Just a small update today.
About a week ago, lesath asked: "Is there any way to browse through tags to find interesting ones to follow?"
There aren't enough.
To begin to remedy that, today we extended the lists of tags and users on the community page. Now, on the bottom of each list is a "more..." link, which will bring up an extended version of that list.
In addition to this, we are currently building out the Hubski search functionality, so that more information can be returned, and more advanced searches can be performed.
As always feedback is much appreciated! Specifically, if you have some ideas to improve Hubski content discovery, let us know.
Question for you, MK - how hard would it be for you to build a plugin for me so that I, personally, can merge tags (as far as my experience is concerned?) For example, I'd love to follow #writing and #writebetterdammit but if I indicate to you that I consider them the same, does that get you closer to assembling a "metatag cloud" whereby like subjects can be shown with some overlap? Trying to think ahead and make things self-organize without curation and this is the off-the-cuff notion I just had. When in doubt, make the users do your job for you... there are more of them.
Quick suggestion on this front: Associate tags by frequency of them being together on the same post, and drop the 2-tag max. Let a post be tagged #science, #biology, #biochemistry, #homelab, #pcr. Then, when tags appear to be used more often together, above some certain threshold, list them as associated on the #tag page, similar to the similar posts feature that was removed. That way, if I think #writing and #writebetterdammit should be related, I can always just use the community tag for relevant posts myself. Additionally, you could have a suggested tags field to quickly add commonly associated tags, though this might bias the algorithm in the long run... Only trouble would be that it could be abused, letting me associate #murderer and #yourmother if I'm feeling malicious...
"How it might work" is kind of secondary to my thinking. "How you might mine easily and harmlessly for data" so that "how it might work" ends up being easy and durable strikes me as far more valuable than imposing order from above. But then, I can't code worth a shit.
Yeah, I'm just thinking in terms of making the curation worthwhile for the user. We already have community tags that need to be improved. I'm just wondering if there might be some nice synergistic way to make the curation process bring immediate returns for the people doing it. I had the thought that community tags could be rolled into this functionality, but I don't think that necessarily works. Community tags aren't necessarily used as 'similar' tags, but for editorialism or greater or less specificity.
To make it worthwhile to me, all you need to do is work it into tag discovery. Say I go to the page of all things that are titled #writing. It asks me if I want to follow #writing. It might also suggest #writebetterdammit, #writingsucks and www.readwriteweb.com. If I can click "similar" next to #writebetterdammit, then you have metadata suggesting that at least one user thinks #writebetterdammit is similar to #writing but #writingsucks and www.readwriteweb.com are not. If I also have a radio button that says "follow and group" next to all, that's another piece of metadata for you. Finally, on my page let me rename my "groups" so that "#writing#writebetterdammit" becomes "Writing" (or "kleinbl00's magical writing group"). That's more metadata for you. The trick is to make that grouping pleasant and fun. I'd go as far as to say handing out badges for every (20? 40? 100?) grouped tag would be altruistic; after all, those badges don't do you any good, they're the gift that keeps on giving. Spitballing.
That's not a functionality problem, that's a UI problem. Take it from an engineer - the two are NOT the same. Wanna see how MOTU's Digital Performer won Digital Audio Workstation of the year for 2011?
Looking forward to search functionality increasing! Could you do something about the hit or miss nature of the url checker? Specifically if there's any way to check, upon submission, for urls that link to the same article but are slightly different.
Thought of another thing I've been wondering about for a little while -- advanced search options along the lines of [posts:flagamuffin "word"] or [shares:mk "word"]? Maybe an option to search title text exclusively? EDIT: [domain:www.wsj.com "word"] etc