A nice essay generalizing about what people might be willing to pay for, in a world where the marginal cost of production of products that people used to pay for has dropped to zero.
One category that he missed is embedding digitized products like software into dedicated hardware. It's a form of DRM that is harder to crack (I think).
Here's some examples from the audio world, that are variations of this idea.
The main audio software platforms, known as Digital Audio Workstations (DAW) have all evolved to a point where they have to support open plugin formats. Plugins are implementations of digital signal processing software, that are used in combination within the DAW to produce the end result, which is a complete, finished sound file,
Because audio production and engineering is hard (basically, things tend to not sound good) there is constant development, and fierce competition, in this niche market.
Various forms of DRM, or licensing systems, are almost universally used. This provides enough friction, meaning you can get cracked versions of most plugins, but it comes at a cost of inconvenience, malware, or compromised stability, that a modest number of small companies have built business in the market.
But the competition is fierce, and the trend in license prices has been steadily down. The cracks do hurt the sales.
One company that has thrived in this market is Universal Audio. They put heavy development into making premium, we'll respected plugins, but they only run on their proprietary DSP systems. For a while, this could be seen as a genuine advantage, as users commonly ran up against the limitations of their CPUs.
This is no longer the case, but the company has steadfastly stuck to their proprietary system. One technique they used was to embed their DSP in dedicated sound interfaces.
The sound interface market is also hotly contested, and companies are constantly fighting against commoditization. So they developed high quality sound interfaces, which is something all audio producers have to have, and use their catalog of exclusive software plugins as a "value added differentiator."
CEDAR is the pre-eminent developer of specialized software dedicated to challenging issues of noise reduction. For a long time, they limited the use of their algorithms to their own DSP hardware. If you wanted these industry best algorithms, you had to buy their, relatively, expensive systems. While they now do offer some of their software as plugins, they continue to use dedicated hardware as part of their product strategy.
One interesting possibility is the embedding of otherwise unremarkable software into dedicated hardware, because of the user interface advantages. By giving the user access to physical controls, that do nothing but mimic their virtual cousins, the goal would be to dramatically increase the usability of the software. There has been some movement in this direction, which is actually a kind of throwback to how the first generations of audio DSP devices, back when dedicated hardware was the only way to implement such processing.
You can see this tension around user interface play out in the realm of audio mixers for life performance. They use dedicated interfaces to run the real time DSP, but combine various virtualization strategies. Some of the biggest audio plugin companies, like Waves, have released versions of their popular software plugins to run on some of the modern live mixing hardware systems, thereby generating new revenue streams from existing products.
At this point, while it is entirely possible to run an entire mix of a live show on a PC with a mouse and keyboard, it is such a sub-optimal user experience, that I have never witnessed anyone do this. (Though I'm sure some foolish, Braveheart do, and budget challenged audio engineers have done it!)