Primal Fusion Primal Fusion / Items

First Impressions

Get Feed

Just got my alpha account.  My very introductory first impressions:

 

- interface is limiting, faceted browsing would help vs drop downs and display of content is put in too small a restricted space, better communication of 'location' in the information space is necessary

- potential for publishing collections of information very good (collect some things then 'make a web page'

- needs more data sources

- access to 'thoughts' content in primal fusion itself would be helpful, though right now there would be limited content anyhow

- hooking 'thoughts' up to delicious tags or other uri systems would be hot - already going in that direction as if you add content from flicker it pulls in the tags, though in a strange way - needs some getting used to

- this is not a Twine replacement but Google Reader is getting there :-)

 

Overall, very interesting, one to watch - a new take on search and publishing - though it runs the risk of being absorbed by other technologies (blog platforms, google) as it is more of a feature than a product.

 

-

Comments

  • Public Comments

    • 9 months ago


      My invite hasn't arrived yet. Can't wait!
      Primal Fusion
    • 9 months ago


      HEY!! HOW COME YOU RATE??? You're in the 5th row in these parts, but they let you in first? Sheesh.

      (can ya put in a good word or two for a few of your aquatic friends, and Francois, too?" ;-) ;-) ;-) nudge nudge, know what I mean?
      Primal Fusion
    • 9 months ago


      It arrived today! Unfortunatey no time to test it seriously till next week. Grrr...
      Primal Fusion
      • 9 months ago


        mine too--same problem! OH well, it's an alpha--which means it will be a totally different product next week, right?
        Primal Fusion
    • 9 months ago


      My invite showed up today!
      Primal Fusion
    • 9 months ago


      Take a look around their feedback form! Rich experience :)
      Primal Fusion
      • 9 months ago


        I do like their customer support and feedback--they are using GetSatisfaction, which seems to have it's act together. Oddly, Twine experimented with this service, but I don't think they are active with it anymore. The people at Primal Fusion however are very responsive, and actively utilizing GS' system. +1 for that at least.
        Primal Fusion
    • 9 months ago


      Agreed. I finally had a chance to try out my account this morning, and although the idea of having searchable inputs from the likes of Flikr, Yahoo, et.al. is appealing, I don't really see (yet) how this is much different than doing global searches on Google and documenting (linking) finds of interest. The web page feed idea may be beneficial at some point when they open up the data resources to other services besides the few they have, and this becomes an active feed based on related tags. Google reader does that already with custom feeds. And I'm concerned about the tag issue in general.

      They are only searching/filtering via tag sets that others have assembled on their particular content--web site, picture resource, news item, post, etc.--which makes PF's ability to recognize and catalog limited to the tags that have been created. It isn't really scanning content and cataloging based on semantics, it's just grabbing items that match a predetermined set of tags. Certainly not very revolutionary (yet). As Kurt says, it may be better to revisit in a while, after they've figured out what to do with what they've got.

      I also like your comment JonRod, regarding public domain content, in light of all the current debates about copyright, sharing, and what is or isn't public access, particularly if there is an intent to reassemble and republish items that have been located elsewhere. If it's for private consumption however (since they are an Anti-Social network) then that really doesn't matter much.
      Primal Fusion
    • 9 months ago


      Was able to do an initial look yesterday. Seems to be an outliner for tags based on content in wikipedia, yahoo, and flickr. User enters keywords for thoughts which seems to translate to tags, and outputs the links to html, doc or feed. The tags are saved in a "train of thought" and can be combined so that matching data can be checked off, saved as "remember selected content", to yield final set or "thought network". It is described as an anti-social network since it is not for collaboration. There is no import capability to work on previous output, nor an api though may be possible to do some scripting which would be temporary since this version is labeled as a prototype. Researchers probably need to add other search and semantic utilities for coverage. It might be possible eventually for a business or government perhaps to provide and fund additional data sources. On the other hand, it is good that it is capable of handling public-domain sources so that data is not eventually wholly owned by private commercial interests. Overall this is yet another datapoint in the external-brain marketing train. May not immediately displace twine, twitter, delicious, zemanta, evernote, mindmeister, 3d and the like. Thanks.
      Primal Fusion
    • 9 months ago


      Thanks, all, for the feedback. It's very helpful. We're pulling together a paper on the innerworkings of Primal, but until that's ready, I just wanted to surface a few points in response to some of the comments above:

      - the intent with the app is to allow consumers an easy way to create a semantic graph representing their unique thoughts (what we call a thought network). We're not trying to situate users within an existing graph but rather build one around them.

      - the technology synthesizes the graph by assembling (fusing) more elemental (primitive) concepts into complex concepts and relationships. Synthetic systems are very different than analytical/retrieval-based systems. The challenge is in providing new concepts and relationships that people will feel are coherent and proximate to the topic at-hand. We use multi-dimensional rules that guide the construction of the concepts and relationships (much like grammatical rules dictate how words form sentences). Using a small amount of data, synthetic approaches can create a massive amount of possible concepts and relationships of varying degrees of quality, and critically, can involve the consumers as active participants to it in real-time. So much of the work is directed to effectively reining in and controlling all those possibilities. (Note that at the moment, we're limiting the output to only a few dozen related concepts for any term.)

      - we need to throttle the amount of data in the system while we're measuring the performance. What you're seeing today is a fairly shallow dive on a number of axes, but hopefully it communicates the idea. We're looking forward to opening it up over the coming weeks. There are a number of different levers you can pull to affect changes in the resultant graphs, so it'll be fun to expose this more in the apps.

      - re: integrating with the external sources like yahoo, flickr, etc., we're using the users' thought networks to run a term analysis against the results returned from the sources. A relatively small network of a few thoughts will have a number of related terms (often much broader than the superficial tag cloud presented to the user, including concepts both intrinsic to any given complex concept and extrinsic in the related complex concepts). This provides a substantial lattice of terms that we can use to filter and categorize the result set. We only send a single query to the external source so as not to bog down those systems. The filtering agent itself is very lightweight and a key extensibility point in the app. The brains of it is in the users' thought networks.

      - we're using a tag cloud as a visualization for the graph underneath, with the font sizes corresponding to the relative number of nodes under each term. It's an opaque way to represent a graph and one that's easily mistaken for a flat tag-space, but we're hoping users will find it more familiar and easy to use.

      - re: anti-social, I think thought networks will be powerful tools for collaboration. My point was more that thought networks (and semantic networks in general) don't require social/human intermediaries to interoperate. Social networks are often tasked to knowledge exchange and I don't think the potential for semantic networking in this area is well understood within other communities.

      Again, thanks for the feedback and please send along any ideas or concerns you might have. The app itself is still very much in its infancy, so we completely understand if it's too early for many. But if you find this work interesting, stay tuned and we'll keep you appraised of our progress. I'll cross-reference this thread on our GS site (www.getsatisfaction.com/primalfusion) in case any of those folks want to chime in on this.
      Primal Fusion
    • 9 months ago


      So I got mine yesterday and started playing with it, definitely an interesting concept, nevertheless tag cloud very limited, I thought this could be used for research (now researching the concept of Ambient Intimacy for a coming paper) no available or relevant tags showed up, therefore at present I agree with the fish, the usage of previous web users of tag sets diminishes the ability to create new content. lacking sources (using only flickr & yahoo +wikipedia doesn't cut it).
      at present it appears that the meaning of 'Alpha' (see prototype) is accurate, not much can be done there.
      having said the above I can definitely see a huge potential as a utility (add-on or else), concur with Kurt here, it is more a feature than a product.
      overall I think the need for such a 'thought' following network is desirable and much in need, but it is not there yet.
      Johnrod summed it up perfectly:.."another datapoint in the external-brain marketing train"
      Will follow development with interest.
      Primal Fusion
      • 9 months ago


        Reading their marketing materials, they are well aware of the 'utility' function of this type of service, and by all appearances are looking forward to an early extension of their API for others to develop around. [LOUD APPLAUSE TO THAT]

        @Peter, I understand the desire and need to keep inputs limited until you are able to build out the system more robustly. I have found it difficult to really explore that as wildcat says, given the existing resource input zones. The few search sets I tried either returned nulls, or returned just themselves in the mix, so I saw nothing meaningful. (searched on Design as a general term and then began adding tag associations that came up with it.)

        The tag limitation is a difficult one to resolve--based on what we've seen here at Twine when bookmarking links and allowing this system to capture and catalog, Many websites and resources seem to apply their tags differently. Some are in meta-data, some are in the body of the page frame, some are just a list. So there is difficulty in simply recognizing the tags that the web builder themselves would associate with their content since the isn't really an apparent standard. (I don't know much about code and webpage design, so there's probably something I'm not aware of here...)

        I think what might be really helpful for any and all of your initial front of applicants, would be to build up a functional mock-up of a specific term set--something like what you showed on the web demo video, but one in which the user could step through the process, and possibly link off in different directions within the test set. That would at least provide more of a hands-on guided tour so to speak, and allow users to interact with what you've got in a somewhat controlled manner. This would also provide you good user feedback on interactive functionality and possibly direct the product in a more beneficial manner. Just a thought.

        Very impressed that you are monitoring our little corner of activity--thanks for be open and willing to listen!
        Primal Fusion
    Add a Comment
Report This

Twine is about discovering, collecting and sharing the content that interests you. Learn More

Join Twine

Stats

First Posted By

First Comment By

Who's Interested In This?

Forgot your password?