
To this day I am wondering if eMule would still have an active userbase if it had some sort of mechanism that (via opt-in) allowed clients to download from trustworthy „friends“ only and pipelined those downloads from the „public“ ed2k network through these friends and their friends' networks. Great times that basically ended because of „p2p sheriff“ companies, Torrents and the dawn of One Click Hosters like Rapidshare that promised privacy and better download speeds. Download speeds were much slower than Torrents, even back then, and that was an advantage because that kept so many files alive.

It was actually relatively hard to find a file that was dead, because people kept sharing as long as they downloaded, often much longer. The main benefits of eMule/ed2K, compared to torrenting, were file discoverability (serverless thanks to Kad search) and longetivity of files. Also so many discussions where the network should be heading with the main Client devs famously mute in regards to those discussions and drama. But also really innovative ideas like downloading from clients that used a thin HTTP server within the client, leveraging your ISPs proxies (Webcache/Peercache feature) for higher download speeds. So much drama with leecher mods (leecher was negatively connotated in regards to the ed2k-protocol) that tried to download more than uploading or upload to clients only that had proven to pay back with higher speeds.

There was a very active scene back then with some eMule Mods that really pushed things further, MorphXT, Xtreme, StulleMule, eFmod, Sivka, ionix, NeoMule and ScarAngel come to mind. I remember vividly being active on various eMule mod forums back in ~2005.
