Re: Boid writeup idea

On Sat, Mar 27, 2010 at 7:50 PM, thunk <[e-mail]> wrote:

>
>
> Andrew,
>
> Systems are rather like languages.  I'm not sure there is any one
> reserved word/command that makes ruby "better" than another language -
> but somehow the sum of all the parts turns into something very
> powerful.
>

Ruby has its place. However, the programatic paradigm (functional, oop,
Procedural ,etc) and the syntax are only one aspect of a language. At some
point, you must look at how the language is implemented from a lower level.
Ruby does a lot stuff in the background. This is both a blessing and a
curse. It is no secret that Ruby isn't as fast as
a compiled language. There are pretty fundamental reasons for this.

>
> And yes, heads on, "get it done wise" this is an INSANE approach to
> any one  problem,  but it also has some general utility.   I'm
> struggling to keep the lines so there is a framework to get pulled out
> of it BUT that is not as simple as it might seem.
>



> What justifies the efforts, to my thinking, and this was a
> "requirement" from the git-go is this:
>
> 1   People experience things anywhere, anytime, anyplace
>
> How is this point relevant to a program? Is there data being collected
based off these obserations?


2.  Some people experience exceptional things of exception worth to
> other people  (We are all a big Swarm you know....
>

How does this get inputted into this proposed system? Is this a numerical
value system like a star rating or is it tagged?



> ((3   Text is based on language, we don't all share it.  We are
> overwhelmed by blobs of it. They get lost, confused, misunderstood, re-
> retranslated and eventualy just plain boring.  Our CPU based friends
> have a bigger contribution to make than dynamic tags to point to more
> of the same.))
>

There is an entire field of study within  information theory that deals
specifically with the issue of understanding large blobs of text.
Infact, the vast majority of the people studying that aspect of Information
Theory probably work for or contract indirectly with the NSA.
How does this proposed system contribute to resolving the loosely stated
problem?

>
> 4. In steps a "BOID" - its a crisp, clear, set of basic assertions
> that can be authored into a basic "form" by a "Expert Session" and you
> have the "Shell" for something useful.  Its a mini-plug-let.   Its a
> class.   Its a text file.   Its doesn't DO anything - it gets done to,
> but it makes a small contribution momentarily to something bigger -
> give me 1000 of these things and you can replace a average WalMart
> clerk, and an average "department"
> depth of knowledge wise.  (Like I heard myself tell Andrew - it might
> be in charge of "vacuum cleaners" and the individual Boid might be one
> of 100 dedicated to the world changing task of calculating the average
> monthly cost of filters for your house size / family, and number of
> long hair cats - ) Well Andrew didn't let me finish but he
> understood.
>
> If a program isn't crispt and clear, its either meta programmed, poorly
programmed, or both. I am not seeing the significants of what appears to be
an abstract factory pattern or plugin architecture.


> 5. Back to #1!   yes these things could get blasted at info on an
> exchange and so on BUT I think a salient point is getting
> transliterated out of things here: THE WEB DOES THE TIME/PLACE thing
> for us, the "most exciting" thing I see about these things is that
> they can put the Doctor's experience in Israel into a usable FORM (out
> pops the boid) - and that Boid can get fired!  at your Medical Records
> before your surgery in Boston.  Routinely - no midnight calls, no blue
> lights, just a Right Boid at the right time (but one of millions)
> making some sort of contribution to the world system.
>
> Its a probability thing, like a "Hero pet cat" - there are millions of
> them sleeping/eating/and pooping all across our country - eventually
> one of them is going to wake us up before a fire - well, something
> like that - if we bred smoke sensitive cats that were pre-programmed
> to jammer when there is smoke - they could be called fuzzy friendly
> feline Foids or something..... but they'd have a USE and if I have a
> point in this paragraph that is probably it.  And this isn't entirely
> off the wall because I'm learning from actual experience with really
> "pumped up patient records / 10 or so health conditions) that about
> 1/2 my boids DON'T FIRE - and that's at "two passes" - I'm seeing
> 3-4-5 passes for many problems domains.
>
> I would argue that pets have plenty of uses as is.
This is a end user problem not a pet entity problem. This again is fuzzy.

This sounds like event driven programming.


>
> BTW - I woke up thinking about the best name for this, and I think,
> unless there are working Boids somewhere that are offended by it, I
> like the "Boid" name best precisely because they contribute somehow to
> the confusion that must proceed the "ah ha" moment when it all clicks
> into place.  If really bright guys like Andrew  hear a phrase he
> thinks he knows, no flag goes up, and he thinks "oh another Rinda
> based sextuples" how boring is that?  These little guys are not
> pretentious - they are working class, pabst drinking, tv watching
> little worker boids that go to work and get something done.  (like
> their SwarmShepherd - me)
>

Boid has a very specific meaning to me. I can assure you that Rinda is not
the
first thing that came to mind.
Again, it sounds to me like useful idiots. That is these are limited in
individual scope, are given a direction, and do one task.
This correlates well with American Politics.


>
>


> and one last BTW - I have picked a domain with really good data for my
> commercial path - DRUGS - but the OTC / Herbal medicine based ones -
> the 380 or so officially sanctioned guys, some of which were used by
> Roman legion doctors, passed to the Arab lands, came back, jumped the
> ocean and are appearing on the counter at your neighborhood
> Walgreen's.  All of which I find fascinating.  I have a 300 page book
> by Canadian writers, based on German "monograms" the guts of which
> I've maybe 40% translated into Boidalese.   Give me a list of your
> health issues, and so far I catch all the matching conditions, and
> list all the contra-indications and give me 50 more boids and I'll
> list all the adverse effects.  They will easily estimate your daily
> costs and such if/when actual instances of these drugs are plugged
> into the system - all info readily available on the net.
>
> There is a lot of money in this field. I am surprised that Intel isn't
willing to fund it.
Have you talked to them?

Again, this sounds like an emergent Expert System. There is nothing wrong
with the idea, but its worth calling a duck a duck when it is a duck.
How is this not an Expert System?




>
> And that's my addition to Andrew's comments - I think he sees uses in
> domains near and dear to him and he has given himself a few hours to
> think about it, I've been staying awake nights for months working with
> it, he might be 100X brighter than I am but I don't think he's had the
> chance to think out what all makes up the value of the results.  the
> Doctor in Israel thing is highly concocted, but that's a life to get
> saved and there's is a point in there that nothing like that appears
> to me to be happening yet.  So if all the university stuff is so
> advanced, then WHY NOT?   The technology is just sitting there - it's
> in your plain Ruby without a single gem to pump it up until this gets
> connected to the Web.
>

I have spent the last 4 years thinking about async distributed systems.
I am not a AI researcher but I do work with statistical classifiers for
computer vision.
This system sounds like an expert system that asserts things in a
distributed manner.



> Sorry if I rambled.


>
> TheSwarmShepherd,
>
> thunk
>
>
> Andrew