the dotmatrix project

Saying that Last.fm doesn’t have a clue in how to structure information retrieval in their own domain is the understatement of the century. I’ve become a pretty faithful user of the music service, but I have a hell of a time finding what I’m looking for with their search engine.

How bad is it? When I search for an artist (for example: Molly McGinn) I expect to get direct navigation to her artist page if there’s only one “Molly McGinn” in their artist database. Instead, I receive this result page:

Information Retrieval 101: Precision is better than recall. And if a system can’t be precise with a return, then it needs to be smart with how it displays relevant recall. In this case, since there are actually two “Molly McGinn” artist assets in their database, the recall should be limited to those two entries.

How did a query with the two terms, “molly” and “mcginn,” return artist results that don’t have both (not either) terms in their title? I tried searching both with and without quotes around the full name, but I get the same results.

It’s almost as if Last.fm reinvented the very premise of search for their own purposes, yet I can’t figure out what business or user objective they’ve supported.

Non-structured artist discovery during a precise search?

But that’s not the worst of it. Check out what the very same search query gets you in the soon to be live beta redesign:

Not only is there far too much recall, but the two most relevant returns don’t even bubble up to the top of the results page. “Molly McGinn” doesn’t appear until page 2, while “Molly McGinn and the Buster Dillys” don’t show up until… shit, I’m on page 6 of the results and it still hasn’t shown up.

The top results are now based on popularity. Who’s in charge over there? They’ve actually managed to make search far worse in this platform redesign.

Google To The Rescue

For all you frustrated Last.fm users, here’s a super simple way to work around their popularity-driven search results when you’re simply looking for a band page:

• Install Google Toolbar (link)
  
• When you’re on Last.fm, enter the band name, song name, shit, even your Last.fm friend’s profile name into the search field
• Don’t click on the “Search” label to the right of the field; be sure to click on the small arrow to the right of the label. Choose “Search Site” from the drop down list.

Following this method, my search for “molly mcginn” netted me these results:

It sucks that I have to step outside the domain to get solid domain results, but I have no other choice when trying to find specific pages. I’d love to close this post with a catchy one-liner to the extent of “and that’s why Google is worth so much” but this is best practice, standard approach stuff we’re dealing with here.

It’s kind of pitiful that Last.fm can’t get it right.

After 10 minutes of playing with it tonight, there are a bunch of positives to report. Information navigation is much more contextual in it’s nature, the visual design is much cleaner and there’s an interesting addition of an activity feed at play (think News Feed from Facebook).

That said, there are a few major pain points remaining.

Search Is Doing Too Much For Its Own Good

I tried to get to Molly’s artist page by using each of these queries:

• molly mcginn
• Molly McGinn
• “molly mcginn”
• “Molly McGinn”

Normally, I don’t worry about proper capitalization when searching, but Last.fm has always been finicky like that, so I figured that hadn’t changed. It looks like that has been fixed, but now the results are much worse. Now a search for “Molly McGinn” — an explicit bit of structured data in the Last.fm database — doesn’t return her in the results until page 2:

It seems as though someone decided that marketing trumps precision in a search result. If you click on the image, you can see that the return of artist matches seems to be based on popularity. Popularity is a great search filter or sorting criteria, but implemented as the default driver of relevancy?

That’s a terrible user experience.

Page By Page Players?

The Last.fm embedded player strategy must have been hatched with the end goal of driving downloads of the player/scrobbling software. Not embedding the site player in a manner that allows for navigation across the site without losing the stream — you often launch a new Last.fm window when trying to navigate from a radio page — is very problematic.

While there’s a bunch of great information on a radio page about the artist, group or tag, similar artists, listeners, nearby artist events, etc., the display of this information shouldn’t be driving the listening experience across the domain. It literally makes me stay on one page to hear my station, instead of following me across multiple pages.

Why not design the player experience so that it lives in a thin, consistent frame and when someone wants more contextual details about the song, artists, group, etc., they can puppet the info display into the page through an explicit interaction with the player?

Playing music on the site kills my experience moving about, adding info to the site or discovering new music. I literally have to make a choice: listen to the end of the song or do what I need to do on another page.

Last.fm already has my $3 per month, so maybe my pain doesn’t matter, but when non-subscribers get this redesign it might impact traffic numbers when it comes to the bottom-line.

Beta Review Wrap

Last.fm’s contextual navigation is great — I’m able to discover new music in a variety of ways– and I really dig most of the re-organization of features, along with the new UI presentation.

But getting search right is essential to a majority of my tasks on the site. Along with the implementation of the music player, this current iteration has me leaning towards stamping a huge FAIL on the beta release.

Michael Arrington at TechCrunch breaks news of a $5 “tax” that Warner Music is cooking up to impose on ISP’s — per individual customer — to guarantee protection from future liability caused by their customers downloading music, regardless of the fact that only a minuscule fraction illegally download music, let alone download music through proper channels at all.

Admittedly, this degree of hubris reaches a new level for the Industry, but dying business models will be dying business models.

The important question to focus on is: What’s going to replace these dinosaurs?

Cause there’s no way this scheme will ever fly.

In the comment thread, Dennis Ramirez makes a few good points:

[...] no, a band does not need a manager nor a label to play a local coffeeshop nor VFW hall. sometimes, they may even be able to organize a small tour for themselves.

if they are content with that, fine. more power to them.

but the majority of artists themselves are not, and there are only 24 hours in a day, so it is much more efficient if a band can hire someone to route touring, get them P.R., reviews, solicit labels or investors, etc., so the band can find time to write songs and record them.

a couple can do it, like Ani DiFranco, but they are overwhelmingly teh excpetion, not the rule.

and the manager needs to be paid (i dont know where you get the 25 manager thing from, bands only have 1), as does the booking agent, the recording engineer, the producer, the artist who designed the cover artwork, the web designer, their rent, their bills, etc. etc.

and that’s if they distribute everything themselves.

a music label can help a great deal with that, just like a VC helps a great deal in getting a startup off the ground. it’s not hard to understand that.

Ramirez names at least eight disparate communities that must be involved in an artists world to assist them in their quest for exposure and compensation. Right now, that’s what the label offers. So what could take its place?

Sounds like the perfect storm necessary for creating an industry-specific social network.

Sketching…

As I popped around town yesterday, looking to knock a few errands off my seemingly never-ending list of shit to get done, I caught a NPR segment about a relatively new book publisher making waves in his industry.

The grand idea behind Jonathan Karp’s TWELVE is actually quite elegant:

TWELVE was established in August 2005 with the objective of publishing no more than one book per month. We strive to publish the singular book, by authors who have a unique perspective and compelling authority. Works that explain our culture; that illuminate, inspire, provoke, and entertain. We seek to establish communities of conversation surrounding our books. Talented authors deserve attention not only from publishers, but from readers as well. To sell the book is only the beginning of our mission. To build avid audiences of readers who are enriched by these works – that is our ultimate purpose.

Karp spent 16 years at Random House Publishing Group prior to founding TWELVE, beginning in 1989 as an editorial assistant, working his way up to Editor-in-Chief. I’m guessing that during his run, he probably noticed something peculiar about the structure of a major publishing firm that interfered with the creative process.

From a 2005 BusinessWeek interview, Cutting Through The Noise:

[...] I’m going to personally edit every book. I’ve learned that you have the most fun and you can have the most impact when you work directly with the authors. I think I’ll have better publishing ideas because I’m also editing the book. I’ll be close enough to the content and spirit of the book that I’ll be able to communicate what’s special about it to audiences [...]

Sound familiar?

Well, it sounds familiar to me.

Over the last few years — after making a huge life change leaving NYC to setup camp in Greensboro — I’ve slowly altered the definition of dotmatrix from strictly being an online design and strategy consultancy to include what I now like to refer to for now as a “next-gen music label.”

Forget the consultancy aspect of the equation for this particular conversation (to my current and potential clients: I’m still in the game); I’ve been stressing about what I should call my work with promoting shows, local musicians and, in particular, Molly’s career.

And the term label just doesn’t sit right with me.

Why?

My mental model of what a “label” represents has been corrupted over the years to be squarely centered around the business of making money, much more so than the business of enabling the growth of musicians.

That said, I’m not about trying to create some kind of old school, micro-managed, opaque, middle-management, tired ass A&R tiered nightmare organization to pull down a buck while skimming over artist development and promotion. Hell, I don’t want to “sign” artists to anything even closely resembling the notion of a contract; I want to empower them to make it happen for themselves.

Bob Lefsetz weighs in on the future of the label:

Will there be labels in the future? Sure. But they won’t look like and won’t have the same names as the big four companies today. Because the new labels will be about building acts and maximizing revenue in all areas of exploitation. They’ll be about transparency. They’ll be run by geeks as opposed to mini-mafiosi. There will be a level of trust between performer and businessman. All things today’s majors abhor, which will contribute to their marginalization.

Lefsetz is on the right path here — transparency, non-mafioso business types, trust — but there’s still a degree of traditional thought buried in his perceptive noggin’; the percept that an artist needs a businessman to make shit come together.

See, I’m thinking that what replaces the traditional label will most likely be more of a service — something that brings together and overlaps all communities of the current industry in a way that enables artists, producers, engineers, venues, merchandisers, lawyers and fans to self-connect, collaborate and/or support one another — rather than a business model where non-musicians represent the artist simply to play matchmaker, get muddled up in the creative process or push avenues of exposure on artists that might be more about their own agenda.

Jonathan Karp experienced the craziness within his industry and created TWELVE as an answer for dealing with the insanity. As a relative outsider to the “music industry,” I don’t have the muscle memory of 15 years within the business, but my goal isn’t to replicate this diseased model locally or to arrogantly focus on a new angle based purely on assumptions derived from creating information architectures for the live web over the past 10 years.

Over the next year I’ll be digging in locally to promote shows, expose artists, book acts, learn how to both mix sound for a show and a live recording, read up on copyright & revenue sharing and push the edge as far as possible in freeing up music and media in order to build community.

In other words, I’ll be asking questions, listening, supporting, learning, getting my hands dirty, modeling, designing… then building.

Free music, from indie to major label, is all over the web these days, but quickly (and legally) finding what you’re looking for isn’t quite as easy as one might think — even in the age of Google.

Google has structured it’s search interface to return by object type (web, images, video, news, etc.), but it relies on the user to add specificity through query strings for returning more particular content types. Most normal people (read: not living between San Jose and San Francisco) give up searching after a few different keyword attempts, so finding free, full-length tracks by particular artists for playback or sharing online can actually be quite difficult.

Until Songza was born, that is.

Songza has removed the mind numbing query construction out of music searches by simply allowing the user to search by artist or song name. I’ve run a number of obscure, indie artist searches, and each has brought back a decent sampling of full-length tracks.

Revolutionary? Not quite.

Under The Covers

How does Songza do it? It’s pretty simple — they mine YouTube’s flash movie pages for keyword matches.

That’s it.

When Songza finds a match, it returns the YouTube video title along with the audio extracted from the video. Might I add that in most cases, it’s of extremely shitty quality, as it completely relies on UGC.

Without including services other than YouTube as input for it’s search results, Songza really can’t be touted as a “music search engine.” It doesn’t even scrape the internet for MP3’s to include in its results (which is completely understandable — I wouldn’t put that kind of chum in the water for the lawyers over at the RIAA either).

The YouTube reliance isn’t a guaranteed sustainable route either, though. YouTube could easily decide to shut down Songza’s access, similar to what Rhapsody did to YottaMusic this past week.

So why use Songza? What’s the draw to keep me from just running a Google video search for “obscure” acts and songs?

Not much, really, but for the sake of this review, I’ll focus on the only two features unique to Songza in comparison to YouTube itself.

Music Embed

Out of the four Share features (Send to friend, Link to song, Watch on YouTube and Embed it on my site), the embed feature is the only differentiation from the YouTube experience itself. Since I’m a blogger that uses embed often, there’s a slim chance that I might use this service just for that feature alone.

But since so much of the music that Songza returns is of such shit quality, the feature becomes useless to me. I mean, what would my readers appreciate more:

A decent video with shit quality audio ?

Or the same shit quality audio, but without the movie to distract you from the shit quality audio?

It’s a nice thought for a feature, but quality audio matters. Well, at least for me it does. So if I need to embed label music, I’ll keep using Rhapsody’s free embed service and probably continue to use YouTube in the other cases.

That’s a nice segue into the other Songza feature differentiation from YouTube.

The Playlist

With all the choices available to me for listening to music these days, why would I decide to string together a playlist of semi-decent to shit quality music here?

When I fire up a playlist, I move onto other tasks, both on the computer and off. The utility of a playlist is 100% centered around the experience of listening to music, not constructing the playlist in the interface. And in this day and age, losing the scrobbling of my iTunes music or the suggestions of a Pandora just isn’t acceptable.

Aza Raskin has done a nice job with the interface — it’s simple, elegant, Humanized even — but he’s completely missed the boat on modeling the usefulness of the end product.

I highly doubt creating a full-fledged music service was Raskin’s goal in the first place. There’s no money behind it, no sign-up functionality and according to Raskin, it was an idea he came up with in the shower and it took only a few weekends to pull together.

Sans the utility of the actual product, it serves as a nice business card / portfolio piece for pitching Humanized interface projects.

(originally uploaded by michele pedrolli)

We’re not yet living in a completely digital world, but to say that the heyday of brick and mortar music stores is in the rear view mirror would be the understatement of the decade. Distribution has moved squarely online, with digital downloads and music subscriptions consistently marginalized CD sales profits, which has caused music stores / chains of all shapes and sizes to fold left and right.

While shopping online is a huge advancement from the good old days, the amazing thing about today’s internet isn’t its capability to support e-commerce ventures (we’re now a decade deep into that angle). What’s amazing is that for every major music “store” experience you find online these days (iTunes, Amazon, Wal-Mart, etc.), numerous innovative hybrid services have found their own markets to provide supportive elements (online radio, SEO event listings, artist social networks, etc.) that help all musicians — particularly independent and small label artists — become exposed to potential fans and customers around the world.

You know, the types of support services that once made music labels invaluable to an artist.

So, in my never-ending quest to deconstruct the music industry, I’ve decided that I’m going to profile each and every music service found online today, focusing on the user experience from both the customer/fan and artist’s perspective. And instead of waiting to publish a cumulative findings post, I decided that I’m going to use this particular post to:

• Add newly discovered services to the review list
• Iterate the service categories (store, radio, etc.)
• Establish the review criteria for individual categories
• Add links, one at a time, to each full-post service review
• Construct a succinct, at-a-glance comparison list to expose my findings

What I ask of you — whether you’re a long-time reader or you’ve just stumbled upon this post two years from now — is for your help in crafting the approach to this long-term focus. I’m open to all types of feedback for how I should structure this review to make it the most useful it can be. So, whether you know of a service that I’ve overlooked or you simply think that I should re-categorize a particular service or you’d just like to share your own experience using one of the services that I plan on covering, please let it all drop in the comments.

If your suggestion impacts this evolving post in any way, I’ll link back to your blog in the body of the post. And if you don’t blog, well, I’ll be greatly appreciative nonetheless.

Here’s the current list of services that I plan on reviewing:

Amazon MP3
Amazon Music
Amie Street - Review score: 8 out of 10
BroadJam
BuyMusic
CD Baby
CreateSpace
eMusic
Fiesta (mp3fiesta.com)
iLike - Review score: 8 out of 10
iTunes Music Store
IndieMusic
KnowTheMuiscBiz.com
Lala
Last.fm - Review score: 7 out of 10
Live365
Lulu
Magnatune
MOG
MP3tunes
Musicovery
MySpace Music
Napster
Pandora
Payplay.fm
Radio Indie
Rhapsody
SeeqPod
Social.FM
Songza - Review score: 3 out of 10
thesixtyone.com
ThePirateBay
Virb
Virgin Digital
Walmart
Yahoo! Music