By Jeff Gould (Peerstone Research / Interop News)
Several readers weighed in with interesting comments on my post last week about Microsoft's plans for Yahoo's open source software platforms.

Of course we now know that Yahoo officially doesn't want to be bought by Microsoft. But that's no surprise. By some mysterious law of human nature, founders and CEOs who are comfortably ensconced running vast but declining corporate empires rarely seem to recognize spontaneously that it is time for a change.

Reader Michael believes that Microsoft "HATES" open source so much they will get rid of it at any cost. But he contradicts himself in the next breath when he adds that the deal is "only about beating Google". That last part is indisputable bedrock truth. But surely it's also true that Microsoft hates losing more than it hates open source. After all, they're not insane up there. If Yahoo's open source can help them beat Google, then they will use it.

Reader Andrew more plausibly suggests that Microsoft's bid is "all about the subscribers" and that Redmond will eventually dump Yahoo's code when they've converted everybody to the new system. In an ideal world I'm sure this is exactly what Microsoft would try to do, at least in those (fairly numerous) areas where Yahoo's performance against Google seems to be just as pitiful as Microsoft's.

But in the real world it may not be so simple. Wall Street Journal reporter Kara Swisher reiterates in her blog the investment community's view that Microsoft will try to replace Yahoo's search and ad platforms with its own. This would be an obvious win for them, and it seems technically feasible at least in principle, because most of the changes would occur behind the scenes. But I would advance against this easy-as-pie view the unpleasant reality that in large software projects simple ideas are often shockingly hard to implement and can result with surprising frequency in catastrophic failures. Big software ain't as easy as it looks.

Even if Microsoft does manage to swap out the search engine algorithms and the ad management tools, there is a lot more to Yahoo than those two pieces. There is also the whole vast publishing platform, with all of its countless moving pieces and intricate imbrication of code and content. There is a huge amount of open source code embedded in this platform, and changing it would be a real, um, female dog. It's a well-known fact in online publishing circles that any time you change your platform you are likely to take a big though perhaps only temporary hit in your audience numbers. I strongly suspect Steve Ballmer will not want to take undue risks in this area, especially if the payoff seems less than spectacular.

Reader "Just me" agrees that Microsoft may want to exploit Yahoo's open source code, but anticipates a Machiavellian move to "privatize" it by shifting it to Microsoft's "shared source" license. Keep in mind though that the most valuable pieces of Yahoo's software are the already proprietary and closed source algorithms written in C or C++ that are invoked from open source middleware such as PHP and MySQL. Some of the mechanisms behind this are described in Yahoo PHP guru Michael Radwin's 2005 talk outlining Yahoo's open source stack. As for the Hadoop parallel computing middleware I described in my post, which is Yahoo's open source answer to Google's MapReduce, it is an Apache project and thus subject to the terms of the Apache license, which does indeed permit "privatization" of open source code. So reader Michael may have a point there.

This whole Microsoft-Yahoo caper calls to mind nothing so much as the recent string of storm-tossed hostile take-overs Larry Ellison has piloted at Oracle. In every case a nearly identical scenario played out. The CEOs of PeopleSoft, Siebel and BEA ducked and dodged, feinted and bluffed, and generally ran around in circles trying to stir up rival bids or anti-trust trouble. In the end they all failed, and their companies were swallowed up, to loud cries by the press and the pundits that Oracle would soon sweep away the acquired code and replace it with its own. Clients and the engineering teams at the acquired companies trembled at the thought, and prepared for the worst. But funnily enough the worst never came. Oracle soon discovered that it simply could not forcibly migrate the purchased companies' code to its own "Fusion" middleware platform. To do so would have courted technical disaster and commercial suicide. So they backed off. Today Oracle is devoting hundreds of millions of dollars to upgrading PeopleSoft and Siebel applications, and will undoubtedly soon be granting pride-of-place in its middleware catalog to BEA's WebLogic.

Hmm, I wonder if Microsoft isn't heading for a similar "blended code" scenario with Yahoo?