You have to give 64-bit x86 some credit. Intel decided it couldn't be done successfully, and developed IA-64. If Intel said it couldn't be done, I would consider it a breakthrough. Next up is helium filled hard drives, for lower drag, less power, more platters and more storage. The idea has been around for some time, but helium is notoriously hard to contain (e.g. balloons deflating). To compound the problem, hard drives are made on razor thin margins, severely limiting the price of any solution. There's supposed to be some product announcements this year, if not actual drives.
By comparison, Edison's light bulb "invention" just used a different filament material and better vacuum than competitors, with the basic device demonstrated more than 40 years before Edison. The "fundamental invention" of the transistor was based on a combination of advances in radar semiconductors and knowledge of crystal radios. It was built to be a better replacement for the vacuum tube, not anything fundamentally new, and vacuum tubes were used to replace the function of mechanical relays in early computers.
Slashdot has high quality video (at least VCR quality) and connection to off site resources. The fundamental change is being interconnected connected (open many web sites at once) vs the single connection (dial 1 BBS at a time). What manner of BBS's were you on that could scale to 621180 users or more? Do you think any could have scaled to Facebook's billion users with 500 million daily visitors and an insane amount of data flying around? Am I just too cheap with my processor and hard drive to process and store 500+TB of new, extremely-interconnected data every day, or do I need to run a beowulf cluster?