I think what "went wrong" can be summed up in a sentence. The player was never built to be a publishing platform or had any other monetization angle built into it.
As soon as it started getting features that would support this, the technical audience saw it coming a mile away and never adopted the required usage patterns needed to leverage such a monetization strategy. And not just because they wanted to avoid paying for music, but because the service model is just a pain in the ass compared to the self-service model.
"It could have been Pandora" ignores the fact that Pandora basically sucks and people just tolerate it. People want to be in control of their software despite what the business fucks running "IT" think now, conceptualizing a "service" where none is needed, trying to pull infinite profit from such trivial functions such as routing packets correctly or parsing compressed audio and routing it through hardware to a speaker without pointless and pathetic DRM algorithms trying to detect time-skewed copyright audio.. sigh. by setting the bar so low that people think that shitfuck-suck software is what you have to put up with unless you have "long term support" or other wedges into your conceptual/computational sovereignty to do something as simple as publish a document and expect it to be readable on another computer in 10 years, a problem that was solved 200 years ago is now the cutting edge of business technology and development, fueling well known multinationals such as Apple, Microsoft and Amazon.
Computing never was meant to be a platform for economic revolution, it just abstracts away paper work. You still have the exact same problems as with the internet as without it. I won't say the internet bubble is over until the world is actually simpler with computers instead of way, way more complicated and outright retarded in many aspects.
NO, WINAMP DIDN'T FAIL, in contrary, economic motivations failed to make Winamp suck, failed to steal a public resource and public work. Failed to co-opt the direction and passion of artists. AOL failed to capitalize on selfless engineering that actually solved a problem the correct way, it failed to corrupt the best solution with advertising indicators and economic feedback loops built into a market, shoehorned into an audio player. It failed to take us back instead of forward, like technology is supposed to.
I think the clearest indicator that WINAMP is a success, is the fact that XMMS exists, the fact that people use version 2.x without AOL's "support", 10+ years since release. It's a simple idea: a media player that supports plugins via a standardized API. (a real standard, not a "living standard" business-oxymoron) People still use it, for business and personal usage, and it doesn't tax them to do it, any more than a screwdriver taxes you, but not like the way a "service" taxes you.