bad! bad meme! (was Re: [extropy-chat] NEO deflection)

[Adrian Tymes]

--- Bret Kulakovich <bret at bonfireproductions.com> wrote:
> I thought it was worth writing to the list since this is the kind of 
> thing we are dealing with as we move forward - I learned this  
> ideathing about 20 years ago, and it gets fixed, cited, notarized  
> etc. in 15 person-minutes effort. Of course it is on top of a  
> developed infrastructure (pinyin) and I had to access the fix from  
> Adrian manually (click, go to website) and then process it  
> (understand and read language)
> 
> But how soon to skipping those other bits? I don't know if I'd trust 
> my e-brain to Norton Utilities, but, Wikipedia and NPOV? maybe.

Part of the problem is that it's largely self-service - and might have
to be.  How do you know that an idea you've held for 20 years is wrong?
For example, I believe in gravity roughly consistent with Newton's
laws (as occasionally modified by unusual or extreme particulars, such
as in quantum situations)...and so far, I do not have any evidence to
contradict that belief.  On the other hand, I hold the apparently
significantly minority opinion that much of our scientific theories on
the behavior of the universe, especially beyond our solar system, is
suspect at best since it's based on data from one location; is there
some facet, some important data about the data, that I'm missing that
would show that astrophysics is just as well established as, say, the
physics of steam engines?  I literally can't spend all my time
questioning all my beliefs: to do a proper job of that would take more
than 24 hours per day, and the results would just be more beliefs that
themselves would be open to question (even if hopefully more solidly
backed up by evidence).

And even once I do identify an idea to question, how best to do it?  I
like to think I've made Web-enabled research into one of my
professional skills...but I have to admit, there's almost always a
significant degree of luck involved.  Will I happen to think of a set
of keywords that will lead me to the answers?  Will I happen to
understand the answers when I read them?  For that matter, will I
happen to think of the right questions to ask?

There are, of course, systems and strategies to help solve those.  But
they are not perfect; they just weight the odds.  (Moreover, I tend to
grok and internalize them, feeling through their lessons until they
become unworded instict.  It's sometimes ultra-convenient for being
able to come up with the right answers, assuming I've seen something
like what I'm running into before, but it makes explaining the answers
problematic.  Running science programs on the animalistic part of my
brain, I suppose.)

There's also a present-day disadvantage to e-brains: they work best
given real-time access to the Web.  In settings where that connection
is not present, such as most of the face-to-face meetings that almost
all of the world continues to rely heavily on, the only data you've got
is the data you've cached locally.  Wireless connections and wearable
computers would only help with this if they could aid without
interfering with the meeting - a high bar (which unfortunately can't
realistically be lowered much without destroying most of the utility)
which no actual (as opposed to sci-fi or design only) wearable computer
I've yet seen comes anywhere near satisfying even when they're doing
nothing, and then there's the AI programs to grep for potential
misunderstandings and obtain the corrections, and then there's the
problem of presenting it back to the user without interrupting the
conversation.

...but perhaps just listing the problems like that is itself a start.
So, we have:
* Unobtrusive hardware (either socially accepted, or perhaps more
  feasably, something that a luddite looking at your face couldn't
  distinguish from an accepted object and thus couldn't take exception
  to)
* AI able to understand the ideas being spoken about in a conversation
  (at least well enough to identify and search for the ideas)
  ** Voice-to-text and other subcomponents of this idea have been done;
     part of this challenge may simply be identifying and assembling
     them
* Automated understanding/translation of search results into
  human-digestible chunks
* Presenting the chunks to the human without interrupting the human
  (which will probably take some skill on the human's part, but is
  there a way to make this easier for the human?)
Anything else?