Umair Haque has two great analysis pieces, applying transaction cost analysis to the problems facing the music industry and its customers, and proposing potential business model fixes.

Regular readers here will probably have noted that I'm a fan of transaction costs economics. Since this theory appeared on the scene after my grad school days, my background is the autodidact's. I've read chunks of the original transaction economics and related market and organizational literature, and make attempts to apply it in my own analysis. Haque's background is more recent, up-to-date, and formal and it shows in the clarity of his analysis of the problems of the industry. More to say about the specific recommendations, but that will keep for later.

Via this week's Canival of the Capitalists, which see.
5:40:23 PM    

Since my own rather aged post on Bayesian and Latent Semantic spam filters still draws a fair number of hits, I thought I'd mention that Steven Den Beste has a post discussing the details and implications of Bayesian filters at - well - DenBestian length.
2:44:36 PM    

Need a reason to drive up to Carson City this weekend? If you're not already qualified to enter, you have two years to 'practice' for the next event.

Hat tip to Himanshu
2:03:18 PM    

(This is a continuation of Friday's post immediately below.)

I've already confessed that I initially fudged the landscape metaphor by implying its stability with words like peak, valley, tower. Makes for an easier image to start.

Real audience interest isn't that way of course. There is a landscape, but it's a terrain in motion. To take the most obvious driver, real world events can cause sudden and large shifts. The big peak over on 'war and politics' wasn't nearly so large before 9/11, at a time when the US audience was widely scorned as well-nigh isolationist in its interests. Many people's attentions were changed that day, perhaps permanently. This weekend and week, there's a large bump in the landscape where the topic 'Santa Anna Winds & Wildfires' takes up an annual residency for us Californians. Our landscape is not really static, it's in continuous motion with seismic shifts.

Such events are (mostly) exogenous to the media. There are other ways to draw attention, however, which are more interesting from the POV of analyzing strategies in an attention landscape such as music. We are a race of monkeys. Flashing lights and loud noise do work, they get our attention, tuned by ages of evolution to watch for the changes that mean threat or opportunity. Advertising, public relations, the very media themselves have all been designed and co-evolved with our baked-in responses to stimulus.

This introduces another type of media strategy. Unlike the abstract broadcast/niche/vanity models I described last time, these strategies do not work with the existing landscape, they seek to (re)make it. In the landscape metaphor, the scrapers and dozers can be called out to build a mound above the water, in the light of profitability. Rolling all that virtual construction gear costs real money, so if you really want a net profit, the peak you build had better be good size. This is the hit-driven model that Roji called 'directed.' Some years ago, there was a generic territory labeled 'adolescent female pop star' on the music-scape, but there was no peak upon it labeled 'Britney'. Now, some tens of millions of dollars of cash flow later, there is such a peak, largely an artificial construction.

That's not the only implied model. At the other extreme, there's (cynical take) trend-surfing / (populist take) 'paying your dues'. Both of these essentially consist of setting up camp where the landscape later happens to move upward, with a benefit that is largely not of one's own doing. I guess if you did it with malice and design aforethought you're an opportunist, but if you're just suffering from dumb luck / survivorship bias, you've paid your dues, and grown 'organically.'

Do you notice that I've just staked out the two ends of the music industry power law curve? The up-from-the bottom micropublishing model? The high end hit-driven RIAA territory? A few weeks ago I tried to provoke a discussion on attacking the high peaks in the music-scape using the Internet as a platform, at least in part to avoid some of the more starry-eyed populist assumptions plaguing the micropublishing discussions. I think it's succeeded in part, thanks to a variety of folks including Kevin Laws and Roji. At any rate it's opened my eyes to both the real structural problems of incumbent labels, and the actual difficulty of assailing their core franchise, even in its decay. Being 'not evil' doesn't help too much when charging entrenchments on peaks.

Now it's time to drop the pose. I actually believe the most interesting territory is in the mid-market, both in music and other media. The terrain of foothills and coast ranges, newly exposed as the Internet drains out the water of fixed costs. We may also surmise that the 'net reduces the costs of building new hills, as word-of-mouth and focused advertising augment or supplant conventional PR and promotion. To borrow an ecological metaphor, there are a lot of niches being opened as the water drops. Many that the BigCo peak dwellers may be ill-adapted to exploit. That's where I want to go next.

This whole metaphoric exercise arose from several weeks of frustration trying to bend the power law framework enough to talk about the mid-market, and eventually chucking it in frustration. I'm not trying to suggest the power law is fallacious. To the contrary, it is demonstrable. But it's just not very useful in talking about business models that rely on teasing apart distinct areas of interest, rather than lumping them. I'll be back around to try that exercise when time permits.
1:49:58 PM