What flavor of Stone Soup is Twine

I have been playing with Twine for a little while now. Some aspects I really like but there are one or two things that puzzle me, and one or two that I am not sure about. I think the twines themselves have potential, I have slowly started building up one that I hope will be a good way to share some resources, particularly when Twine goes public. I am sure that some will think that the aspects that bother me are simply down to not knowing enough about how Twine works, or more likely because I am not using Twine enough yet. This "not using Twine enough" is the thing that bothers me most, the mantra used to answer criticisms of Twine is that "the more you use it, the more you get out of it". Well that seems reasonable but I also think that a cutting edge semantic web app. should do some things almost from the get go.

There is the story of the stone soup, it doesn't really matter if you don't know it and don't wish to read the Wikipedia entry, hopefully my point will still make sense. Is it that the flavor of the Twine soup is mainly derived from all the ingredients that are added to it by the users. A cunning but relatively simple starting framework could attract a critical mass of interaction and this mass then be used to develop a semantic application. Of course you need content to be added and categorized at some level, but once you have content why does it seemingly require a lot of activity to guess what kind of new content a user may be interested in.

My best example so far is a story on Mandarin immersion learning , here are the things that Twine could know.

1. I have created one twine so far, it is tagged with mandarin


2. I have posted a high percentage of items tagged with mandarin


3. I have read a number of other items tagged with mandarin and shared one or two to other twines


4. I have conducted a search for mandarin


5. I have pulled an RSS feed against the mandarin tag search


6. The mandarin tag is relatively rarely used on twine

Yet when a new item is bookmarked on Twine and has a mandarin tag, I am not alerted. Given all the information above and not withstanding any smart programming that may be monitoring my activity across twines or connections I cannot imagine not picking up on at least enough of the low hanging fruit above to alert me somehow on the item. In fact if I had been involved with the programming I would have a red face at this point because Twine is touted to be smart etc. There is a remote possibility that I missed something I suppose but there doesn't seem to be any mechanism to discover or tell me of the important fact, also the twine where the article appeared wouldn't normally hold my interest enough to make it worthwhile subscribing to it, so the standard Twinerian approach of subscribing to a huge number of twines wouldn't have helped either

There are a number of other things I am looking into that worry me. There is a "bug" where if you post a bookmark that is already in Twine you get duplicated items, even if you post to the same Twine. Putting on my programmers hat this is more than a bug, unless I am mistaken it seems that the url of a bookmark is its unique identifier, this should be enforced at the back-end and unit-tested, not allowed to bubble through to the user interface. Other opportunities to cut down on information clutter have been missed for example my Twine digest email informs me of items I have posted myself.

What I have at the moment to discover things I may be interested in a marvelous hack-up of RSS feeds, pumped and tweaked and combined through Google Reader, Netvibes and now increasing Yahoo pipes. This chaotic system has the amazing ability to (assuming I apply a little regular maintenance and tweaking) to supply me with a pretty good overall picture of things I may be interested in. Now I am increasingly adding Yahoo pipes to the mix I can see another big step-up in the ability of my ad-hoc distributed tangle to serve this purpose. In fact if you didn't know how it worked you could even be mistaken for thinking the system I am using has some kind of intelligence (Yahoo pipes probably being the major potential contributor here).

Twine is a useful social repository to drop information into, they have convinced me of that, also it is supposed to automatically tag things (well sometimes), why haven't the simple quick wins to link people with interesting information been picked up on, especially when I can get so much mileage from tweaking and manipulating RSS feeds from disparate sources all over the Internet? The recommendations that are supposed to kick in once Twine has learned enough about me haven't show me anything yet. My experiment is going to be to see if I can get more interesting and relevant content to read from Twine simply by the manipulation of its many RSS feeds (and become a Yahoo pipes power-user at the same time ;)). At the moment I am betting on my system to win, to be fair I am interacting with Twine in a manner that should help their recommendation system, but not as fast as some (anyway a system that only works for power-users is not a mainstream system at all).

As it happens a helpful Twine user called Twain (thank you twain) told me about the Mandarin post in a reply to a comment on an unrelated thread, which was nice, even though I was already aware of it. This however does not count as Twine informing me of it.

Ultimately the stone soup story can be interpreted in a number of different ways, with both a positive and negative slant, lets hope the Twine story has a happy ending.