Two trivial corollaries of Goedel's first incompleteness theorem and Algorithmic Information Theory are that: 1) for any program P there exists program Q such that program Q is both larger and stronger than P, and 2) it is not the case that for any given program P there exists program Q such that Q is not larger than P but is stronger than P.
So basically, for any program there is always a stronger program. If we take any infinite series of such increasingly powerful programs, the individual programs must eventually become larger.
This mechanism suggests via practical reasoning that a necessary but insufficient strategy to create more powerful programs is to collect as many existing programs as possible. Practically speaking, I wish to collect as much free/libre open source software as possible, and make it coexist on some set of machines. The best way to achieve this is to utilize the various packaging frameworks for software.
For the last 20 years I have been working to develop tools to index, download and package as much free software as possible, with preference given to tools that enable this very process. For instance, I have the FLOSSMOLE DataMart and ghtorrent indicies of software systems, as well as spider researchers' and research groups' pages for capable software to index, and have created a packaging queue. I then collect, packaging, and integrate these tools into our systems.
For more information, see https://altruisticsoftware.org
My next post will cover how I stovepipe many of these software systems into the FLP (Free Life Planner) project, as an application of these tools to the problems of daily life. By helping to solve daily life, FLP will enable the creation of an online collective of teams of FLOSS programmers who are working together towards the creation of increasingly powerful AI systems.
Top comments (0)