I would imagine that the goals align with the task of "augmenting human
intellect", to borrow Engelbart's phrase.

The STEPS project, in particular, seems concerned with compact
representations that approach the entropies of the systems being
simulated. Computing, to me, anyway, is very closely linked to
simulation. A compact representation is (hopefully) easier understand,
thus making it suitable for educational purposes. However, it should
also be more computationally efficient, as well as enabling greater
productivity.

I think it's also about regaining control of our technology. A modern
computer system is composed of layer upon layer of ad hoc mechanics,
short on architecture and long on details. There are few people who have
a truly good understanding of the complete system from firmware to UI,
including all the details in between, and it's not because the details
are fundamentally complex - they simply involve huge amounts of rote
learning. Something like Linux has grown somewhat organically, without
any of the robustness that organic growth might imply.

Given concerns about security and privacy - not to mention demonstrable
correctness of operation - an easily decomposable, understandable system
is hugely desirable. There should be bonus side effects, such as running
well on lightweight mobile devices.

I hope to see computing systems becoming vehicles for training
intelligent agents that assist human endeavours - by automating menial
tasks, freeing humans to concentrate on more interesting problems, while
also leveraging the abilities that are trivial for computers, but hard
for humans (large scale data processing, correlation and statistical
analysis, particle simulation, etc.). I also hope to see more of the
abilities that have traditionally been described as A.I. entering
mainstream computation (goal-seeking behaviour, probabilistic reasoning).



Disclaimer: http://www.peralex.com/disclaimer.html



_______________________________________________
fonc mailing list
[email protected]
http://vpri.org/mailman/listinfo/fonc

Reply via email to