THE WELCH COMPANY
440 Davis Court #1602
San Francisco, CA 94111-2496
415 781 5700


S U M M A R Y


DIARY: April 4, 2000 08:36 PM Tuesday; Rod Welch

Eric Armstrong reviews scope of DKR to augment reasoning.

1...Summary/Objective
2...Augment Human Reasoning
......Feedback Metrics Improve Communication and Productivity
.........Mind Anticipates Future Efforts While Engaging Current Work


..............
Click here to comment!

CONTACTS 

SUBJECTS
Eric Armstrong
Augment Not Automate Thinking for DKR
Augment Thinking with Computers
Augment Small Scale Capabilities, Group, Individual
Knowledge Management
Knowledge Representation, 000405
Augment Reasoning by Technology

1009 -    ..
1010 - Summary/Objective
1011 -
101101 - Follow up ref SDS 70 0000, ref SDS 69 0000.
101102 -
101103 - Eric submits ideas for guiding design of the DKR that augment human
101104 - reasoning.  This part is pretty well focused, and seems doable.
101105 -
101106 -
101107 -
1012 -
1013 -
1014 - Progress
1015 -
101501 -  ..
101502 - Augment Human Reasoning
101503 -
101504 - Follow up ref SDS 39 4736.
101505 -
101506 - Received ref DRT 1 0001 from Eric Armstrong to the Colloquium, and
101507 - responding to Jack Park's question...
101508 -
101509 -      So, why not let's enumerate the problems of knowledge
101510 -      representation we want to solve, then start from there.
101511 -
101512 - Eric says...
101513 -
101514 -      The goal is to *augment* human reasoning, not replace it. So,
101515 -      while I applaud efforts aimed at teaching machines to understand
101516 -      "tree" and "apple", there are simply too many nouns in the world
101517 -      to make that approach useful any time soon. Even if we *do* reach
101518 -      the point where machines can understanding everything, I'm not
101519 -      sure I care. If the machine doesn't make *me* smarter, then I am
101520 -      fundamentally dependent on it in ways I don't like it.
101521 -      ref DRT 1 2106
101522 -         ..
101523 -         This aligns with objectives of SDS reported on 000106,
101524 -         citing background on AI. ref SDS 38 1316
101525 -
101526 -         On 000120 consideration was suggested on "intelligent
101527 -         processing" of agents. ref SDS 39 4736
101528 -
101529 -         On 000125 computer aided thinking was reviewed by the
101530 -         Colloquium based on referenced resource. ref SDS 41 3008
101531 -
101532 -         On 000324 at first project meeting the record cites this issue
101533 -         as requiring a decision. ref SDS 59 0638
101534 -
101535 -         On 000330 the project plan did not address it. ref SDS 67 1479
101536 -
101537 -           [On 000407 listed as possible agenda item for project team
101538 -           at SRI to understand this distinction. ref SDS 72 2752
101539 -
101540 -      Machinery acts as a tool, as an enabler that helps make me
101541 -      smarter -- one that relieves me of much of the repetitive labor.
101542 -      ref DRT 1 3358
101543 -
101544 -      When I think about what I do in the design process, it really
101545 -      looks very much like a logical system. There are alternatives
101546 -      (or), aggregations (and), implications (therefore), and negations
101547 -      (not). There may also be syllogistic inferences (a->b & b->c & a
101548 -      => c), as well as contradictions (a & ~a). ref DRT 1 0630
101549 -
101550 -         Performing these operations requires a lot of daily working
101551 -         information called out by Doug, reviewed on 000327.
101552 -         ref SDS 63 3971
101553 -
101554 -      Outlines on a computer can easily rearrange things to *build* the
101555 -      organization.  The difference from manual practice is like night
101556 -      and day. The outline becomes a tool that helps get organized.
101557 -      ref DRT 1 1938
101558 -
101559 -         POIMS makes a similar point about the power of editing.
101560 -         ref OF 2 3742
101561 -
101562 -
101563 -       ..
101564 -      Feedback Metrics Improve Communication and Productivity
101565 -
101566 -      Multiple feedback loops and the need to track implications from
101567 -      one document to the next are tasks we leave up to individual
101568 -      designers. Their numbers are limited, due to the difficulty of
101569 -      integrating large volumes of information and maintaining all the
101570 -      "mental links" necessary to do the job properly. But what if
101571 -      there were a tool that assisted us in that process? Might it be
101572 -      possible for average programmer-jock to perform like
101573 -      super-designer? ref DRT 1 1450
101574 -
101575 -         On 000227 Eric cited similar scenario for promoting an
101576 -         effective DKR. ref SDS 49 1248
101577 -
101578 -         POIMS explains feedback is critical to effective knowledge
101579 -         management. ref OF 2 3774  New World Order... explains
101580 -         resistance to feedback. ref OF 4 2670
101581 -
101582 -         On 990525 there was a report that engineers do read and write,
101583 -         they just want to create technology. ref SDS 26 0966
101584 -
101585 -
101586 - Eric cites...
101587 -
101588 -          ..
101589 -         Mind Anticipates Future Efforts While Engaging Current Work
101590 -
101591 -         Design process develops problem statements, functional
101592 -         requirements, functional specs, data structures, design specs,
101593 -         help systems, user guides, and other documentation. The mind
101594 -         does not stay rigidly fixed within a single document space --
101595 -         it happily jumps to the nearest logical node, regardless of
101596 -         the space it is in. So, while thinking about part of the
101597 -         functional spec, the mind may leap to related ideas in the
101598 -         design space, the data structure space, or the requirements
101599 -         space. ref DRT 1 6438
101600 -
101601 -             Understanding human mental machinery is helpful for
101602 -             designing knowledege management.
101603 - ..
101604 - Eric proposes tools should support...
101605 -
101606 -         1.  ...capture associated ideas easily and naturally, without
101607 -             having to "change context" to another document to do so.
101608 -             ref DRT 1 1102
101609 -
101610 -              Seems to align with Doug Engelbart's reference on 000327
101611 -              to his 1972 paper explaining knowledge management handles
101612 -              daily working information. ref SDS 63 3971
101613 -
101614 -              [On 000405 Eric indicates a journal would not harm the
101615 -              DKR but the value is not settled. ref SDS 71 1972
101616 -
101617 -         2.  ...tag ideas as "design" "functional spec", or whatever --
101618 -             and it must be possible to do so after the fact, rather
101619 -             than requiring the author to accurately predict the
101620 -             correct category in advance. ref DRT 1 1960
101621 -
101622 -         3.  ...collect all ideas (nodes) of a single type to create a
101623 -             "document". Collect all of the data structure notes, for
101624 -             example, produces at least the initial version of the data
101625 -             structure document. ref DRT 1 3060
101626 -
101627 -               These are good requirements.
101628 -
101629 -
101630 -
101631 -
101632 -
101633 -
101634 -
101635 -
101636 -
1017 -
1018 -