P2P filesharing. I loved the idea of this when it was all the rage c. 1999-2003. Cut out the big corporations, cut out the ISPs, no need for servers, just people sharing data with each other directly. Only problem was: it was slow, unreliable, hard to get started with, and the RIAA would sue you for it.
I do find it interesting how so many prominent companies grew out of ideas or founders that were deeply connected to the P2P boom of the early 2000s. Facebook (Mark Zuckerburg worked on a media player before Harvard and a filesharing network in parallel with Facebook, and its first COO Sean Parker co-founded Napster). Uber (Travis Kalanick ran a P2P company that was acquired by Akamai). Skype and Rdio, both by the founders of Kazaa. Y-Combinator (RTM's research at MIT was in distributed hash-tables). Bitcoin, which builds on P2P algorithms for exchanging transactions. Bittorrent is now widely used legitimately for distributing software updates.
It actually wasn't hard to get started with at all. The likes of eDonkey/eMule, LimeWire, Kazaa and Napster all became hugely popular (and some remain popular in certain jurisdictions) precisely because of how trivial they were to use.
Then, you already mentioned BitTorrent. P2P filesharing wasn't a terrible idea at all, and BitTorrent is evidence.
It's all relative to the alternatives. P2P was easy by desktop app standards, but concurrently with its development, the web started taking over. With webapps, you didn't need to download a program, you didn't need to deal with your firewall settings (consumer firewalls were just starting to become popular during this time period as well), you didn't need to configure upload/download rates or servers - you just clicked on a link and got a file.
Did anyone apart from the VCs hoping for another Snapchat think it was that good? Even the founders, infamously, cashed out early.
Same applies to the little-lamented Color, which was universally panned on here for more than just the excess represented by a $30m initial fundraise.
As for companies I thought could be huge and didn't really make it, a 4x return on Hunch was pretty unspectacular. I think there's a huge company waiting to happen in the [meta]event search space, but that's not borne out by the companies that have tried it.
I do find it interesting how so many prominent companies grew out of ideas or founders that were deeply connected to the P2P boom of the early 2000s. Facebook (Mark Zuckerburg worked on a media player before Harvard and a filesharing network in parallel with Facebook, and its first COO Sean Parker co-founded Napster). Uber (Travis Kalanick ran a P2P company that was acquired by Akamai). Skype and Rdio, both by the founders of Kazaa. Y-Combinator (RTM's research at MIT was in distributed hash-tables). Bitcoin, which builds on P2P algorithms for exchanging transactions. Bittorrent is now widely used legitimately for distributing software updates.