Maybe Tim Westergren has a point. The Pandora cofounder has been making the rounds in the last few months, arguing that the music licensing costs his company has to pay are crippling and should be changed by Congress. Music labels and many artists aren't thrilled about the proposed changes. So what's the deal? Could licensing costs actually kill Pandora? 

Probably not anytime soon. But it's certainly true that these high licensing costs make it much more difficult to do business. For evidence of that, look no further than Last.fm. Next month, the algorithm-fueled Internet music service is scaling back its most radio-like feature in several countries.

In the United States, Canada and the UK, Last.fm's radio streaming feature will still be available via Web browsers, but using it from their desktop client - just like its mobile apps - will become a subscription-only feature. The reason? You guessed it: The cost of securing rights to that music is too high for Last.fm to make the end product freely available, the company says.

In the UK, music licensing costs are even more onerous than they are in the United States, says radio futurologist James Cridland. It's exactly why Pandora isn't available there, nor is there an equivalent service. Last.fm is probably the closest thing the UK has, and it's about to see its functionality scaled back. 

"I'm not sure what the magic bullet is," says Cridland. "But at some point people need to stop and ask why there isn't a Pandora in the UK, which after all is one of the countries that produces the most music in the world." 

Cridland has done the math. Running a service like Pandora in the UK would require a company to pay out more money in licensing fees than they could conceivably earn in revenue.  

Internet Radio Is Expensive And Hard to Monetize

For Last.fm, licensing costs aren't the only part of this equation. It's also worth looking at the other end of the business model - namely, the revenue side. The fact that Last.fm is cutting its free radio service out of its desktop app suggests it's tightening up its freemium business model overall. 

It makes sense. Last.fm's desktop app collects a lot of listener data, but it doesn't appear to directly make money. I can play personalized stations from this app all day long without ever seeing or hearing an advertisement. Starting January 15, I'll have to go to the Last.fm website, where the entire layout is wrapped in display ads and before my Notorious B.I.G.-inspired station starts playing, Taylor Swift will try to sell me some perfume. 

The fact that Last.fm is pulling radio out of its desktop app all together - as opposed to augmenting it with audio ads - suggests that audio-only Internet advertising isn't quite as potent a force as once thought. Personalized, targeted radio ads seem powerful in theory, but they're far more lucrative when paired with video and display advertisements. Paying subscribers are even more valuable. Clearly, Last.fm is hoping to push users in either of those two directions as it continues to navigate a digital music marketplace that looks very different than it did when Last.fm was founded almost a decade ago. 

Last.fm isn't the big fish here - it's the canary in the coal mine. Pandora is far better-positioned than Last.fm, its chief U.S. competitor. Yet even at the head of the pack, business in the streaming music space is brutal. Music licensing costs may not kill Pandora, but Last.fm's troubles make it clear that the issue can wreak havoc on a music service's core functionality. No wonder Tim Westergren won't shut up.

The Swedish sure do love them some Coldplay. The Japanese? They're more into The Beatles and Radiohead, but not quite as much as they adore Utada Hikaru, a 30-year-old J-pop singer based in the U.S. 

These musical insights and many, many more can be gleaned from a massive, interactive data set published by Last.fm, the Internet radio and recommendation service. Its user base might not be a representative sample, but the numbers are still fascinating. Last.fm has also put together an interactive timeline breaking down music milestones and trends tied to its own music playback data. Rick Astley saw a big spike in listens in 2008, for instance, when "rickrolling" was at its peak. 

Last.fm has now been tracking everything its users listen to for a decade. Its "scrobbling" technology has been watching us hit the Play button since the iPod was a chunky white brick with a big click wheel. Today, it's baked into Spotify, a Chrome add-on and a host of digital music platforms and apps. 

The Value of Last.fm's Aggregate Listening Data

The service may not get the attention it once did, but Last.fm holds a uniquely significant, if underrated asset: Big Data. By embedding itself on desktops, mobile devices and within music services, Last.fm has an unparalleled, platform-agnostic view of what people are listening to. As a user, I can see which artists, empirically speaking, I listen to the most. Which songs I played excessively six months ago. 

Last.fm, of course, uses this data to recommend new music to me, which it actually does quite well. But there's also huge value in the aggregate data it collects.

What good are traditional music sales charts when people download torrents, stream albums on Rdio and check out new bands on YouTube? Not much actually.

But if we permit it, Last.fm can watch all of that activity, compiling rich and detailed data about what we listen to, both as individuals and as a group. That aggregate data could be a goldmine of insights for radio DJs and other tastemakers, argues radio futurologist James Cridland. He's right.

I hope Last.fm keeps spreading its API far and wide, and that more people opt to use the service, even if it does lack the buzz of Spotify or the headline-grabbing power of Pandora. The company should push hard to forge partnerships with music apps, hardware manufacturers and mobile platforms to keep collecting all that data. 

The insight we stand to gain would be pretty incredible.

Spotify wasn't built for discovery. The Swedish music streaming company realizes this and instead of trying to natively bake a zillion features into its service, it launched a platform for third party developers about a year ago. 

Spotify's app directory now features almost 60 HTML5-based add-ons for the service's desktop client. These apps perform a lot of different functions - some are social, while others sonically augment album reviews from big name publishers. The thing for which they're probably most useful is discovering music you might like but may never have heard otherwise.

1. Moodagent

Since launching on Spotify last year, Moodagent has been one of the most interesting apps on the platform. That's because it takes standard algorithmic music recommendations and beefs them up with emotional intelligence. 

There are at least a dozen apps that let you build a playlist based on related artists, but Moodagent factors in the mood of each song to build out something that feels more consistent. The options look broad, but are surprisingly powerful. A playlist can be sensual, angry, happy, tender or some combination of all four. You can even base them on tempo, playing back a series of similarly paced songs. Tie these characteristics to the same kind of artist-to-artist matching algorithm that fuels so many other music-discovery apps, and you have a uniquely intelligent system for finding new music. 

2. Last.fm

Last.fm has been around for a decade now, but the Internet radio and music recommendation service is still a reliable tool for discovering new artists. It works by keeping track of everything you listen to and using a Pandora-style algorithm to recommend related artists and albums. It's a simple concept, but one that apparently holds up quite well over time. 

Existing users of Last.fm will feel right at home in its Spotify app, which more or less frames a slightly modified version of the service's usual interface into Spotify's desktop client. The results occasionally need to be tweaked, but on the whole the recommendations are pretty solid. A few albums in my own physical record collection landed there thanks to Last.fm's ability to turn up hidden gems.  

3. Swarm.fm

Oh great, another social music-discovery app. Ho-hum.

Actually, Swarm.fm is pretty useful. It uses data from Facebook to show you what music your friends are listening to, even if they're not signed up for Swarm.fm. If they are, that data becomes much more detailed and easily explored. Swarm.fm will also let you know if any artists in your own collection have new releases, which is far more relevant than the new releases coughed up by Spotify itself. 

That tag cloud on the home tab might look like just another collection of metadata, but it's actually informed by your social music data. I listen to a number of artists who don the tag "space rock" - and when I click that tag, it shows me dozens of similar bands. I can then sort those artists by popularity and what's trending on Swarm.fm, which is a good way to pinpoint worthwhile listens.

4. ShareMyPlaylists

When I first opened ShareMyPlaylists, I thought "Oh, this is looks fairly generic." Alternative, Classical, Blues, Dance. One-size-fits-all playlists.

I was wrong. 

When you scroll down, you see a wide variety of very specific playlists: Beatles covers, the songs sampled by Nas and music from Quentin Tarantino films, songs featuring Moog synthesizers. It's a random conglomeration of curated listening experiences, but one that is well worth browsing. 

ShareMyPlaylists has something for absolutely everyone. Devotees of popular music from the charts can browse the "Top 50" tab while those with more under-the-radar tastes will find plenty of new stuff under the "Recommended" tab, which finds playlists based on the artists you listen to the most. If nothing in either section suits your mood, you can always run a search or use the app's built-in playlist generator. 

5. The Hype Machine

It's been a wildly popular MP3 aggregator on the Web for years, so it only makes sense that The Hype Machine would find its way into Spotify's app store. It's right at home on top of the streaming service's massive library of music. 

The Hype Machine eschews the complex algorithm in favor of human-curated playlists. Specifically, it aggregates tracks from popular music blogs across a wide range of genres, each of them very heavily populated. Dream Pop, for example, isn't exactly a top 40 genre of music, but the Hype Machine pulls together no fewer than 100 different blogs classified as such. It's loaded with music, all hand-selected by Internet tastemakers and guaranteed to introduce you to something you haven't heard before. 

A Growing Universe Of Music Discovery Apps

Narrowing this list to just five selections wasn't easy. There are plenty of discovery apps on Spotify worth checking out - top charts from We Are Hunted and Billboard and social music from TweetVine, Soundrop and Sifter. Depending on your tastes, the critic-curated recommendations from Pitchfork, Rolling Stone, NME or KCRW can be invaluable. 

It's also worth mentioning that the new, supposedly Pandora-killing Spotify Radio feature is worth playing with. Its Echo Nest-powered recommendations are not quite as granular and effective as Pandora's, but they're quite good. Not only can you create a station based on any album or artist, but you can build one off of an entire playlist. This is pretty powerful. For instance, if you've starred a lot of music on Spotify, you can build a radio station based solely on those favorites. 

Here's another Spotify Radio trick: The Last.fm app will let you generate a Spotify a playlist based on your dozen or so most-played  albums of all time. You can then start a Spotify radio station based on that playlist, which is sure to contain a few tracks you'll love, but have never heard before. And isn't that the point of music discovery?