First the good news: Twine is a much more powerful take on social bookmarking than Delicious and its rivals. A Twine is a collection of links and other Web information, and you can create them on any topic you like and either make them public or keep them private. Other Twine members share their Twines, too, so you can browse through ones on an array of subjects, or peruse the Twines that a particular member has created.

Here, for instance, is one on the election:

And here’s one on the financial meltdown:

You add items to a Twine using a bookmarklet that sits in your browser’s button bar, giving you access to it whenever you’re on a page you want to share. One of the things that makes Twine more interesting than Delicious is that it has semantic understanding of some types of Web content. For example, when I added a page at Amazon about a DVD collection of Smothers Brothers TV shows, Twine understood that it was a DVD, that it was published by Time-Life, and that it was released on September 16th, 2008–and all that information showed up in the page it created on Twine for that link:

And when I added a page at YouTube, Twine knew it was video and automatically embedded the clip on that page into Twine:

So far so good. Pretty neat, actually.

Twine is bursting with features–and that turns out to be a problem as much as a plus, at least in the service’s current incarnation. Some features just aren’t very clear. Each Twine bookmark has a Collect button: I’m still not sure what it does, even after having pressed it. There’s also a Catch Up button; when I pressed it, I got a dialog box asking me if I was positive I wanted to do it. Said dialog didn’t explaining what it means to Catch Up, though.

When I visited Twine’s Help section for information on what these features do, I found that that help is provided mostly in the form of a FAQ that is itself a Twine. Which didn’t help with my questions.

Twine uses its semantic understanding of the bookmarks and other data it stores to try and connect the dots between various types of information. But these connections often seem tenuous. I did a search for Red Sox, then clicked to see “Related Tags,” and got this:

And then I did a search for myself (go ahead, mock me) and clicked on a link for “Related People.” Here’s what I got:

I’m not sure whether to be flattered; mostly, I’m confused.

Twine’s basic search feature was also a little iffy in my tests. Here are the first results it gave me when I searched for “cars”:

Twine does contain plenty of stuff about cars–it’s just that none of it came up at the top of my results.

I still think Twine is full of potential. I chatted with Nova Spivack, the founder of Twine’s creator, Radar Networks, yesterday, and he spent almost as much time talking about what the company plans to do with Twine in the future as he did discussing its current version. And I suspect that some features will work better when there are simply more Twines that contain more information.

Spivack told me that the Twine that launches today is version 1.0, not a beta, but I think it might be best to think of it not as a finished product but as a pretty rough draft. That said, it’s a rough draft that’s worth checking out.