walkitout (walkitout) wrote,

A few remarks about predictions

I've said a variety of things about predictions (laughing about them when they are wrong, tracking pundit predictions to get a sense of how seriously to take their future predictions, the easiest prediction to make is more-of-the-same, etc.). But I'm not sure I've posted anything about predictions that turn out to be right, in an amazingly remunerative and useful and important way, that in no way benefits the person who made the original prediction, mostly because their model for the prediction coming true was wildly wrong.

Here's a specific example.

When I was in college getting an undergraduate degree in computer science (over 20 years ago), my fellow computer science students (and others as well) participated in the usual bull sessions of college days. Some of these debates had been engaged in so many times they were really painfully boring and we developed idioms for avoiding getting sucked back into them (e.g. which is better: vi or emacs), or some of us did anyway, after a few years. Some of the debates were more interesting: when will ebooks replace pbooks (<-- not the terminology used at the time), and fun to watch play out (mostly because I was _really really right_ about it taking more than a few years to happen). And some of the debates weren't really arguments with each other, but trying to figure out how to make a desired goal happen. I would put Unix Taking Over the World into that category.

Our operating system choices in those days were varied but almost universally proprietary. A full open source operating system was still a distant dream, but people were rabidly backing the Hurd even then. I had friends who started a bespoke software company to develp for the NeXT (and are still in business doing quite well, altho obviously not developing for NeXT any more). While I had a couple buddies who genuinely loved PCs and went off to work at Microsoft, most of my friends were die-hard Unix fans of one flavor or another, and if you knew what to say, you could put a nickel in the Unix Will Take Over the World jukebox and sit back with a sandwich for a half hour of mild entertainment.

I'd like to take this moment to point out that they were all more or less right: Unix is indeed taking over the world, whether as the basis for QNX over at RIM, webOS at HP, Android in any number of places, or iOS and OS X over at Apple. True, neither Nokia's OS nor the two major variants of Windows Mobile OSs (much less Windows itself in any incarnation) are based on Unix. But I think the proportionate market representation and trends suggest that Unix Has Taken Over the World.

But it doesn't look anything like what any of us thought, back in the late 1980s and early 1990s, Unix Taking Over the World Would Look Like. Specifically, a lot of what we, as baby programmers, liked about Unix is mostly if not completely absent from the experience of all these Unix systems, for virtually all end users. On the other hand, the devices do actually work pretty great, which was an important part of our argument for why this should occur.

To the extent that any of us made money off of believing this would happen before it did, we failed to make other money -- that is, we didn't go work for Microsoft, but started other companies or worked for other companies or both. And I think that's a really important lesson about predictions.

(1) You can be right, and barely recognize it when you are (and not just because you forgot the prediction you made).
(2) Even if you are right, and even if you benefit from being right, you might have benefited as much or more by being wrong, or by making choices as if you were going to be wrong.

Visualizing the future in an optimal way is an intractable problem, but the effort is interesting and rewarding.

  • Post a new comment


    default userpic

    Your reply will be screened

    Your IP address will be recorded 

    When you submit the form an invisible reCAPTCHA check will be performed.
    You must follow the Privacy Policy and Google Terms of use.