Horace Dediu — the Harvard-trained analyst who writes the influential Asymco blog — was studying the mobile phone market for Nokia (NOK) in 2005 when Google (GOOG) bought Android, primarily as a defense against the perceived threat that Microsoft (MSFT) was about to do to cellular telephony what it did to desktop computing.

The real threat, it turned out, was Apple’s (AAPL) iPhone, which came along two years later from a different direction.

That’s the context, Dediu maintains, in which to try to make sense of Google’s proposed acquisition of Motorola Mobility (MMI).

Dediue took a crack at making sense of the deal in two venues Tuesday — in an article published by the
Harvard Business Review
and in a 50-minute “Critical Path” podcast on the 5by5 Network. Together, the two pieces constitute the deepest dive we’ve seen — and we’ve seen plenty — into the meaning of the $12.5 billion merger.

Dediu’s conclusion: The deal doesn’t make sense. Not as a pure patent play. Not as an entry into cell phone manufacturing. And not as a combination of the two.

“The big secret here,” he concludes in his Critical Path piece, “is that I don’t think Google knows what it wants to do yet.”

Google’s original strategy, he says, was to disrupt Microsoft by giving away what Redmond was selling — a mobile phone operating system — and making money by providing online services (primarily advertising).

But Google was still using Microsoft’s modular approach, treating the OS, devices, and services as three separate businesses. Unfortunately for the licensees of Microsoft Windows Mobile, that made for an unbearably clunky smartphone.

Apple’s breakthrough — like Research in Motion’s (RIMM) before it — was to build an integrated device that offered users a smooth, seamless experience.

So Google went back to the drawing board and redesigned Android to mimic the iPhone’s look and feel. But it was still going modular — not integrated like Apple. It was giving away the OS, letting others — Samsung, LG, HTC, Motorola etc. — build the devices that ran it, and hoping to somehow make money selling mobile services.

The money in mobile services hasn’t yet materialized and the modular approach opened several “cracks” in the Android strategy that Dediu enumerates in the Harvard Business Review:

“Issues with intellectual property in Android caused some licensees to have to pay royalties to patent holders, increasing the cost. Fragmentation took hold where some versions of the software were used by some licensees on some products without the option or incentive to upgrade. Finally, some vendors modified the software resulting in missing features or inconsistent user experiences — even to the extent that Google’s own services were omitted.”

The 24,000 or so patents and patents pending that Google would acquire if the Motorola deal goes through might help protect its licensees on the intellectual property front — although as FOSS Patents‘ Florian Mueller points out owning those patents didn’t protect Motorola from getting violently sued by Apple and Microsoft.

Trying to compete with its licensees by building its own smartphones, however, creates even bigger problems — as Nokia learned when its Symbian partners eventually abandoned that ship.

Dediu concludes that the Motorola deal was hastily put together and not very well thought out. It’s “reactionary, not proactive,” he says. “It’s playing chess not looking five moves ahead but looking one move ahead, which is not a very good way to play the game.”

The break-up cost of the proposed acquisition — the price Google would have to pay Motorola if it falls through — is an extraordinary $2.5 billion, which suggests that someone knew that the deal had its risks.

In the end, Dediu repeats doubts he has expressed before about how committed Google management really is to Android’s long-term prospects.

“Android,” he says at one point in his 5by5 podcast. “could cease to exist within two years.”

We’ll see.