From a business perspective upgrades of operating systems on existing non-enterprise PCs is non-material to Microsoft. Nearly all Windows revenue comes from sales to OEMs and from Enterprise Agreements/Software Assurance. This frees them to play around quite a bit with the pricing model for retail upgrades. The actual difficulty on the upgrade front is that if all upgrades are free, or super low cost, how do you get Enterprises to purchase Software Assurance? I think we saw one approach with the Windows 8 launch, you make cheap upgrades available for a limited time window (thus capturing the bulk of people who really care) and then return to more traditional pricing. But there could be other ways.
Imagine that 3 years of a variant of Software Assurance are included with every OEM and retail copy of WIndows. So any updates, including new versions, of WIndows that ship within 3 years of your device purchase are free. After three years you must pay for new versions. This very much would approximate Microsoft's current revenue cycles while adapting them to the demands of the consumer market. Alternating "Updates" and "New Releases", as Ian Easson also mentioned, is another option though I don't think I'd do it quite as mechanically as that. I'd base declaring something a New Release rather than an Update depending on some key new capabilities.
But this article wasn't about the revenue situation, it was about how Microsoft has a marketing rythm that is driven by infrequent big honking releases. After decades of operating this way will they be able to generate excitement with an annual release cycle that by definition offers a more gradual evolution of the product?
A likely consequence of the new release strategy is that each release will actually contain a larger number of user-visible improvements, and particularly small "delighters", than its traditional big releases. The big releases often get focused on major infrastructure upgrades with the user-visible improvements lost both in the planning process and then those that make it into WIndows get lost in the messaging. On the former, imagine a planning session where the networking team decides it can't do customer request #7 because it has to rewrite the entire network stack to support IPv666. And even if it does both then which is going to get all the attention? It's hard for anyone to notice feature #7 when the entire networking world is going to hell.
Microsoft's ability to make noise every year rather than every three years is, in my mind, much more positive of an opportunity than the risk exposed by the author. There are probably 10 rather small but significant changes to WIndows 8 that would significantly move the experience forward, generate positive buzz from pundits, and feed user lust IF they are available within a year of the original release. If you take those same items and leave them for three years then the typical case is that no one cares. In the worst case they become negatives ("It's about time Microsoft, you should have done this three years ago") as reminders that Microsoft wasn't reacting quickly enough to its users' needs.
So there is great opportunity for Microsoft to actually generate bigger and better buzz around annual release cycles than the three year monsters. The real question is, can their marketing organizations adapt to this and take full advantage of it?