Tinderbox User-to-User Forum (for formal tech support please email: info@eastgate.com)
http://www.eastgate.com/Tinderbox/forum//YaBB.cgi
Tinderbox Users >> Off The Wall: Feature Ideas >> A meta-map?
http://www.eastgate.com/Tinderbox/forum//YaBB.cgi?num=1245399425

Message started by Mark Anderson on Jun 19th, 2009, 4:17am

Title: A meta-map?
Post by Mark Anderson on Jun 19th, 2009, 4:17am

I've been pondering for a while about the map-vs.-outline struggle.  Maps are great for discovery without forcing hierarchy. Most other tasks, such as review, textual structure and export normally need an outline, breaking the purview of the map as it shows only one level (well 2 if you include container viewports). Indeed one child level of a parent note - i.e. only siblings with the same parent.

I'm positing the idea of a 'meta-map' that shows every note in the TBX - effectively the outline flattened onto a single map. The idea is to allow a global view of the TBX without upsetting the outline - or in early stages without needing to worrying about the outline.

Practical issues:
  • Scale. It might be that TB might have to cap the numbers of notes allowed in this system to avoid overload.
  • Exclusions. Do we want prototypes, templates, utility agent and other behind-the-scenes stuff on deck? Probably not, so maybe we'd need a note level exclusion flag and possibly the option to exclude all templates and/or prototypes via a set of preferences.
  • Initial placement. We can't just use the normal Xpos/Ypos as everything would be in a heap in the middle of the map. Perhaps separate co-ords for the meta-map. If drawing the map for an existing TBX draw down the outline in a left->right top->bottom manner?
  • Where do new notes get added if created outside the meta-map?
  • Adornments from normal maps - do we want to see these?
  • Show container (viewports)? The problem here is that if the meta-map is everything flattened onto one map then seeing child maps that are actually different is potentially confusing both for the map and the application('s design).
  • Agents. If viewports aren't shown, how to deal with agents? Perhaps in the meta-map they'd function like smart adornments (with a visualisation difference to differentiate them from actual adornments).
  • Querying the meta-map. Using hierarchical queries (inside(), etc.) in a meta-map won't work as it's flat plus we need to consider existing map vs. the meta-map. An isCloseTo(item,mapUnits,[mapName]) query might be useful for the meta-map; the 'mapName' optional argument being used if wishing to specify a map other than the meta-map. However, this idea might be computationally expensive, so slow, so not a good thing.
  • Container plots/tables. We lose these in this view but for obvious hierarchical reasons.
  • Feedback to Outline. Use adornments to set $Container. The meta-map location is unaffected but the note's outline location is changed. In a context like a book design, chapters 1 through N can all be seen together on one big map, while the outline can begin to capture the output structure as it becomes available.


My term meta-map is a poor one but it's just a placeholder name. In the spirit of Tinderbox I don't want premature commitment to a name that might drive development of the idea.

There are some necessary constraints vs. existing maps for the reasons laid out above (and those others may yet add), so at best there's a bit of learning for for users - especially long-term users. But if possible it might add another useful analytical layer and break out of the current issue of starting in a map then losing sight things the moment they are assigned some hierarchy that moves them off the current container's map.

I'm not even sure if this can practically) be done but I do wonder if it might address some of the flat-vs.-hierarchy issues I see fellow users having. I don't have a strong personal use as (outside aTbRef) my TBXs are normally small and light - 'use & throw' - for early stage discovery. So, my experience of the above issues is thus somewhat second-hand but it does help one to take a detached view of things.

What say you?

Title: Re: A meta-map?
Post by Charles Turner on Jun 19th, 2009, 6:48am

Hi Mark A.-

I'm responding to this after a quick first read and not yet through my morning cup of tea. But I wonder if Tbox couldn't have an attribute that would assign a note to a map view. That way the control over a note's presence in a map view would behave like much else in Tbox: you could "QuickStamp" to create a view of your notes, or write an agent that would select a set for display in map view.

This would require a change to the ways views currently display, or less successfully, perhaps a distinction between flat and outline maps, but an attribute-based approached would resolve most of the issues you pose above.

I think. ;-)

Best, Charles

Title: Re: A meta-map?
Post by Mark Anderson on Jun 19th, 2009, 8:53am

Interestingly, that's effectively extending my concept of having more than one 'global' map. Maps at present are of the contents of a single container (or just the root). Optional display of items in that context is probably more complexity than gain, whereas on a global scope it's probably as distinct need. To have more than one map is conceptually possible but I'm wondering at what point it all gets too complex and/or too slow. Currently TB is the right side of app needs vs. available horsepower on a desktop OS but as complexity gets added (or rather much additional back-of-house calculation) then performance may suffer unduly.

One benefit of the global map that I omitted to mention is that source/sink of all internal notes will be visible on the map.

Title: Re: A meta-map?
Post by Charles Turner on Jun 19th, 2009, 11:15am

Or you could think of a map view as a container itself, and attribute notes to its view. Agents would work on view attributes as well as note attributes.

As far as efficiency is concerned, right now you can muck Tbox, or any other app, with a rat's nest of interrelationships. Read the DTPO forum, for example, where users have out-of-control numbers of "Replicants" and just want to get back to their original data. You can't save people from themselves. ;-)

Tbox will pick up a lot of speed when it doesn't have to deal with the Rosetta context switch anymore...

Happy dreaming! Charles

Title: Re: A meta-map?
Post by Mark Anderson on Jun 19th, 2009, 12:46pm


Quote:
Or you could think of a map view as a container itself

Actually, that's easily done now.  Make a note and put aliases of other notes into it. What I'm thinking of is a map that - in simplest terms - the TBX outline display in map form. In other words as if you'd taken your outline and promoted all notes to the root level.  I say 'as if' because the whole idea is this exists as an alternate view so the real outline and normal maps, etc., are preserved.

Writing is a popular topic so lets assume each chapter of your book or thesis is a container.  This means notes in Chapter 1 can't be viewed in the same map as Chapter 2 - something it seems people want to do.  My idea (if technically feasible) would allow this. We'd even be able to see link lines from notes in Chapter 1 going to their target in Chapters 2/3/4/etc. I hope that makes it a bit clearer...

Title: Re: A meta-map?
Post by Charles Turner on Jun 19th, 2009, 3:00pm

Right. Silly me. I guess the tea wore off, and my IQ dropped about 40 points...

C

Title: Re: A meta-map?
Post by Charles Turner on Jun 25th, 2009, 11:38am

Hi Mark A.-

I re-read your post and have thought about it some. Here's a little take:

Right now, the Map View provides a different graphic rendition of the Outline view, or perhaps vice versa. A note is placed in a container: a single location. Aliases might be considered the paradigmatic manipulation of note organization in this context. It represents the possibility to have a note be in more than one place, or everywhere, if for some reason you have that need. So, singularity moving toward ubiquity.

Your post considers that a property of the Map view be that it's a flat view into the entire Tinderbox notespace. A perfectly reasonable idea. I won't consider whether this should be an additional view in Tbox, I'll simply assume it would replace the current Map view.

This flat view represents a different way of modeling data in Tinderbox. We might ask then, what is the paradigmatic manipulation of notes with a view of this type? I would suggest that it's different than aliasing, essentially moving in the opposite direction of hiding detail, rather than replicating detail. In view of this type, say in Omnigraffle (or the wonderful, but now long-gone MetaDesign) you have the ability to aggregate and reduce detail (subgraph in Graffle-speak). Essentially, the ability to put things into containers to reduce surface complexity. So, ubiquity moving towards singularity.

So how would this be expressed in terms of Agent behavior in Tinderbox? My guess would be that Map Agents would have to behave differently than Outline Agents. They'd move notes into containers instead of creating aliases.

My guess is this could be accomplished practically by having the current path for a note become two paths, and move from a "general" property of the note, to a (separate) property of both the Outline and Map view. Then note organization, the data model, could be different for each of the two views. In Outline view, agent queries would create aliases, in Map view agent queries would move notes.

Finally, I think also you'd want to consider what are the paradigmatic mouse/manual manipulations of notes for each view. Currently, Outline view supports dragging notes around in the hierarchy. I think a Map view should allow one to lasso/select a bunch of notes and create a container/sub-map. Map view should also support the dragging of notes into and out of containers.

So that's my thought. Probably stuff I haven't considered. What say you?

Best, Charles

Title: Re: A meta-map?
Post by Mark Anderson on Jun 25th, 2009, 1:38pm

Interesting. although the meta-map is a much simpler proposal it could be a step on the road to the more complex concept you describe. I'm deliberately proposing something more simple as even in just flattening the outline there's still plenty of challenge in deciding where things could go - in advance of any user manipulation giving things a specific Xpos/Ypos. I think there's a danger in getting too ornate with the map metaphor lest the UI become a work in itself and we lose sight of the task of analysis.
Still, my view is perhaps a bit utilitarian as I'm not from a hypertext/academic background and some of the more formal terminology goes over my head - so I'm sorry if my analysis lacks insight!

Title: Re: A meta-map?
Post by Charles Turner on Jun 25th, 2009, 1:54pm

I'm hardly a hypertext/academic type as you can see from my previous lack of insight in this thread. :-)

If all you want is a current Map view with "everything" in it, couldn't you do that now with an agent that made aliases of everything? One of your containers would be an everything container...

Thanks for the dialogue, Charles

Title: Re: A meta-map?
Post by Mark Anderson on Jun 25th, 2009, 3:33pm

Yes you could, though all the contents would be (a) be aliases and (b) not be arranged in any order close o=to that of the outline.  Actually, sorting the agent on OutlineOrder before turning it off would at least give some semblance of current organisation.  However, without re-starting my original post, I'm suggesting a bit more than that.



Title: Re: A meta-map?
Post by Charles Turner on Jun 25th, 2009, 4:50pm

No need. I haven't stared at what aliases do with Map View coordinates, so I'll put that on my list.

While we're on the subject, have you seen this example of CSS?

http://nedbatchelder.com/blog/200805/css_homer_animated.html
Best, Charles

Title: Re: A meta-map?
Post by Mark Anderson on Jun 25th, 2009, 5:16pm

Re the link - wow, I've never got that bored at work! Pretty amazing stuff considered how it's done.

Re Agent maps and moving things, you need to turn an agent 'off' (see here) to be able to move items but wait until v4.7.0 (imminent) before trying it as it likely won't work in v4.6.2. - so leave off trying this for a day or so. Then, once an agent is 'off' you should (will!) be able to move aliases around like on a normal map; turning the agent back on will re-order your map - so bear that in mind. Note too that aliases of container notes show as notes (i.e. no sub-map viewport).

Title: Re: A meta-map?
Post by Paul Walters on Jul 29th, 2009, 5:53pm

I think of maps as they exist today as layered - each layer corresponds to an OutlineDepth, as does each level in Outline View.  I compose my map layers with adornments, note layouts, aliases, etc., to suit the purpose of that layer.  My own take on the thread, above, is that I'd like to see a quick way to toggle through layers.  For example, at the bottom of a map, between the selector for changing zoom level and the selector for performing a cleanup, there might be a "layer selector".  So, if a document had notes at OutlineDepth 1, 2, 3, 4, & 5, then when we clicked on the "layer selector" we would see 1, 2, 3, 4, & 5 and clicking on any one of them would display the map for that depth.



Title: Re: A meta-map?
Post by Mark Anderson on Jul 30th, 2009, 6:35am

But, what is there is more than one container on the present map? When you go down a level which (container's contents') map will be shown? Or are you proposing to all level X notes on one (meta-)map?

Title: Re: A meta-map?
Post by Paul Walters on Jul 30th, 2009, 6:56pm

On second thought, though it seems there is a "layers" metaphor implicit in the Tinderbox outline order that would be interesting to exploit in some sort of map browser, as Mark A pointed out what I proposed is too confusing to be useful.  

Title: Re: A meta-map?
Post by Rob Forsyth on Sep 10th, 2009, 3:29am

I've enjoyed this exchange, and pondered similar issues elsewhere on this forum before. I recently discovered Prezi (http://prezi.com) and enjoyed their approach to what in some ways is a similar problem to structuring information for presentation/overview. I don't know how familiar people are with this but would encourage people to visit.

It seems to me you could in principle use their approach of extreme variation in font sizing and "hyper-zooming" to create the illusion of exploring a nested structure although in fact everything's on one "level". Perhaps you could use this approach to show the meta-map? I suspect however that the implementation (particularly such wide variation in font size) within a standard Mac app may be non-trivial!

Rob

Title: Re: A meta-map?
Post by John Quanrud on Sep 19th, 2009, 10:13am

What you describe, Mark, as a 'meta-map' is feature I have wanted from the very first time I started working with Tinderbox. I prefer working visually and so have created many maps for my research, but always struggle when it comes time to covert my thoughts into an article. The ability to visualise all the layers of an outline in map view would be a most useful addition to my Tinderbox experience.

Title: Re: A meta-map?
Post by Amafortas on Sep 27th, 2009, 10:21pm

Yup. I can see how this sort of view would help with organizing information for writing. Is this sort of thing possible in Tinderbox's programming language or is it stretching the parameters in ways that are simply unfeasible?

Have a good one,

~A

Title: Re: A meta-map?
Post by Amafortas on Nov 27th, 2009, 12:42pm

Hello all,

I was wondering if there is any more work done on this topic?

Thanks,

~A

Title: Re: A meta-map?
Post by Mark Anderson on Nov 27th, 2009, 1:36pm

Today? It stretches what might be done, but look at this page, as the concept of separator=adornment offers some scope for mimicking map areas in an outline.

Title: Re: A meta-map?
Post by greg ibendahl on Jan 21st, 2010, 1:03am

What about a meta-map view with "panes" where each pane would represent a different level of the outline. These panes would be adjustable for width and also with different zoom levels. You'd still have all the issues that Mark mentioned but it might help with visualization. I'd also like to see a dotted line link (or something) that would connect parents to children in such a view.

Title: Re: A meta-map?
Post by Mark Anderson on Jan 21st, 2010, 2:27am

Greg, how do you see that scaling past a few containers. Once the doc grows you'll need lots of panes - my current TBX has >150 containers (i.e. maps).

Title: Re: A meta-map?
Post by greg ibendahl on Jan 21st, 2010, 10:14am

Mark, A current map view shows the children of a single parent. What I'm proposing is that a "pane" show all notes at the same outline level. As an analogy, My parents have 3 sons and me and one of my brothers have 3 kids each (my other brother has no kids). That makes 6 grandkids for my parents. If I was to create a map view of my parent's grandkids, there would have to be one map for my kids and another for my brother's kids. With the pane view, the 3rd pane in the map would have all the grandkids.

While your current file may have lots of map views, how may levels deep is it? I'm guessing that may folks only have files that go 3 or 4 levels deep. And making the panes' width adjustable should help. The first pane could probably be fairly narrow in most cases. However, you do bring up an issue that with a lot of levels in your outline, you could end up with a lot of skinny panes. Perhaps you could allow for a second row of panes but this makes reading the map somewhat more difficult.

Title: Re: A meta-map?
Post by Mark Anderson on Jan 21st, 2010, 11:15am

The doc is about 2000+ notes/agents/aliases/etc going down as far as level 9. OK, in my case it's primarily an outline, but I've learned never to assume.

So the idea might most easily be though of as like Finder column view except each column is a site-wide slice as that outline level except the selection at any given level doesn't alter the pane to it's right (as Finder would!).

OK so if column 3 (outline level 3) has 10 containers at that level, then column/pane 3 has to have 10 maps drawn into it. Except those are 10 discreet maps then you can't just merge/overlap them - without affecting the outline - each maps is a single container.  So column 3 now needs 10 rows, one per column (or more/less as the number of containers. Interesting idea.

Title: Re: A meta-map?
Post by greg ibendahl on Jan 21st, 2010, 12:12pm

Mark, Actually what you are proposing (i.e., a pane with rows for each parent) is what I had first considered. I'm now thinking more that each pane would be basically a new type of map view that would require an alternative attribute for X and Y position in the meta-map pane. Consider my previous analogy with my parents grandkids: In the 3rd pane all 6 grandkids would be listed. However, I might want to group them within this pane so that all the boys are together and all the girls are together (which would require mixing my children with my brother's children).

I wonder though if you could use the X and Y position from the regular map view to assign an initial position in the meta-map with some way to offset the vertical position relative to where the parent note is in the outline. This would help when opening a new meta-map view so that all the notes were not grouped together in the middle.

Title: Re: A meta-map?
Post by Mark Anderson on Jan 21st, 2010, 12:29pm

I see, so in effect everything has two map positions - one on the normal per-container map and one on the meta map for normal map's overall outline level. The two positions might be the same but rarely will be, I suspect as all maps use the same basic co-ords and mashing them together would require some resetting of per-container map positions. I suspect that would be difficult to program, but I'm not a software engineer. A hard part of the appreciation is knowing, without actually building it, whether such a complicated feature will actually live up the the concept.

It's worth noting this is still a constraining context in that it only suits documents where note that need to be on the same meta map are on the same outline level.  If you have stuff on level 2, spatially  interacting with level 3 content is out as they're in different frames.

Your comment about most active maps not being too deep did make me think that a more effectively push to 'map flattening' would be to allow a 2 level outline/map transpose where on going from outline to map the containers become adornments. OK fewer levels supported but probably a more achievable solution. not that I've strong feeling on this - I was originally picking up on users comments about the problem of starting out with Map and then maturing to Outline when the need for export/output beckoned.

Title: Re: A meta-map?
Post by Brian Gregory on Feb 23rd, 2010, 12:02pm

Admin has moved this cross-post to a new thread.

Title: Re: A meta-map?
Post by Greg Korgeski on Feb 26th, 2010, 11:16pm

It seems that rather than "panes," a more three-dimensional view which allowed one to expand thoughts to view their links or link-networks, would be helpful.  Panes seem like thin slices, to use a biology lab or CAT-scan analogy -- adorn all you will, they are still two-dimensional, static views, whereas something more like a three-dimensional view of thought-networks (or thought "galaxies") would be nicer.  I like PersonalBrain for that (from thebrain.com); I can build networks of linked thoughts and then expand either entire documents or more often, selectively expand certain thoughts to see their link networks; it allows a dynamic drag and view kind of function in which some thought clusters trail off into "space" and seem further away, and others are closer.  It allows for a visual perspective of near-and far clusters of relationships which don't require one to click on containers and so lose higher level views in order to see lower level views -- you can effectively see as many levels as you want in a three dimensional space.

PersonalBrain lacks some of the potential functionality of Tinderbox documents, such as agents; but it provides a much more flexible and aesthetically exciting visual tool for visualizing complex idea networks.

Frankly I keep vacillating between relying on Tinderbox and PersonalBrain for exactly this reason.  It would be nice to have a way to have the three-dimensional map complexity of PB (along with its less daunting "learning curve," which is just another way of saying that only a few people will ever master your software)... but with the potential power of Tinderbox as well.  

If you look at how apps develop over time, either Tinderbox will eventually have to go that way, or someone is going to write its replacement which will combine these features into a next gen of software.   I can't imagine slogging along without these changes in five or ten years, esp given the amazing potential of touch-screen user interfaces which just cry out for three-dimensional mapping applications.

Greg

Title: Re: A meta-map?
Post by Mark Bernstein on Feb 27th, 2010, 9:19am

Greg:

Personal Brain's map isn't actually three-dimensional.  It's a hyperbolic browser.  We've been following that technology with interest since its original appearance in the research literature.  It's not without problems, and it appears to be encumbered by *two* separate and mutually contradictory patents.

The history of real three dimensional hypertext browsers since Lai and Manber's _Flying Through Hypertext_ (Hypertext 91 -- I *hope* I've got the right paper, but I'm in Florida and can't easily check the reference).  To their very great credit, they found tjhat people had trouble actually using a 3D representation, although they liked it.  

There have been other interesting examples of hypertext browsers using three dimensions or non-Euclidian spaces.  Marshall and Shipman won a Nelson award for a rubber-sheet spatial hypertext system with a fine paper, but neither retained the feature in the later work. Andreas Dieberger's _Information City_ (at IBM) was very cool, and should probably be followed up again.  Adding a third dimension to hierarchical browsers like Tinderbox's Chart and Outline can be intriguing too, or at least attractive: see for example, Durand and Kahn's MAPA.

A problem to keep in mind, though, is that while people are fairly good at looking at 3-dimensional representations, they are not generally adept at navigating or transforming them.  Ask any fighter-plane instructor!  (That's not a rhetorical figure; I'm pretty sure Lai and Manber did just that)

Title: Re: A meta-map?
Post by garyturner on Mar 22nd, 2011, 12:54pm

Greg, I have to agree with you 100%

Tinderbox User-to-User Forum (for formal tech support please email: info@eastgate.com) » Powered by YaBB 2.2.1!
YaBB © 2000-2008. All Rights Reserved.