Re: One vs many directories

[Jean Louis <bugs@gnu.support>]

* Texas Cyberthal <texas.cyberthal@gmail.com> [2020-11-28 11:20]:
> Hi Jean,
> 
> > What should it be or do?
> 
> Dbmind does things that Postgres handles better than Org.
> 
> > As you have specific thought order in directory names then maybe
> > such could be parsed, maybe slashes / removed to show a full path
> > to the file. This becomes long but could be useful in some lists.

> I don't intend to do so.  Textmind maximizes path dynamism via
> Dired+Treefactor.  Links shouldn't break that.

I enjoy these thoughts.

Maybe you refer to browsing the tree.

I am switching completely from browsing the tree into index type of
access. Index helps to find the thought which in turn finds candidates
and file is in front of me. How I think now is little different than
past and different than future. If I think of term "Insurance" related
to Germany those are two words to search for. Maybe in future I think
of "health" then I should be able to find "insurance". There is no
fixed pattern on how to think. Repetition in locating the node creates
habits that next search is even quicker found.

Those most searched nodes could be ranked automatically for even
quicker access. And all searches for nodes could be recorded for
future as inexpensive and automated log that helps person find what
happened in the history and which nodes have been located at certain
months, dates, years.

org-store-link would get its hook or additional function that each
time when Org mode related heading is is stored that it ranks the
file the heading in a separate list which can be stored on file
system.

After a while user will get a subset of highly ranked headings in
their corresponding Org files. That subset then can be used as quick
bookmarks or get bound to keys.

> > Alright and I find that it is the case on my side, and previous work of Engelbart, then also within some other information management systems, like Semantic Synchrony.
> 
> Some of that might qualify as an algorithm, but not a natural thought
> algorithm.  A natural thought algorithm must manage substantially all
> natural thoughts while satisfying the definition of an algorithm.

I wish I could understand definition of "natural though algorithm" in
the context how you refer it to.

From other people's experiences I can see they are thinking
different. It is questionable if there is one algorithm corresponding
to many people's natural thinking.

The algorithm in locating specific file on my sides is programming
algorithm based on what I find most quick for me personally. There are
several such algorithms programmed that help me locate things.

I have 19489 nodes, everything is together. There is no visible delay
for completion to show me list of nodes. The list of everything is
offered for completion and I can type what I think that I need.

If I search for PDF related to "mercury" I will locate it as quick as
1-2 seconds and can open it already. Or I could search by word, tag,
attribute and get a list of related hyperlinks and
hyperdocuments. From there on I can locate the right one.

> The things you mentioned are not even as sophisticated and complete as
> GTD.  And GTD is merely a personal paper-management algorithm, not a
> natural thought algorithm.

I did look on the hyperlink you gave me. It looks like algorithm
deciding what to do with tasks for me. But that is not how we manage
tasks in the group.

We have processes that are defined from A to Z and tasks are managed
with to achieve the overall purpose. A purpose could be project
accomplishment such as "bridge built over the water". Then tasks are
steps bringing the project closer to accomplishment, such as purchase
timber, saw, meter tape, screws. Project planner is the key here as
one should not put nothing more and nothing less, and none task shall
be defined that is not doable. No algorithm should influence tasks to
be pending, or archive them or decide to do task if it is very short
as relations to task and from the task to/from other objects are not
so simple. There can be 10 tasks each 2 minutes long that cannot be
conducted right now as the phone call and subsequent travel may have
so much more important.

I was referring to this:

> After reading appendix IV to Making it all work (another book about
> GTD by David Allen) it occured to me that the Process phase of
> "Mastering Workflow" would be well-rendered using psuedocode. This
> is probably due to the fact that there are several decisions to make
> along the way (logic) and this happens for each item in your inbox
> (loop). So, without further introduction here's the code:

> while length(inbox) > 0:
> 	inbox_item = inbox.pop()
> 	thing = analyze(inbox_item)

> 	if not actionable(thing):
> 		toss(thing) or tickle(thing) or file(thing)

> 	else if less_than_two_minutes(thing):
> 		do(thing)

> 	else if is_single_task(thing):
> 		wait_for(delegate(thing)) or assign_context(thing)

> 	else:
> 		create_new_project(thing)

If the bridge gets built, all the undone tasks become redundant. The
algorithm shown tries to manage something but would be detrimental to
the actual task and project accomplishment we are doing here. Overall
this discussion does help me to see your insights and thoughts, other
people thoughts and to build upon it and create new ideas and
implementations. 

> I doubt there are multiple ways to design a natural thought
> algorithm.  For example, all natural thoughts occur in a
> chronological sequence.  This necessitates a ramblog to accurately
> reflect them.

Memories are saved chronologically. By thinking what I was about doing
in the years about 2004, I can jump to images sorted chronologically
and access maybe image in 2005 as first thought was maybe not so
precise. But it is about there.

But I cannot access image related to business ABC that is located in
2004 quickly unless such image is indexed somewhere by its meaning,
maybe "boat purchased" would be meaning related to such picture. Then
if I wish to see boats I have purchased I would type most probably
"boat" and from there find various other attributes such as dates,
types and similar.

> This is the GTD inbox algorithm:
> 
> https://michaelwhatcott.com/gtd-workflow-processing-algorithm/

I cannot relate it as useful for work we are doing. And we did
accomplish many projects successfully. Principle is like this:

* Plan

- purposes
- goals

** Plan steps

1. do
2. do

If plan step cannot be easily conducted, it has to be branched into
Project.

* Plan

- purposes
- goals

** Plan steps

*** Project 1. do

- task 1
- task 2

*** Project 2. do

- task 1
- task 2

If any purpose get accomplished, the undone tasks become redundant and
can be forgotten, but could be reused in future.

> GTD is usually called a method.  I've started calling Textmind an
> algorithm to emphasize the finiteness aspect of its design, a key
> feature.  I think I could construct a pseudocode Textmind algorithm.
> It would of course rely on human judgment for some decisions, and
> judgment is fuzzy.  But the algorithm itself would be unambiguous.

I have fully understand concepts of 10 Bin. But I cannot adapt to the
thinking pattern that I did not yet understood. I did read but did not
understand it.

Since we talked about it I have developed more so that I can soon
forget about the file system and only use conceptual access and
filing. I am in transition of all files sorted hierarchically to be
sorted semantically while program is to make sure of hierarchies.

Let us say video is there, I can file it by choosing the genre which
is node name and place some interesting tags or attributes, maybe
regisseur, movie director if necessary, there are movie researchers
who do exactly that. Once necessary attributes are there it becomes
pleasure later to access it in future and play it or send to
somebody. But there is no need any more to hierarchically browse the
tree of directories until movie is found. Any related concept will
lead by associations to the movie. If one does not remember the movie
name, maybe remembers movie director or actor or maybe genre or even
approximate movie year or other attribute.

If all associations such as node name, attributes, tags, etc are
indexed and quickly accessible (similar to database view but rather
table with the index to get blazing fast query) -- then locating any
item among huge list of items becomes rapid.

Think of a concept what you want to get? To play movie related to
"eagle"? Type word and get "Where the Eagles Dare", but maybe one has
to type also "movie" to distinguish it from subject of birds.

Think of a concept what you want to file? Designate the type, is it
movie, PDF, meeting, TODO task, note, survey, designate place, date,
group of people, person or other attributes or in other words
integrate the relations of a hyperdocument. From there on, forget
about it.

Mnemonics techniques use exactly the same conceptual and associative
remembering.

Number 2 is often associated to a swan. It should be swan doing some
action, for example, swan that is eating something where something is
the object to be remembered.

Let us say that object number 2 is bucket of metal screws. All one has
to do is to connect relations with the fixed relation of swan on the
number two that is eating something.

Then it becomes a swan drowining in the lake as swan has eaten a full
bucket of screws which shape can be seen on its neck. Just imagine the
picture as picture itself is relational object as associations are
relations. Forget about the picture.

When it comes to tell which object was at the place number 2, one
associates first 2 to swan, and immediately remembers the drowning,
the lake and that screws in the bucket were heavy and that is the
object that had to be located.