Using diigo's bookmark tool it hits me that if the system knows the tags ordinarily associated with a group are (by frequency of use even) you should be able to select a group by tag matching (so imagine the group drop down populating with the most likely choice and you just validate it, or if multiple groups match they are listed first (above the line) in alpha, with less likely starting again below the line in alpha order)

very nice, I especially like the reducing group list as the user selects tags that can simply be accepted if correct
the tag to group matching needs to be somewhat 'fuzzy' fit as I wouldn't want a group to disappear from the selection just because I added a novel tag.

Thank you Francois, excellent description.

I didn't follow the Common Tag saga, though from the outside it seems a standards effort by small players without the support of the major ones (delicious for example) which is pretty much doomed from the get go. I was impressed way back when when the founder of zigtag disclosed his cheap and dirty approach to semantics on tags with a wikipedia integration, a brilliant insight at the time, since much copied.

Zigtag suffers from a loss of its founder, it has not done anything of value since the funding parties pushed him out. It has also since lost its lead developer, because there is nothing interesting to do. The company has more or less disintegrated and the site continues on auto pilot. I had way too many issues with the tool not saving or sending and gave up as no one is at the wheel.

Ultimately I think re tag/group a group is just a tag with a few additional relations (member of, for one) and should actually be defined at run time rather than design time. Group is only one of the interesting combinations of relations that would be possible. Even what is referred to as 'tags' is not one thing. we have simple text, multi word text, defined text, tag hierarchies or rhizomes (meta tags if you will) and semantic structures all the way up to ontologies. So tag is not a simple thing even by itself. The simple named grouping function that is a basic text tag has many younger siblings.

In a world of real time feeds, repetition is an attractor

ping

repetition, reverb, you can set up 'remind me' or 'read later' with time options to have an item pop back into your 'real time' feed

ability to subscribe to a tag - perhaps a lower class form of group (bent's extensional vs intentional?) is automatically formed when >1 person sub'd to a tag

aside from the obvious feeding one group into another, the ability to specify two groups to be merged where the merge takes each post and tags it with the name of the group it was from

long sought after at twine, the ability to feed one group into another, and to create a group from a static saved search, or feed a dynamic saved search (rss feed of new results on that search) into a group

if posting to a group, multiple groups are possible in one posting, and clusters of groups can be defined, both as logical groupings and as equivalencies - those marked as equivalencies are mined by system for referrals and merges

I've discovered I must be able to inspect the current state of something I am bookmarking (who what where when etc) as I am bookmarking it

It would appear I've hit a hard limit with diigo of 50 groups created, to which help's reply has been unhelpful. While fish will jump with joy, this is a non-starter for me, my tool should not arbitrarily get in the way of me thinking. I am going to build my stuff elsewhere (likely amplify) (Eric if you are still paying attention, does Amplify have a limit on groups/topics or whatever you call this level of aggregation?) thanks all, will send postcards

that would assume everyone is interested in everything, until such time as users can filter using facets to a feed they wish to get, then absolutely I will do just that, in fact tags and groups are very closely related and you should be able to freely turn a tag into a group when someone else wants to socially contribute with you. lateedah. skipping down the lane with my pink balloon, not bothered by you at all - oh look a butterfly

the tool should not require me to stand on my head to think.. but that may be a temporary solution, time to experiment with multiple personality disorder, I fear I may enjoy it too much

zigtag "search, save, share"
ebay "buy,sell"

possibles:
share: recommend, propose, discuss, support, agree, disagree, express interest in, acknowledge interest in - consideration needs to be given to the intent of 'sharing', is it to get opinions, is it to collect highlights to combine and subordinate later (as a group, by yourself), is it to feed an interest of someone else you know (and you have no real interest in the thing)

collect, read later, save, remember

search: browse, get updates, measure buzz, locate friends in infosphere, collate, track, filter - we search for various reasons, what are they all?

I have no time to continue right now, but I believe this is a rich discussion

I do like hilighting to show, but would like to be seeing only my and my colleagues hilighting within the context I am navigating to that page from, likely with toggles for 'all comments' 'all highlighting' 'my highlighting only' - zigtag used a left side sidebar for contextual information (tags, who had saved etc) and on a wide screen with the normal widths of websites this was usually less annoying than the horizontal split diigo uses. I find the menu bar on diigo sufficient usually for non ambient context as I can always click the bookmark link and it will show me tags and where the url is shared already. It would be nice to fitz law that gesture like shooting the mouse pointer to left side of screen, or some gesture to pull it up, finding the little itty bitty bookmark target is annoying

friend lists,

should have public and private friend listings possible (you may have one category for the other party to see and another for your own internal organization, filtering, trust, attention etc settings - don`t want anyone getting upset now)

should be able to use these lists for:
messaging
filtering
assigning trust levels to
paying attention to
- presence alerts
- posting alerts
inviting
...?

twine had toggle, neither really works, consider what we are trying to do:

know what has occured since last here (very important!!!!)
understand how conversation developed (chronology)
understand structure of conversation (threaded)

the 'occured since last here' touches on synchronous vs asynch in that if you are constantly synchronous you cannot miss anything

IM is synchronous 2 party for the most part, but I still have multiple topics in one IM with one other person, need threading there too

also tied up with rich presence as well (understanding who is synch who is not, who posted synchronously / asynch'ly - following a conversation is complex, if you are 'there' you need to follow multiple threads and multiple participants, if you are not there, you may want to 'catch up' in an organic way or by scanning the structure of the conversation, you may be disinterested in certain threads and want to de-clutter by hiding them, you may be disproportionately interested in certain parties and want to somehow hilight them

screen real estate is an issue here as well - I may not be able to view the whole thread but I may be very interested in jumping into a hot spot in the conversation that is off page currently - following the interesting crowd at the cocktail party if you will. This is clearly not because you will miss something (unlike the cocktail party) but that you will miss the rich interaction and ability to influence the conversation and maximize leverage with valuable minds.

I have suggested that presence blips on the scroll bar would be useful, some change if active vs 'last at' also indicate attention this way - show where frank is looking at a thread right now, even if not typing - gwave has issues with simply showing activity so there may be technical hurdles here

brings in brin's concepts of social filtering, attention, keyword filtering etc

forking / splitting and Perspectives come into this as well

boy diigo is pissing me off crashing firefox, have switched to chrome - long posts cause issues and then I lose them (auto save people!!!) anyhow. /rant

The predominant model of systems to date has been to analyse your domain, model it, implement it in a rdbms then code access to it (forms on tables). Simplistic but largely true. RDF triplestores among other things allow 'stacking' of data which does not require schema in advance, only equivalency mappings (or standards). We need to decide what we consider foundational to the system (schema in advance) and what will be provided as structure after the fact - architecturally we need to allow both modes - and given extensibility, we must also provide reference structures to provide minimum functionality with a set of "structures after the fact' provided up front (yes I'm aware of the cognitive dissonance there)

if a post is made to a group that is a sub-group of another post is promoted based on settings (if setting is 'ask' then system will ask on posting whether to promote) this adds 'enable post promotion' to the group settings and I suppose 'disallow post promotion into this group' for parent groups to avoid hyperspam

canonical example, threaded conversations, any entry in a threaded conversation should be able to be simultaneously a comment in that thread and the head of its own thread (a topic) these two 'types' could have both different interfaces and different affordances (the comment is just text in context, with associated order time author metadata, the topic may have subscribers, rss feed etc - a party to the original threaded conversation may ONLY be interested in that comment and its sub threads, and may wish to elevate this comment to a topic for example and invite others to discuss it - see perspectives also

using system 'Concepts' and user/group/social/domain specific types, and or not operations, result set type specification