I was the project lead on the IBM Microkernel. I did the technical analysis that was used to convince Taligent to use IBM’s Microkernel in place of their own internally-developed Opus microkernel. I subsequently moved to Taligent to work on the low-level tasking, synchronization, and memory management facilities for the Mach-based TalOS. I have a unique perspective on the failure of Taligent since I was a low-level player on both sides of the IBM-Taligent Microkernel struggles at one time or another. (Caveat: That doesn't mean anything I say is correct. These are all just personal opinions and observations, nothing more.)

Here a number of things that come to mind:

1. Joint ventures are hard. Apple and IBM entered into the deal, each with their own set of priorities and they were not necessarily aligned. Apple seemed to treat Taligent with benign neglect. IBM tended to give it too much attention at times. I'm sure that trying to keep the parents happy and engaged took considerable executive attention away from running Taligent as a viable independent entity.

2. IBM didn't really want a separate Taligent Operating System. The person in charge of IBM's Taligent strategy was also in charge of IBM's Workplace OS strategy, which was an IBM Microkernel (Mach) based OS that could run OS/2, AIX, and DOS personalities (VMs). A Pink personality was certainly given lip service to, but as time went on, it seemed that IBM was only interested in propagating Workplace OS with OS/2. In the months prior to Taligent abandoning its operating system, IBM made a play to force the TalOS group to switch to just writing device drivers for IBM's Workplace OS. They didn't succeed. And it was only a matter of time before TalOS was abandoned. I'm not privy to what was discussed in the C-level suite, but I cannot imagine IBM was wishing Taligent well with its OS.

3. What Taligent was attempting to do was unprecedented. It's class dependency diagrams were circular and hellacious. There was an architectural attitude that OO changed the rules and the type of layering of services used in other systems was out-of-date thinking. The issues created by these circular dependencies were serious and made startup and shutdown of the system impossible without some restructuring. Though we got the lower-levels of the system working on the IBM Microkernel and the upper layers (what would later be known as CommonPoint) were working on top of a compatibility layer on AIX, we never got TalOS put together as one piece.

4. The OO-based I/O driver model never got completed. This was partially due to political pressure from IBM trying to get us to drop the model and just adopt Workplace OS procedural drivers (essentially making Pink a personality on Workplace OS) and partially due to it just being a hard problem that needed more time to implement than the inter-company political runway allowed. That said, the I/O group did a remarkable job of getting parts of it working, even after the decision was made to kill the OS.

5. Worse than pressures from IBM over the direction of the OS work and the lagging schedule, Taligent was unable to find much demand in the market for a new OS. People loved the value that the upper layers of Pink would offer, but it seemed no one loved the idea of yet another operating system. With neither Apple nor IBM champing at the bit to sell TalOS-based systems, there was little reason to continue working on the OS.

Less than a year after Taligent dropped the OS to concentrate on CommonPoint, I left the company because my interest was in low-level systems. It was just too disheartening for me to stay.

Here are some of the reasons I think Taligent was unable to make CommonPoint successful:

1. The complexity of the frameworks. We used to joke that the frameworks were so powerful that you could write any program in three lines of code, but it would take you 6 months to figure out what those three lines were.

2. Changing priorities of Apple, IBM (and by that time HP) caused shifts in company outlook and direction.

3. Shortly after I left Taligent, the first CEO, Joe Guglielmi resigned. His replacement was Dick Guarino from IBM. Unfortunately, Dick Guarino died of a heart attack very soon after taking over. The instability at the top may have influenced how Taligent's parents viewed the company. But, I suspect the writing was on the wall with Joe G's departure. By spring of 1996, IBM had bought out Apple and HP, leaving Taligent as a wholly-owned subsidiary of IBM. At that point, I suspect IBM was merely looking to use whatever Taligent technology they liked in other IBM products.

Could Taligent have made CommonPoint a success had they had sufficient time and autonomy? That's a question that will never be answered.