| Let me just ask - When has anyone ever actually invented anything totally new? The telephone? It was referred to as the "speaking telegraph" (telegraph + speakers). The car? Horseless carriage (engine + wheels + steering + brakes) The plane? Glider + engine. The Wright brothers invented powered flight - not flight. The difficulty was getting the weight to lift ratio high enough with a primitive heavy engine on board. Gliders already existed - you just couldn't go anywhere with them! You can't just jump off a cliff and glide to China. Google? A more advanced MITS algorithm + inktomi's commodity cluster map-reduce architecture + inktomi's PPC. Indeed, had inktomi doubled down on search instead of their CDN, we might well be talking about inktomi and not Google. General and Special Relativity? Nope - http://www.quora.com/If-Albert-Einstein-had-never-existed-at.... There are no really new ideas out there - merely combinations old ones that "hang in the air". There are no new ideas - merely old ones combined in unique ways. Everything is a remix (https://vimeo.com/14912890). |
Arguing that things come about as combinations upon combinations of small, 'not completely new' inventions is begging the question. When you compare the state of human affairs in 5000 BC with the state now, you can't possibly mean that no 'completely new' things have been created.
This is the same problem as e.g. distinguishing species in biology. Of the countless intermediate forms, you cannot point to a single one as a 'completely new' species. However, 1000 generations after an earthquake, the species on both sides of the created ravine have diverged so much they can no longer interbreed. There are now two species where there was one, so a new one must have come into existence, right?
Andreessen is completely right in suggesting you don't want to aim to invent something 'completely new' on the spot. Nevertheless, if you can see where something could be going, you could aim for something new say 10 years down the road.