And emphasize those two things on the main page.
Personally, I'd drop the recommendation system, or at least drop talking about it. There's so much tripe surrounding that kind of thing that, when I read it, I noticed that your credibility instantly dropped in my mind. I'd rather you put your effort into making it really easy for me to browse/search what's going on so that I can make my own choices. Recommendations won't make me want to use your site if there isn't a lot of value there already. And if you do develop a kick-ass way of recommending related stuff, don't talk to me about it, just show me a few suggestions unobtrusively and I'll eventually catch on.
I also like the slogan "chat rooms that really work".
Edit: on reflection, I seem to notice a pattern which, if it really is there, I suggest you avoid like the plague. It is the idea of the system being "intelligent". It comes out in details like your choice of the word "Theory" and the title of this post, as well as in the emphasis on the recommendation system and its overtones of AI. To me this is a big turn-off. I don't want a chatting system, and I certainly don't want an "intelligent" chatting system. What I want is intelligent conversation. It seems to me you've got the kernel of something that could actually offer me that, which could be genuinely valuable. Focus on that and try to get the system out of the way.
Edit 2: while it's fresh in my memory... I noticed another thought going through my mind which was, "I wonder who is on here that I know from somewhere else", like HN for example. I wonder if people would be willing to link their user names to their names from other communities. I would, if I knew that lots of other people from HN were talking about stuff on your site.
It's clean and straightforward and mostly gets out of the way and creates a good space for conversation.
Karma should not be limited in time., think of it this way; in CofC karma is like facial expressions. If you want to experiment with Karma try a decay function where it returns to 1 after a few minutes, and where a users karma donations reset the clock.
Originally there was no recommendation system planned, but once I realized how easy it would be I had some infatuation-based feature creep. Going back to the drawing board with 'don't tell them, show them' as priority 1.
Also, showing more tips hides the input box. It prevents people from being sure they're implementing tips right, and adds more complexity to saying things for new users.
I think it's pretty clear websites continually evolve, grow, etc.
Your approach to user signup is superb.
There should be an explicit "tag this conversation" link. Otherwise nobody will even know that they can do that.
Can I post a link to an ongoing conversation in some other place, like Twitter or, say, here?
The "Theory" page is... a bit too theoretical. I mean, everyone here will get it, but not the mass market. I doubt that many AOL Chat users understand what "asynchronous" means. Don't delete the existing page... just make a slightly simpler, shorter Theory page and relegate the existing text to a second-level "Theory of the Theory" page.
Other than that... it's hard to know what to critique. Many of the essential design features of such a site will only become obvious when it's flooded with traffic or overwhelmed by griefers. That's hard to test. Maybe you should start a really provocative conversation and then post a Digg link to it. ;)
You can post the link from your address bar, but the next step will be to make each tag and conversation a static URL which can be crawled by Google.
The theory page was basically written for the HN and reddit/r/programming audiences, because I wanted to get some serious feedback on the concept itself. In the next month or two everything will be made more palatable.
Thanks for the feedback.
Suggestion: Add up-down voting for tags/topics, then rank the tags/topics on a redesigned front page that looks more like HN or Reddit. You could make the site behave like a HN for which all link submissions were self-referring (like this one), and all commenting/conversation happened in real-time.
Tagging a conversation with a pronoun like "I" will probably means it sticks around forever.
Can you join multiple conversations at once? What happens if two conversations share a tag? Are they merged?
To stop conversations being tagged incorrectly, you could perhaps ask users that join via that tag if it was accurate or not.
It'd be awesome if you could automate tagging of conversations. I'm not sure how many people will be motivated to maintain metadata about their conversations (the only way to find that out is to get more users though).
1. Some conversation and/or the site gets dugg
2. Thousands of incoming users waltz in with good karma
3. Everything goes to hell
Wikipedia, Everything2 et al. have some resilience because the audience is more than just who's using the site at the moment. Dunno if you've thought about that in great detail or not. Also,
* Friend system. Encourages people to band together against the trolls
* Bookmarking
Perhaps an obviously marked URL to copy/paste or something like that?
"Coupled with an analysis of your actual chatting history, this data lets Circle makes helpful suggestions as to what conversations you're likely to enjoy."
"makes" --> "make"
