A Grand Time to Make $1 Billion Teaching Computers to Write Software December 18, 2007January 6, 2017 A GRAND TIME TO BE RICH AND POWERFUL IN AMERICA Thanks to James Musters for sending me this link: . . . The average Forbes 400 member has $3.8 billion. . . . We have a record 482 billionaires – and record foreclosures. We have a record 482 billionaires – and a record 47 million people without any health insurance. Since 2000, we have added 184 billionaires – and 5 million more people living below the poverty line. The official poverty threshold for one person was a ridiculously low $10,294 in 2006. That won’t get you two pounds of caviar ($9,800) and 25 cigars ($730) on the Forbes Cost of Living Extremely Well Index. The $20,614 family-of-four poverty threshold is lower than the cost of three months of home flower arrangements ($24,525). Wealth is being redistributed from poorer to richer. . . . It’s time for Congress to roll back tax cuts for the wealthy and close the loophole letting billionaire hedge fund speculators pay taxes at a lower rate than their secretaries. Inequality has roared back to 1920s levels. It was bad for our nation then. It’s bad for our nation now. ☞ And still the Republicans seek to eliminate the estate tax on billionheirs. KURZWEIL FOR FREE AND IN COLOR Bruce Stephenson: ‘The Kurzweil C-SPAN interview video is actually available on-line for free (though it requires RealPlayer), here.’ ☞ I repeat this because I think you’ll really be fascinated. COMPUTERS WRITING SOFTWARE Peter Kaczowka yesterday suggested, ‘We may be near the limits of what computers can do, because of the limitations of their human programmers.’ Peter’s credentials are astonishing, and yet to me this had the ring of all those infamous (if sometimes apocryphal) quotes doubting the future of technology. (‘There is no reason anyone would want a computer in their home.’ – Ken Olsen, founder of Digital Equipment, in 1977.) So, to Peter’s comment on the difficulty of humans programming computers, I asked, But can’t computers themselves take over much of the job of writing software? (If not now,one day soon?) Jonathan Edwards: ‘In a word, no. Kurzweil’s vision may someday come to pass. But in my opinion if it does, it will be after a radical change in the basic structure of computers themselves – the hardware will have to look much more like a biological brain than it does at present. And the problem then will not be getting a computer that can ‘think,’ it will be figuring out how to build one that thinks about the things you want it to think about, rather than following its own interests. (Your humble correspondent has undergraduate degrees in Computer Science, Electrical Engineering, and Mathematics from Brown University and a Master’s Degree in Computer Science from the University of Kentucky; has been employed in the software industry for 23 years; and wrote his first computer program [to play tic-tac-toe] in 7th grade.)’ Paul deLespinasse: ‘Kaczowka has made the best statement about the stupidity of computers that I have ever seen. Although all of my credentials are in political science, I taught a few computer science classes every year during the second half of my 36 year career at Adrian College, and I did a lecture every year about the dangers of personifying computers and of exaggerating their ‘intelligence.’ As Kaczowka notes, the only sign of intelligence that can be found in computers is that of the people who designed and built them, and of the people who wrote the software that runs on them. When “Deep Blue” won that chess match against a human expert (whose name escapes me at the moment) the expert was really playing against the team of human programmers who wrote the software, not against the computer. ‘It is true that a computer can write software, sort of. I myself have written a short program that produced hundreds of lines of code that would have been very tedious to do by hand, but they were very repetitive lines that varied only in small but systematic details which I furnished to the program as data. But the computer did not really write the program and did nothing creative – it simply followed my instructions blindly. (The most dangerous thing about computers is that people who don’t know anything about how they work tend to personify them, and that the more we personify computers the more plausible it becomes to machinify people – to think of ourselves and others as mere machines with no cosmic significance.)’ Ron Goldthwaite: ‘Yes, software can be written … or evolved … to write software. The most common method is John Koza’s ‘Genetic Programming,’ which he popularized (and patented) in the early 1990’s using the odd programming language LISP to facilitate the arbitrary shuffling of program parts which the genetics-inspired recombinations require. This technique is effective in some specific application areas, but not all (and LISP is no longer necessary). A quick Google search reminded me of this Salon discussion, which is a good general overview including this summary: ‘Software programs that evolve using genetic programming techniques are often convoluted, bizarrely multilayered creations, nothing like the software a human might write.’ I’m an evolutionary biologist and recognize this as a necessary result of evolutionary dynamics. My wife is a software engineer (Sun, Oracle, etc) and would fire any programmer producing such code and expecting to maintain it through its life-cycle. It’s unreadable and unfixable. And intrinsically unreliable. That’s only different in degree from our current software. But a large enough quantitative difference can become qualitative.’ John Leonarz: ‘Years ago I had the opportunity to work with the great software theorist, Dave Parnas, whose observation on artificial intelligence is still pertinent: Artificial intelligence is to real intelligence as artificial flowers are to real flowers: at a distance there is a resemblance, but the closer you look the more clearly you realize the utter inadequacy of the artificial.’ ☞ All fair enough. But I think you will find the Kurzeil video compelling. Technoprogress of the last 50 years has been astounding, and the pace of that progress, he says, just accelerates . . . exponentially. The implications – even as regards your own mortality – are dazzling. * [Mark Centuori: ‘On the subject of predictions, I strongly recommend William A. Sherden’s The Fortune Seller’s: The Big Business of Buying and Selling Predictions. It is a bit dated example-wise but the overall message is timeless (and not math-heavy). The first chapter is online here.’]