Maybe every tool that we invent and use is two-faced: it makes life both easier and harder, and therefore the wish to make programming easy could result in one of the most difficult journeys so far, comparable with inventing an AI indistiguishable from a human. If i imagine a success in 75 years, then i would probably sit infront and talk to a human-like android: "play me some music based on Bach with the voice of Eddie Murphy, mellow, with a hint of tragedy, but not too sad" - then it would in the blink of an eye play me some music and sing along with it. Chances are i would change Eddie Murphy to some other singer that is readily available at the huge human voice archive this machine has access to. But i am bored with archived and old fashioned music, i want something new and exciting. So i try to teach the machine a new voice. As it turns out it is pretty good at mixing characteristics of multiple voices and synthesizing a new one out of it. I let it do some hundred random variations with hints on what parameters i prefer. "that rough whiskey vibe - keep that and mix it with Presleys guttural hiccups"... "no , not like this, listen". I try to sing what i have in mind, but i am no good singer, so the next test comes out worse than before. "go back ... one more". We spend the rest of the day analyzing Presleys archive and filtering out the guttural characteristic that i meant. I heard i am an expert programmer, my grandma called me that, and i sit down with this android for a whole month and explain in all details the music i want to hear, constantly listening and tweaking my instructions in a flawless instant feedback cycle. The most time spent, or what i sometimes feel - wasted, is actually searching for the right words in my own head to describe what i mean. After 2 weeks i found out that we have broken down the characteristic of a singers voice into twohundredandeighty relevant parameters. Relevant not for everyone of course, but for me and this music i am working on. Maybe i am taking this too far. "delete MelodramaticChorusAccentuationTongueModulationMode two to four". I spend the next week with condensing these parameters down to thirtythree. At some days in the next week i am really without any inspiration and just let the android play thousands of random variations based on the current version that i judge on a 1 to 10 basis while doing some gardenwork. Sometimes i would only change a little note, or the single expression of a syllable. "uah insiide meeee... you see the "siide" must sound more desperate because this guy is on the verge of losing his lifelong dream at that moment." - we spend the rest of the day tweaking that "desperate" thing. This android is amazing. After a month and three weeks the music is ready. I save it and do some other things, mostly gardenwork and watching the drones fly at the evening - they really got some nice new formation techniques that i enjoy greatly. The next day i speak with Andreas (another real human) and tell how good the android has learned my personal preferences for singing voices and how fractalising Bachs harmonic structures to the fifth degree is worth it but no further without structural simplification at the base while keeping the dimensionalitys denominator intact (that was really a complicated talk, i cant get into all the details here). We agree to exchange our Androids preference patterns - my bach-presley11786 for Andreas painting-images9904, and our androids get automatically updated. Next day i try it out. This may take a while, Andreas has warned me. But over the next days i can watch the android painting a really amazing photorealistic but nonetheless astonishingly dreamlike image with Andreas pattern, although the process is, as he said quite slow by design, and one image at a size of four square meters takes about 3 weeks to finish since it includes a complicated technique of iterated de- and reconstruction with additional time for the oil paint to dry up inbetween. But the result is definately worth it. After the image is finished, i call Andreas and congratulate him for his exceptional good taste on imagery (afterall he worked two and half a years on it). He also thanks me for the musical pattern and asks if he can use the guttural Elvis thing in one of his next Crazy Donkey Singalong Performance. That asking just being a polite convention, i naturally give him full permission. We agree to hold another meeting to talk in detail about his structural approach to synthesize dreams without crossing over to kitsch in two weeks. The rest of this day i am back in my garden, the drones fly really low this week... maybe a thunderstorm is coming... but this is great! Every human is a programmer now, which essentially means that he tells other beings what to do, in more or less detail, and of course the fact, that these to be told are not humans anymore, but human-like machines. So, if one of these humans does not tell a machine what to do, he usually enjoys not telling other humans what to do (besides the endless debates about how to tell another machine best what to do (and not to forget - the debates about how it could be better to directly think to a machine rather than speak to it and why it has not been implemented yet)).