ORB
Visualization
Technological Innovation as an Evolutionary Process
3/19/2004 2:35 PM
Comment from a respected scholar and
responses from <Director of
BCNGroup.org>
A couple of points:
* You have a number of good, bang-on ideas, as well as a few naive or misconceived notions.
* So, you need to be careful about technical areas where you don't have deep grounding lest you discredit yourself more generally (eg, AI, collaboration infrastructures, innovative operating systems & programming languages).
< I
actually do have this background and part of the team has even more. I understand neural network architectures
and history, expert systems, computational linguistics, the uses of database,
XML, RDF and KIF, I do not know a lot of computer languages but a few. I have designed and coded several large
intelligence systems, including systems used in automated de-classification
(1996), systematic analysis of trouble tickets for one of the three largest
telephone companies (1995), a system for the analysis of massive data bases on
Internet transactions (1999- 2001) and a system for the thematic analysis of
Islamic social discourse (2002).
In my opinion, the problem is that all of
programming languages are based on a first round of thought about what computer
control should be like. We take the
principled position that all programming languages should be totally replaced
by Sandy Klausner’s iconic programming language – which I also have used and
understand to a deep point. The point
is well taken that my expert knowledge of things “artificial” cannot compare
with John Sowa or yourself or Jim Hendler. However, in pure and applied
mathematics or neuroscience, I can hold my own in almost any company. >
* Semantic Web is
a rehash of DARPA ontology efforts from the 1980s by mostly 3rd stringers, but with a smattering of 2nd stringers.
<most agree with this, and most point to this as the core practical problem>
DARPA explicitly intends to "deploy known AI techniques" in simplified form for the Web. Unfortunately, distributed application in unrestricted domains is not known aspect of description logics (DL) or FoL (First Order Logic). (Fortunately, some of the smarter DL and FoLs people know this and are doing interesting research.)
<Yes and I would count Jim Hendler as one, as well as John Sowa, yourself, and others.>
The key trade off is reduced expressibility for computational tractability
< ah, this is radically important, if one can
show that there is a , well look at the Primentia patent
and work as a starter.>
(DLs, FOLs), but if you can't prove plans (yet with DLs) what can you do? { + }
Restrictions on expressibility limit the scope of application, so the main critique must center on what the actual scope is or can be. Ask for examples of successfully deployed, applications. In military and intel domains, these systems will be very dangerous if allowed to operate autonomously. Use the thought experiment of spending money for your business automatically.
< I claim that the problem is set up wrong.>
* Collaboration infrastructures for intel applications can make an organization more intelligent and effective with proper architecture informed by organizational theory and empirical study. The SRI SEAS (Structured Evidential Argumentation ) systems is probably the best of the bunch.
* There are new-wave AI approaches in some labs that can address the issues of brittleness, just as there are organizationally-ground approaches to collaboration infrastructures without debilitating problems.
<this is to be shown?>
* DARPA was originally a small agency that funded geniuses to produce break-throughs -- which is what geniuses do. If it is not failing 90% of the time on research projects, then it is not pushing hard enough and they might as well drop "advanced" from their name. Fear of failure by program managers and administrators has narrowed the research horizon to 1-2 years, rather than 10-20 years. Today, it has been captured by lesser rank intellects and, for institutional reasons, is failing to attack the paradigm shifting problems (ie, failing to respond to dramatically innovative proposals) that industry cannot address.
<we agree completely>
The task of the intellectual leaders is to conceptualize the new paradigms and frameworks. First rank intellectual leaders are in short supply, yet the country has the people; we just need to put them to work.
No Federal R&D efforts ("national projects") will succeed on fundamental science and engineering tasks without a high caliber of team lead by intellectual leaders. People get dissonance and resentment about intellectual hierarchy, but it is real and we can work with it.
So we need to mobilize the intellectual leaders to create the frameworks for us to work productivity and effectively, according to normative regimes with which we feel comfortable.
<The National Project that the BCNGroup will be proposing is NOT an R&D project, It is strictly an educational project, designed to shift the capability of the population to understand the science that has developed regarding complex systems, human memory – awareness – anticipation, and computer hardware and software.
Ask yourself if you can achieve your aims with honey rather than vinegar. How can you help the busy official see the rewards in taking a risk on a new approach rather than muddling through with previously-safe yet stale or misconceived approaches.
<At one point we are willing to shift, but for
twenty years I have seen my own work and the work of hundreds of others killed
based on ignorance and commercial greed.
When the Congressional hearing occurs I will let other speak to the
issue that I have forced to be put center stage.
<I did not want this responsibility, and I do not
want to take leadership. As soon as someone or some group takes over, I am
going to retire to teaching at a small college somewhere.
<We have developed the BCNGroup Charter to organize an institution dedicated to open science that is not hindered. When the BCNGroup becomes “real”, then the professional organization and membership will address the issues in a way that is consistent to the Charter. The main idea is that the National Science Foundation should be replaced with a private by not-for-profit organization governed by this Charter. >
If you can collect and document misdirection of science policy under the current administration, you can tack it onto the litany that Kerry and the Democrats are busy preparing, and thereby get it onto the agenda for both parties, because science and technology policy has traditionally been bipartisan.
Most important will be positive proposals for better outcomes.
<We agree.>