Volitional Science

9

Wednesday, October 13, 1999 - Ownership - the Answer to Systemantics

Okay. Enough theory.

Sometime in 1984, I think, my wife made a comment that something was screwed up because no one had a proprietary interest in fixing whatever-it-was. I don't even remember what we were talking about. But I remembered the idea of "proprietary interest."

This single idea has driven my entire career. Of course, she had gotten the idea from Galambos in one of his courses.

After I heard that idea, I started to look at every single project and process in terms of "who has a proprietary interest in making this work."

Do you want to know the ultimate answer - the fix for Systemantics? Systemantics, as you will recall, is a pseudo-science about systems and how they fail. But... some systems work. Big systems grow from smaller systems. Really complicated systems tend to degrade. Systems don't do what they are supposed to do. A system doesn't do what the system itself says it is going to do. And so on.

But some systems work. Simple systems often work. Systems often work when they are first created. Why is that?

How come the Unix operating system, which started small, and grew tremendously, and fragmented into many flavors, still works? Why is that?

What do all systems that work have in common with each other?

What do all systems that are dying or decaying have in common with each other?

Here, finally, is the answer.

Systems that work ... have an owner - a competent owner. If a system is working, then you can bet there is someone with a proprietary interest in keeping it alive.

Systems, in general, are not capable of automatically balancing themselves. Especially systems designed by human beings. Those systems are never complete. They require active control. That active control is provided by the owner. The owner is the person who controls the system and is interested in maintaining it. That includes tweaking it as the environment around it changes.

Let's look at some counter-examples. Why does the number of laws seem to grow without bound? Because nobody owns all the laws. Even the guy (or gal) who originally sponsored the law doesn't take ownership of the law. Nobody owns the process of cleaning up the system of laws. Sometimes, in a piecemeal fashion, the courts clean things up, but only when they are forced to.

Why do big operating systems tend to degrade? Because there are huge parts of them that nobody owns. Even if it is true that Bill Gates and the stockholders of Microsoft own Windows, I can tell you from personal experience that there are huge parts of Windows that nobody pays any attention to anymore.

I was in the beta for Windows 95. Everything was great. There were problems but they were being fixed. I was impressed. But I had a lot of concern about MIDI support. And apparently even after Windows 95 shipped everything was great. Then, one day, I had to reinstall Windows 95 on my MIDI machine. Suddenly, my machine could only enumerate up to 12 MIDI devics, whereas before it could do an unbounded number of devices. Sometime in the last few weeks of the beta, a contractor put in a stupid bit of code - a bug - that made Windows crash when it tried to get past the 12th device while booting. It turns out I had old beta code on the machine so I didn't catch the error. You can imagine I was quite annoyed.

I sent in lots of bugs about this since there were a number of service pack betas. You know - those OSR things. The bug never got fixed. I got told there was no plan to fix it. I said, "do you want me to buy a Mac?" That always gets 'em. Well, almost always. That didn't work either.

Finally, Windows 98 was coming. The bug was being ignored again. This was not acceptable to me! By now I was involved with Meltdown (a Microsoft hardware testing event) and I met more and more Microsoft people. I decided to find out who owns MIDI. It turns out it'd been orphaned! Well, not quite. It'd been handed off to the lady responsible for DirectInput! I decide to track her down. I finally found her at a Meltdown event. I said, "Hey, what's the deal with this MIDI bug? It's been there for years. This is our last chance [Windows 98] to fix it!" She said, "Yeah, you're right, we should get around to dealing with that legacy stuff." Legacy! I almost choked. She's the one that tells me about the contractor that put in the silly bug. She also says, "And guess what, he's back working with us, so he can fix it." One guy in the world can fix this bug and he's a contractor! I encourage her with profuse praise and tell her how incredibly grateful I'll be if she gets this thing fixed.

And she did fix it. She was the only person in all of Microsoft who could put the resources together to get this bug fixed. (She was a program manager - they can do things like that.) A few thousand people who wanted to play lots of MIDI keyboards weren't enough to raise this bug to a high-level of attention without her personal involvement. And any of you MIDI people out there using Cakewalk Pro or some such with a lot of MIDI devices can thank me personally for getting this fixed for you. In fact, the people that make the MIDI sequencers should thank me too! This OS bug was limiting the services they could provide.

My point is this - nobody really owned the MIDI part of MMSYSTEM at Microsoft. It turns out a lady sort-of owned it, but only because it was dumped into the DirectInput group for legacy support. In fact, you could say I owned this problem more than anyone else, because I had the greatest proprietary interest in getting it fixed! I sent e-mail to Twelve Tone Systems (now Cakewalk Systems) to try to get TTS to put pressure on Microsoft to fix it. I thought that maybe a sequencer company would put enough pressure on MS to get it fixed. It didn't work! I guess TTS had enough sales to people happy to live with 12 devices or less. The Microsoft program manager lady - who was very nice BTW - didn't play MIDI keyboards so she didn't have much of an interest. Fortunately she responded to my plea, for which I'm grateful, because I really didn't want to have to get a Mac and learn a new sequencer. (Macs used to cost $5,000.00 you know. But the learning curve is the real pain.)

Nobody owns DirectX Retained Mode anymore. It's been orphaned. In a way, I own it as much as anybody does, because I'm willing to rewrite the rendering loop!

These are typical examples. How about orphaned device drivers? Why do they stop working? Because nobody owns them and maintains them.

Who owns Unix? How come these fragmented versions continue to survive and even thrive? Because somebody owns each one. Sun owns Solaris. SGI used to own Irix. I suppose technically they still do - but is there anyone still there that understands even 10% of it? HP owned HP-UX. More importantly, there were product groups with managers that had a proprietary interest - their career interest - in keeping these things alive and responsive to customer demand. Who owns Linux? That's an interesting question. I think Linus Torvald does. He owns it because he was the one that wrote it originally and slapped the license on it and made it available. But more importantly, he still owns it because he drives the quality and timeliness of the integration of new features. Even though it seems like no one owns it, because of the sort-of public nature of it, in fact, Linus owns it. Gates owns Microsoft. Even though he really owns 25% and not the whole thing, let's face it, he "owns" it. He drives that place. Ballmer is a good helper. But Ballmer doesn't own it the way Gates does. Microsoft, like Digital, and IBM, will start to fall apart once no one has that incredible proprietary interest in it that Gates has.

(BTW, my idea is that Gates should break up the company on his own. Not for any anti-trust reasons, but because he can avoid the massive infighting that will result sometime in the future. By dividing the company up now while he's in control and giving significant proprietary interests to the managers of the various divisions he can avoid all that fragmentation. Then maybe Microsoft will avoid the "degeneration syndrome" that hits companies once their strong-willed founders leave. He'll still get to own 25% of each part, but he can shift control to the operating managers who will have autonomy and control and accountability. And by breaking it up, these managers won't raid each other for resources and power. It's a good idea. He should be a pioneer and do it.)

So, to revisit why I knew Virgin Interactive was going to die after Robert Devereux stopped running the place - there was no one set up to have the same proprietary interest in the financial success of the company that he had. Robert was the guy who had to get all money that was invested from his brother-in-law Richard Branson, had to justify everything, and kept tabs on every penny. It was their money, literally, and not a bunch of paper-stock money from a big corporation (Viacom). He had invested in me, and knew what was going on. The new owners had no idea the extent of the contribution that Robert made. How could they? Do you think Virgin Communications was going to tell Viacom/Blockbuster/Spelling, "hey, guess what, you're getting a different company than the one we had!" Not likely.

And later when Brett Sperry had the responsibility but not enough authority to revive non-Westwood development. Same thing. Somebody at Spelling 'owned' the company but didn't have the intelligence or experience to understand what they owned. Technically, they 'owned' it. But in fact, nobody had that driving proprietary interest. Brett, who was supposed to 'own' development, didn't really own it, because he had to go through the mire of Spelling bureaucracy. The company was orphaned.

When you're trying to figure out why a process or system is working or not working, you need to invoke Galambos' Razor and ask the question - who owns it? Is it the right owner for the job? Does the owner have the resources and authority and competence necessary to do the job? If they don't, they don't really own it.

Why do new systems tend to work? Because the person that created the system is still hanging around making it work. Why do simple systems work? Because it is easier to assign ownership to something that is easy to understand. Complicated, convoluted systems with overlapping responsibilities soon find themselves without owners who can tweak them and keep them healthy. Those systems start into the horrible cycle of slow death.

(The original source of these ideas is Galambos, but this column has been filtered through my real-life experience working for both entreprenerial [Virgin, for example] and statist-sponsored companies [Northrop, for example].)


Next Volitional Science Article


Back to 'Random Blts' Table of Contents


Back to Above the Garage Productions