Woohoo, getting published

One of the Gumpagraphs
http://www.pastpeak.com/archives/gumpagraphs/index.htm
will be on the 'Working class consciousness' card
of the Public Sphere 'Pattern Language' project,
http://publicsphereproject.org/drupal/node/211

It's available from wikimedia.org
http://commons.wikimedia.org/wiki/File:Well_repair.JPG

I'm tickled they picked it.

Before there were maps

Back then, I would describe the place I traveled to with a linear narrative. I would describe what I saw as I walked along ... a grove of trees to my left, further along a cliff to my right ...

Frustration with communicating via linear narrative resulted in picking up a stick and making marks representing a visual model of the landscape in my head. These scratches evolved into the world of cartography and GIS we now enjoy.

We deal with a lot of complexity today, cognitive complexity, complexity rooted in relationships between concepts. We explain these complex relationships via linear narrative, describing a journey along a path, pointing out the landmarks. The audience must build this cognitive landscape, registering each component and they way each connects to each other. Given the difficulty of this exercise, many get lost along the way.

We need maps, visualizations which place the concepts in space such their relation to each other is understood at a glance.

This is a fairly well known concept, and currently implemented as marks with a stick. My conviction is that it will evolve into tools which spatialize knowledge such that more is more quickly understood by more of us.

On linking

The Unix filesystem concept of ``links`` is quite wonderful. I remember a couple conversations with my father which touched on the question "is anything possible". For some reason my example to prove that, in fact, not all things are possible, was to state that I can't be both here, downstairs, and upstairs at the same time. It seemed irrefutable that a thing can only be in one place at one time.

That may or may not be true, but filesystem links offer cool magic.

However, the quality of explanans offered by Unix could be improved, and
the Python equivilent is flat out confusing.

$ man ln
...
SYNOPSIS
ln [OPTION]... [-T] TARGET LINK_NAME (1st form)
ln [OPTION]... TARGET (2nd form)
ln [OPTION]... TARGET... DIRECTORY (3rd form)
ln [OPTION]... -t DIRECTORY TARGET... (4th form)
...

Target and link_name are reasonable names, although a bit of a mixed metaphor, target implies an arrow or bullet, not link name. I also don't like the order, I approach the link and arrive at the target, why put the target first?

Python is much worse:

>>> help (os.symlink)

symlink(...) symlink(src, dst)

Create a symbolic link pointing to src named dst.

Huh? the destination points to the source? Just Plain Nuts, Don't Make Sense, Oh Puh Leaze.

***********************************************************

I've been for some time futzing with code to wrap link management, eventually want a buildout recipe which describes link wrangling, particularly the ``gather`` idiom, which moves a file to a managed location, replacing it with a link. Useful for versioning configuration.

I like the following spelling to describe links and files:

['ln1', 'thefile'] or ['ln3', 'ln2', 'ln1', 'thefile'] which could also look like: "ln3 -> ln2 -> ln1 -> thefile"

I think it would make me happy to specify links this way, both in arguments and and reports.

That is so wrong!

Since I seem to be settling on 'Gumpa' as my primary brand* it seems only right to act my age and channel some codger energy. Hey you kids, GET OFF MY LAWN. This is my first post which will be tagged TISW.

The metaphors used to discuss hierarchal data organization are SO wrong.

Take the file system.

It's often described as a tree, a tree with the root at the top, you descend to the files, which I've seen analagized as leaves.

Im my world trees don't have roots at the top and leaves at the bottom, in my world 'to descend' is to go down.

No wonder I've always felt somewhat slow in understanding explanations of file hierarchies, a small matter of their up being my down and visa-versa. Not to mention that file systems are not tree-like, they are container-like. Files are 'in' directories, not 'attached to' as leaves are to a branch.

I don't currently have suggestions for improvements, but it seems like it should be possible to provide a mental image that doesn't require hanging upside down to resolve the cognitive dischord.

Maybe the original work on this was done by bat-geeks at night, when they are upside down. No, they tend to be in caves, no trees in caves. I Know! it was the sloths. Of course, sloths live in trees, upside down, trees with roots at the top would make perfect sense to them.

Darn you sloths, TISW

* TODO acronymize gumpa

Better if used before

I've long been a fan of the `Better if Used Before` movement. A sustainable future clearly will require rethinking our relationship to reuse, repair, restore, repurpose, and, if all else fails, recycle. The Better if Used Before folks have been around for a long time. They do a great job of getting their message out, the tagline can be seen on lots of packaging. They understand the value of timestamping, along with the slogan will be a date.

As pervasive as their PR effort is, they keep a very low profile otherwise. All that Google seems to know is the work of an artist, looks wonderful, but it seems unlikely she's to thank for all the boxes proudly declaring 'Better if Used Before'

They evidently are deeply old school, however I think their mission is of such importance that it needs to leverage the advantages of the Web.

I hope to act on that thought.

BIUB has spawned many Web-based initiatives; Make, Instructables, Hackaday, etc., it's time they dipped a toe in the waters of the cloud.

Time as connective tissue

An expansion of applied synchronicity. The principle can be applied more widely, premised on logging timestamps on the following

- audio file playing
- web page focused
- image(s) focused in viewer
- phone # active on phone
- email message currently focused

Voice recognition input provides metadata for one of the above resources, in a format like::

audio put artist bob and ray id hard as nails tag interview tag staply

meaning

- audio = determine currently active audio file
- put = create a new chunk of data about the current audio file
- artist = active field of chunk is 'artist'
- bob and ray = key:artist, value:bob and ray
- id = create a new indexed entry key:id, value:'hard as nails' pointing to this record
- tag = append interview and staply to the list of tags on this audio file

or::

image put category sunrise tag frost

- image = determine currently focused images(s)
- put = create new data for these image(s)
- category = active field of image record(s) is category
- sunrise = key:category, value:sunrise
- tag = append frost to the list of tags on this(these) images

The magic is in *currently*. We know the timestamp of the voice, we've stamped it's begin time, the file knows time offset at any point.

We parse the logs to determine the correct association to make with the voice commands, the logs tell us what audio, web page, images ... we are referring to.

I think there is real potential here, the voice recognition demands are not too great, controlled voice and vocabulary, the logging and persuant matching up seem fairly doable.

Applied synchronicity

I'm relishing the brilliance of Bob and Ray, listening to a collection of mp3's: 5 days and 22 hours of them.

They should be indexed, I don't have 6 days to devote to the project.

How could I multitask, do the indexing efficiently in the background?

It seems it could be done via time-based data matching. The audio player logs timestamps with filename as it plays the set of files. I record voice messages describing what skit is being played, these voice files are timestamped. An application matches words in my voice recording with the index into the file being played.

The ideal would be a wearing a bluetooth mic, speak into it to record an index into the currently playing file.

A more accessible start towards that goal would be to provide the index info via keyboard instead of voice.