« Back to home

The worst mergers and acquisitions

ZDNet ran a story in the week before Halloween about the worst tech mergers and acquisitions. Here are my thoughts on some of them.

Caldera and SCO

Xenix was the first Unix I ever used. It’s how I ended up using the C Shell — I remember the Xenix box as only having csh, though it’s possible that it also had a really feature-poor version of sh. Either way, csh was the only sane option for interactive use.

When I got MINIX running at home, I stuck with csh. On real Unix, I switched to tcsh. Eventually Bash got support for Unicode, which was a compelling enough reason to switch to the Bourne family of shells. Then once Zsh got Unicode, I finally completed my destined path.

The less said about SCO’s business ventures after the Caldera acquisition, the better.


Not mentioned in the article is that Palm OS was modeled on classic Mac OS, and the hardware used low power variants of the same 680×0 series processors that 1980s Macs were built on. PalmOS had resource forks, four character file type and creator IDs — and no memory protection or virtual memory. This meant that Palm OS started to hit a technological wall a few years after the first PalmPilot; a single bad application could crash the entire device. As with classic Mac OS, there was no plausible technical way forward without basically starting from scratch, and by the end of the 90s it was clear to me that PalmOS was a dead end.

I think WebOS could have succeeded, if HP had stuck with it. The problem was, like the Apple Newton MessagePad, the software wasn’t quite ready by the time the hardware launched, and the hardware wasn’t quite the form factor people wanted. Better market research and a couple more iterations and I think they could have made it a going concern. I went with a BlackBerry instead, and watched the train wreck from a safe distance. (And then watched BlackBerry fail to deal with their own technological wall, and jumped again to Android.)

As an aside, I always expected Apple to sue Palm for patent or copyright violation.

Oracle and Sun

I remember when Sun was looking for suitors, seeing employees saying “Anyone but IBM”. I suspect that hindsight may have caused them to re-evaluate that sentiment. Obviously I’m biased, but I think that IBM buying Sun would have been much better for almost all of Sun’s products.

Clearly OpenOffice would have lived for longer. In fact, IBM did its best to keep OpenOffice a going concern even while Oracle was busy killing its community. MySQL, too, would have been OK — IBM has long experience at maintaining multiple competing databases and supporting them all, and I think MySQL would have just been slotted into the portfolio alongside Informix, DB2, DB2 on i, Derby and IMS. Netbeans and Eclipse probably would have had a cage match, and Eclipse probably would have won out of inertia, but Netbeans features would have been rolled into Eclipse — and Java would have stayed an open technology.

The one Sun product I think would have had trouble at IBM is Solaris. AIX vs Solaris was still serious warfare at that point. Now, Linux is where it’s all happening (at least from my viewpoint), and I think Solaris would have ended up sidelined even if it had beaten out AIX for its new master’s affections. Still, Linux would have gotten ZFS and dtrace and the other things from Solaris that people actually like, and maybe SMF instead of systemd.

HP and Compaq

HP’s real problem was partnering with Intel to produce its new RISC-replacement. Given Intel’s history, they really should have known better. They then dumped the working Alpha CPU they had acquired by buying DEC, deciding to wait for Intel’s promised Itanic — and decided to let Intel handle all the design itself.

I guess that’s pretty much standard Carly Fiorina era HP, though — treat engineering as an expense you should avoid as much as possible, and make business decisions without their input.

Microsoft and Danger

The Danger Hiptop was a great product from a hardware and user experience standpoint, but when I discovered that it was entirely dependent on back-end cloud servers I steered clear and went with a BlackBerry instead. I don’t mind cloud services, but the product has to be able to work standalone. When Microsoft bought Danger I knew they were doomed. You could put together an entire article of disasters just by cataloging companies that have developed an intimate relationship with Microsoft: Microsoft and Vivo Software, Microsoft and RealNetworks, Microsoft and Kodak, Microsoft and Nokia, Microsoft and Groove Networks, Microsoft and Oddworld Inhabitants, Microsoft and Bungie, Microsoft and FASA Interactive, Microsoft and Connectix, Microsoft and Rare, Microsoft and WebTV…

Borland and Ashton-Tate

Borland had two big problems which hurt sales of their compiler products: their languages were non-standard, and they were single platform.

Turbo Pascal was great, but it wasn’t standard ISO Pascal. Instead of migrating to (say) Modula-3, they added proprietary extensions to Pascal for OOP. They had C++, but they used their own layer of windowing APIs on top of Microsoft’s, which led to the horror of Borland Buttons.

So if you wrote your code for Borland C++ or Pascal, you were forever tied to Windows. That meant that once Microsoft had a workable set of development tools, Borland’s target market was people who loved Microsoft enough to want to use only Microsoft Windows as their OS, but didn’t like Microsoft enough to use their development tools. That market was approximately 0 people.

Novell and Unix

Novell’s problem was wasting its time trying to keep IPX alive and trying to be a Unix vendor. Once dial-up Internet was a thing and TCP/IP was usable even in small offices of Windows machines, IPX was dead. Windows for Workgroups in 1993 showed the future. If Novell had made a user-friendly fileserver OS by adding full TCP/IP services to NetWare, they could have had a sellable product; but practically nobody who used NetWare wanted to switch to Unix, and nobody who wanted to use Unix had any compelling reason to get it from Novell.

Google and Motorola

It was patently apparent, if you’ll pardon the pun, that Google was only interested in Motorola in order to use its ‘intellectual property’ as a defensive resource in the litigation wars started by Apple. I don’t call this one a failure. Motorola even produced some great handsets during the Google period — look at the Moto X and Moto G.

What has been a disaster, is Motorola since Lenovo took over.

HP and Autonomy

Autonomy had some really neat software when I saw demos around 2000. Sure, it was all based on Bayesian inference, but it was useful stuff. Other than HP paying too much, the real problem with this merger is that HP has never been any good at selling software.

Nokia and Microsoft

Nokia made a lot of mistakes in the 1990s. Like Palm and BlackBerry, they ignored their OS platform’s technical shortcomings. Like HP with WebOS, they made poor hardware design decisions.

However, by the early 2000s Nokia was heading in the direction of a Linux-based handset OS developed in an open source fashion, and they were about to start moving to Qt for their UI to make development and porting from other platforms easier. Maemo was almost at the point where it would have been workable for phones, it just needed a better touch UI, which Qt would likely have given it one.

It’s pretty clear to me that Microsoft wanted to kill Maemo before it was too late, and obtain Nokia’s patents so they could use them against Android. Elop may deny it, but I think this is another outwardly disastrous merger that went exactly the way it was intended to.

I mean, Microsoft had been trying to sell Windows-based mobile phones since 2000 without major success (vertical markets excepted). It makes no sense that Elop would kill Maemo and Symbian and bet the company on Windows Phone — until you realize that he joined Nokia from Microsoft.

Not convinced it went according to plan? Consider that it was Elop who made the decision to sell Nokia to Microsoft once its value had dropped by 85%, and that he got a $25m reward when that transaction was completed — with most of the cash coming from Microsoft.

Microsoft gave Elop the Xbox division to play with for a year, but he was invited to accept a horizontal promotion as soon the media had forgotten about him. They quietly killed Windows Phone, and have settled into their primary mobile strategy of collecting patent licensing fees on Android devices and forcing Android device manufacturers to bundle Office.