I posit that there really can be no legitimate arguments against acceleration of novelty/information complexity toward a Singularity- but if one must choose the least wrong criticism I have always said that it is Lanier's software argument that Jim seems to offer as his main argument against acceleration-
the simple fact is that as long as humans develop high-level software there can be NO accelerating progress because the tools of the human mind aren't accelerating - the brain hasn't changed in a million years- so how can one argue for acceleration in software complexity/utility?
the issue here is the myopic anthropocentric view of the tiny period that humans have been engineering software-
just as the genome project proceeded slowly and linearly at first but then had a vast leap to completion through the sloppy process of shotgun sequencing- Stephen Wolfram's NKS approach that sorts and searches for utility and function in the Computational Universe is about to do the same for software:
http://www.youtube.com/watch?v=60P7717-XOQ http://vimeo.com/7466113by mining the Computational Universe for utility we have a new paradigm that has actually made software catch-up with and out pace hardware in one great leap! by simply running searches through the Configuration Space of all possible programs we can discover ideas/designs for software and hardware/ as well as actual 'kernel'/seeds for SAI and even discover extra-terrestrial intelligence and culture by simply running a
Schmidhuber search algorithm that enumerates and sorts through all possible universes for some instance of utility or function- such searches would find forms that are incomprehensible to us- and perhaps even SIMPLER than what we already have- yet they would produce things like sentient transconscious mind or any software tool you can imagine- a new paradigm where you specify function but get a black box that you could likely never unravel to discover it's secrets and methods- a piece of code that just WORKS-
so we have the software means to find software of ANY possible complexity and function- but now we must wait for the brute force of hardware to give us enough ops to do such searches in reasonable amounts of time-
NKS has defeated and transcended Jaron Lanier's software issue!this approach in AI accomplishes the same method as uploading- that is not trying to re-invent the wheel- Nature has ALREADY invented AI in the human neocortex so we should simply record/copy that process instead of trying to create AI from code- or mine the Computational Universe for ALL POSSIBLE kinds of mind-
Terence McKenna said that mankind likes to think his progress is like striving upward on a ladder or staircase- but if he would stop and look around he would see he was walking in place on an ascending escalator - the evolution of the universe itself drives the acceleration so regardless of our efforts the tools emerge as they are needed despite any road blocks we believe block our path
