OK... just a few quick points to add to this:

1.  *Inclusion of code*.  I believe AGI would *best* be achieved by a
combination of theory and craft.  A joint project / consortium should
actively encourage people to experiment with AGI code.  Also, pure theory is
very dry, having code will increase the fun 10x.

2.  *"Theives harvest and run"* is indeed a serious obstacle to trust.
Especially for things like a good NLP module -- many people can steal such a
module and apply it to their *non-AGI* biz.  The
"indebted-to-mother-project" clause mainly protects against professional,
big, AGI projects.  So we need to compensate for members taking the risk to
disclose easy-to-rip-off stuff like NLP modules, perhaps with cash and a
bigger chunk of shares (and yes, I'm aware of legal problems with shares,
let's assume this at the mo).  That is, compensation should vary
case-by-case.  Perhaps we can set up a fund to pay people in cash, and I can
contribute some seed capital, and some other sources.

That said, remember that the "indebted-to-mother-project" clause will be
attached to all the stuff (including code) that we disclose.  If anyone
steals our code by ignoring the clause, they're still *pretty likely* to get
caught when their business becomes big and famous.

3.  *Making money*.  I think making money in the medium term is highly
probable if we can pool together a group of people with code contributions.
But we need to compensate them for the special kind of contributions they
make and the risks they take, as explained above.

4.  *Accept members as broadly as possible*.  A typical AGI company usually
interviews potential candidates, sign NDAs, and then see if their skills
align with the company's project.  After such a screening many candidates
with good ideas may not be hired.  The consortium is to remedy this by
letting members with disparate views exchange their ideas freely, with the
safety of being credited for them.

A "research consortium" is a joint venture where companies in the same
industry collaborate to bring about an R&D breakthrough that is desperately
needed and yet is beyond the capacity of any single company to achieve.

5.  *Management is needed*.  I was wrong to think that the consortium can
run entirely autonomously via some mechanistic rules.  We do need a managed
organization to run this, though the simplicity of operation is still highly
desirable.

YKY

-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e

Reply via email to