Closest Tools to a Perfect computer interface/organizer/thoughtmap
Jörg Richter
A California-based DeepaMehta aficionado compiled an extensive report about related tools. He evaluated each one and tells the pros and cons. I think DeepaMehta can learn a lot from it, so the report is posted here for discussion.
I REALLY like DeepaMehta’s vision of a full interface where everything resides, and you can visually link any kind of data together!
Closest Tools
Thursday, February 10, 2011
8:24 PM
TheBrain
More thumbnails:
What it gets Right:
1. Fast and Simple
2. Fast search for connecting distant nodes
3. Associative and hierarchical
4. Drag and drop files to it!
5. Animated so you don't loose context
6. Mouse over causes links to get bigger
7. Mouse over lets you see notes under each node
8. Double click on background to make full screen node link view, hides properties, and notes
9. Color wheel icon is cool:
10. Linking icons are neat. Top dot is for creating parent node, bottom is for child, left is for jump
i. 
11. Collaborative (not tested)
12. Database backend (don't know details) (does it store everything in a big file?)
13. Has a fast java web client (and ajax that looks less animated)
14. Physics rules are neat
15. Drag files into the attachments of a node
What it gets Wrong:
1. ! The nodes and links move automatically into !!! Alphabetical order !!!. I think this hinders spatial recognition. So you remember the relationships between nodes, and if they keep moving every time you change a link or node, or flip back, then it's in a different spot, and you have to read all the nodes again. It's like if everything in your room got randomly re-organized every time you looked for an object.
2. Can only see nodes 2 links away, no more. Further than that they are completely invisible
i. Can't see further than a few nodes down. Better to have a default level of viewability, and then have immediate things emphasized and de-emphasized
3. Can't customize the positions of nodes and links easily, have to change modes
4. Can't change size or priority of nodes. All child nodes are the same size in relation to each other
5. Stores everything in a big file?
Biggest things I'd like from Personal Brain
1. Multiple windows, so I could connect distant nodes in window 1 to different nodes I'm exploring in window 2.
2. Stay in extended view mode, modify it so that:
1. Nodes don't overlap each other
2. Default View of Nodes: based on distance away from selected node. So if I select a node, the nodes directly connected are bigger, secondary connections are smaller, tertiary are smaller still, and maybe 4 nodes away is nearly invisible, but still have an idea of how many are there.
3. Size of nodes can also be controlled by user
4. Allow boolean searches by selecting multiple nodes, and give weights to the overlapping nodes
3. Give me more settings control. Let me choose buttons and mouse clicks like VLC media player does.
1. Biggest things I'd like to be able to do:
- Mouse over and wheel up or wheel down to change the size of a node in relation to the selected node to represent it's importance or emphasis
- One click node creation
- Ctrl wheel up to change the node distance visible. Wheel up to see further away nodes, wheel down to see fewer
4. Select multiple items and have it create a container that I can link to other nodes, or move them around. The whole selection box is really fiddly, and requires too much right clicking and menu navigating.
5. Rather than having menus, or right-click-context menus, stick with nodes and links. Allow users to create custom node link menus for themselves populated with the settings they use most often, and in the locations they expect
DeepaMehta:
What it gets Right:
Great white paper. Wonderful idea
They give a lot of thought to the back end
Open source
They see it as a desktop navigator, all applications would be modules in it, great idea
Looks easy to make into an ELN
What it gets Wrong:
Not done yet, very simple demo
No zoom
Can't specify strength or thickness of links
Can't change size of nodes
No auto size changing of nodes, so that more distant nodes get smaller.
Hire the guys who did deepa mehta
Asterisq:
http://www.optimice.com.au/projectinterdependencies.php
What it gets right:
1. Clicked, selected node is automatically centered!
2. Draggable
3. Different thickness and link weighting
4. One click switch between "1 node away" view and "all nodes" view, would be better if you could scroll with modifier key to increase or decrease visibility of nodes that were different distances
5. Has transparencies
6. Mouse over makes links bold, emphasizes nodes connected
7. Physics rules are neat, tries to not have nodes overlap others
8. Fast for small number of nodes (don't like how things keep moving, and the graph get's populated slowly)
What it gets wrong:
1. Things loose their relative positions. When I click on node 17, node 23 is above it. If I browse through the graph and come back to node 17, node 23 is now
2. Things don't stay where you drag them
3. Slower when hundreds of nodes are showing
i. Has to rebuild the map every time
4. Looks more like a visualizer, so not editable in the browser
5. Flash based, means harder to put in ipad, iphone
6. Zoom in and out requires moving a slider, wheel doesn't do anything
ThinkMap:
http://thinkmap.com/visualthesaurus.jsp
What it gets Right:
1. Nice physics engines, nodes don't overlap (much)
2. Node of interest is centered
3. Has an SDK
4. Works in browser
5. Can spread nodes apart, lengthen links
6. Make nodes bigger
7. Physics rules are neat
What it gets Wrong:
1. Can't edit it at all right now, just a visualizer of synonyms
2. Can move things around, but the spring back to the same distance away
Cytoscape:
What it gets Right:
Lots of plugins
Powerful auto sorting tools
What it gets Wrong:
Incredibly hard to use
VUE
What it gets Right:
Open source
Zoom in and out
Drag around
Presenter mode, where you can walk through a specific plotted path
What it gets Wrong:
Gets slower when it gets big
Mouse and keyboard actions not customizable
Limited color set
Not dynamic, no centering
Not easy to resize nodes
MindMapPaper:
What it gets Right:
1. Can change colors and size of nodes
2. Mouse over shows you links to all of that nodes connected nodes
What it gets Wrong:
1. Not animated or dynamic, except mousing over nodes
2. Color and properties hard to change
3. Buggy
4. Really hard to change settings (done with text file)
Cayra (defunct russian software)
What it gets Right:
1. Good use of color to present meaning for large maps
2. Great coloring so nodes would stay the same color and all nodes made off of them would inherit the color from their parent
What it gets Wrong:
1. Slow, Got really slow after 25 nodes
2. Defunct mapping software
3.
Pasted from <http://i.d.com.com/i/dl/media/dlimage/17/58/09/175809_large.jpeg>
OmniPlan
· What it gets Right:
o Colors, clear which portions of timelines belong to which group
· What it gets Wrong:
o Hard to use, not easy hotkeys.
o Lots of places where you have to browse menus with mouse to accomplish something
o Terrible interface, everything has to be done with the mouse, and through a couple layers of menus
o No zoom in zoom out ability
What it gets Right:
Very good execution, zooming in and out, and one left click dragging.
Very fast zoom in and out
What it gets Wrong:
Mindmap rather than a non-hierarchical network diagram
Text on lines is dumb though:
Created with Microsoft OneNote 2010
One place for all your notes and information
x28
we can thank him?
What HE got right:
- that too much dynamics "hinders spatial recognition"
- DeepaMehta is great but not done
What he got wrong:
- Moving between overview and detail needs no zooming. DeepaMehta
facilitates this just by flashing your own eyes from left pane to
right pane (provided that it will be fast enough again so that you
need not switch your mind into "waiting for system's response" mode
for added ~250 ms). Traditional zooming, by contrast, hinders spatial
recognition (see above).
Matthias
mla
Here's a better copy of my notes on various diagramming/thinking
tools, which has pictures of the various tools as well:
http://dl.dropbox.com/u/5852846/Closest%20Tools.pdf
I've been planning and sketching ideas on the "best possible graphical
interface" for a few years as a hobby, and coming across DeepaMehta
(or at least the white papers) was especially fascinating because it's
extremely similar!
About the zooming, I was thinking something analogous to google maps.
From far out you can understand the structural landscape of your data,
even though you can't see the details or individual roads. But you
don't really loose context from the zoom because you know that when
you zoom in to Berlin, Hannover is still to the west and Dresden is to
the south. But you wouldn't try to understand the shape of Europe by
staying at one zoom level. There are a lot of ways to represent
different views of the graph besides making some things smaller and
others larger. But if you have thousands or millions of digital
objects, you have to hide most of them from view somehow.
How is it currently planned in DeepaMehta? Keeping everything on a
massive canvas and screen drag? Minimization?
Really excited about DeepaMehta! I'll continue trying to learn more!
Cheers,
Michael
Jörg Richter
Hi Michael,
Thank you very much for the "Closest Tools" PDF! This condensed report is incredibly worthy for every think tool designer!
About the central problem of displaying and working with large amounts of data objects: DM's topicmaps/canvas concept comes from the insight that, yes, there can be millions of data objects but not everyone is equally important for the user's current work situation. So, in the first step DM strictly separates the data (the "corporate memory") from its visualization (the topicmap/canvas). In the second step DM lets it up the user to decide what to visualize. Crucial here is DM's concept of a "working situation".
In DM a topicmap/canvas ...
... does *not* respresent a self-contained unit, like an image file or a geographical map
... but a particular *working sitiuation* -- the "here and now" of an user or a group. It's up to the user to decide what data is relevant in that particular working situation. She decides what part of the corporate memory to reveal in the current topicmap/canvas. E.g. if the working situation is "Telephone conference with Customer X on date Y", the user would create a topicmap/canvas with that very name and find himself inside a blank topicmap/canvas. Then he reveals the conference participants and create agenda topics. While the conference she take notes and create relationships.
This concept of a working situation has several advantages:
- both, revealed and newly created content is automatically assigned to a particular working situation. Later on -- when the topicmap is already closed -- you can easily e.g. for a certain agenda topic recall (visually!) the entire situation in which it was discussed (by revealing the agenda topic and then following its topicmap(s) relationship(s))
- if the topicmap gets too cluttered you can hide data objects (topics) that are no longer relevant. Hiding means: remove from that particular map but keep in corporate memory.
- if a particular rendition is no longer needed the user can delete the respective topicmap. All the data objects and relationships are still kept in the corporate memory, usable in later working situations.
So, the corporate memory represents a passive global storage, and the topicmap/canvas the active working situation.
Everything said so far focuses on DM's concepts of working situations (the most innovative aspect of DM from my point of view) and its benefit for cope with large amounts of data. However, in working praxis, the tool must not only support the "here and now" but also represent history/time. That is to allow the user to create versioned snapshots of topicmaps, e.g. when it comes to trackable deliverables that are handed out to customers. Furthermore, there might be objects that are relevant only *inside* a particular topicmap/canvas (and not appropriate for being globally accessible). These aspects are not yet conceptualized/developed so far.
From my point of view zooming is one aid when it comes to providing an overview for a large 2-dimensional information rendering. Zooming is not per-se a solution for the problem of displaying and working with large amounts of data. I see the knowledge worker not as a cartographer who deals with 2-dimensional renderings of a 3-dimensional world, but as a navigator through and a creator of an n-dimensional information space who uses the 2-dimensional computer display as a visual thinking aid that represents her's constantly changing state of mind.
However, besides DM's concept of a working situation, zooming might be implemented as an additional visual aid.
Hopefully, I did not completely miss your point here.
Welcome to this group!
Cheers,
Jörg
mla
Ok, I think I understand what you mean: rather than shrink some things
and make others bigger, you remove the clutter, and only show the
things relevant to the working situation, sort of a relevance filter.
It sounds very similar to Compendium where a 'working situation' would
be a 'map'. http://compendium.open.ac.uk/institute/ Definitely check
it out, I've since added it to the list of Close Tools. Each map
opens up it's own window and has it's own canvas that you can populate
with idea nodes, and draw links between them. It kind of breaks
context for me though. The semantic zoom I was thinking of would
still allow you to see the ghost of other nodes, but only focus on a
few of them. Maybe that would still feel too cluttered though.
That's sort of the reasoning behind cupboard doors, right? You hide
the stuff behind them so your kitchen doesn't look cluttered with cups
and plates and stuff.
Ideas for Time Interface:
http://www.tiki-toki.com/ works pretty well for navigating a
timeline. maybe have a slider like that at the bottom for navigating
the history of the graph? here's a couple sketches i just made to
illustrate one possible way of viewing nodes by time:
http://dl.dropbox.com/u/5852846/fathom%20timeline%20sketch.pdf
http://dl.dropbox.com/u/5852846/fathom%20timeline%20search.pdf
x28
I like your idea of "the ghost of other nodes" -- it is much more
colorful than my term "virtual" in my distinction of visual - verbal -
virtual http://x28newblog.blog.uni-heidelberg.de/2007/01/13/think-tools-for-connective-knowledge/
. However, I am still skeptical of the power of zooming.
The geographic maps you were mentioning, use a sort of zooming that
involves cartographic generalization. For example, rivers become much
broader, important streets as well, while less important streets and
villages drop out altogether. For semantic zooming similar to this, it
would need a clear conceptualization of what would count as "less
important". The type of knowledge that can be captured by such highly
structured relationships, is only a subset of the real and emerging
knowledge to be handled by such think tools. Even if you take very
generic relationships such as Max' and Heiko's "conceptual data
structures" in their http://imapping.info/ you cannot cover fuzzy
structures that have not been marked up with these structures yet, or
those that can be modeled with DeepaMehta's Generic Topics and Generic
Associations.
Sooner or later you might end up with a hierarchical zooming that
hides child nodes, thus falling back to the desktop folder cabinet
metaphor (your kitchen cupboard example is dangerously close to this),
where the well-known compartmentalization effects do a great
disservice to creative thinking!
Matthias