Keil Hubert: Prognostication Part 2 (Of Seers and Siri)

It’s often difficult explaining to upper management that you can’t accurately predict the future. Business Technology’s resident U.S. blogger Keil Hubert offers you a metaphor that seems to effectively get the point across.


This last Monday, I wrote about difficult it is for IT folks to predict the future. The subject of new IT solutions deployments came up … indirectly … at the dinner table yesterday during our traditional Thanksgiving holiday get-together. Really. We’re that kind of family.

Here in the USA, Thanksgiving takes place on the last Thursday of November every year. Our Christmas buying season kicks off at midnight that night, as many retailers open their doors ridiculously early on what’s come to be known as ‘Black Friday.’ I personally loathe the Black Friday tradition and avoid it at all costs. For some members of my family, though, it’s a fun and different way to get out and enjoy shopping … As for me, I can’t imagine how being thrust into a scrum of screeching consumers in the middle of the night is ‘fun.’ [1]

Anyway, the topic of ‘good’ Black Friday deals came up. My darling wife asked if there were any compelling sales that she might want to try and take advantage of. Our local newspaper is always stuffed to bursting with sales circulars on Thanksgiving, and the top flier on the stack showed a reasonably good price on a digital music player. That got me to thinking …

This time last year, I wrote about how critical it is for IT leaders to learn how to speak to non-IT people in terms of similes and metaphors that communicate a functional model in terms that the non-techie can wrap his or her head around. Seeing that sales flier triggered a cascade of associations for me, and it got me to thinking that I could probably explain the systems sustainability prediction problem from this Monday’s column in terms of music and music players.

I know it sounds ridiculous. Bear with me.

I know it sounds ridiculous. Bear with me.

Ask your executive to imagine that they’re a runner (it doesn’t matter if they really are one). The problem that you want to solve is you want to run a race and achieve a certain time over a certain distance. So far, setting your own pace hasn’t been working—you don’t notice yourself slowing down as you tire. [2]

A decent solution to the running pace problem is to use an iPod to maintain a constant pace. I’ve done this a lot; you listen to a song with a consistent beat for the duration of the race, and match your stride to the beat from start to finish. If you choose the right song and follow the beat for the entire race, you’re highly likely to hit your desired time. That works for me. It works for many of my runner friends, too.

Ideally, our technical solution will be to pick the artist and the song to play on the iPod that hits exactly the right track length and exactly the right beat to help you make your desired time. The great thing about playing music on an iPod is that once you find the right tune, you can play it many times and will always receive the exact same result.

At this point, even your most non-technical (and non-runner) executives should be nodding along. This example is pretty simple, and resonates with most everyone.

Here’s where the example needs to get a little meta, for education’s sake: explain that this is how tech solutions are often phrased: if you’ll buy this song and play it on this player while you run on this road for this distance under these weather conditions, then you’ll always achieve your desired outcome. In information systems designs, the pitch goes essentially the same way: if you buy this server and load this operating system and apply these patches and load this application with these settings, then these clients running this OS with this browser will always deliver your desired business process result.

The sad reality of IT deployments is that something that works once cannot be guaranteed to work that same way again because there are many variations on any given product, and products constantly change.

This started life as an iPad, but the requester kept changing their list of desired features.

This started life as an iPad, but the requester kept changing their list of desired features.

For the sake of our runner’s pace metaphor, let’s say that you find the exact song that you want – the one that’s perfect for your race. The trouble is, (for the sake of our example), your iPod is doesn’t have a visual display, so you have to use a speech-to-text function. You use something like Apple’s Siri and order the iPod to pull your tune up for you right as you start to run.

That’s a bit of problem if the song you picked was popular enough to have been recorded by multiple artists. When you tell Siri to play a specific song by name, you’ll get the song that you want – but not necessarily the version that you wanted. Every time that you tell your iPod to play the song you want and start running, you don’t know if you’ll get the original studio recording by the original artist in the format that you wanted … or if you’ll get a random variation on the original:

  • An ‘extended dance remix’ version of the song by the same artist that lasts twice as long
  • A live concert version where the artist included an impromptu drum solo
  • A different artist’s cover of the song in a completely different musical style
  • A snippet of the original sampled by a hip-hop artist to make a completely different song
  • A frenetic speed-metal interpretation that keeps the tune and lyrics at twice the speed
  • A languid version repurposed as a child’s lullaby performed with nothing but cellos
  • An avant-garde prog rock ‘interpretation’ that breaks all the standard music conventions for the sake of some higher ‘art’ [3]

That might make your run quite … difficult. You want a catchy pop song that consistently delivers a certain number of beats per minute, and instead get an emo version of the same track that not only comes in at a fraction of the original’s BPM, but also changes the time signature unexpectedly halfway through the track! Oy! [4]

'What kind of lunatic reinterprets "Amazing Grace" as a trash metal anthem?!?'

‘What kind of lunatic reinterprets “Amazing Grace” as a trash metal anthem?!?’

For the sake of our metaphor, the fact that you can’t perceive what Siri is ‘thinking’ when you tell her to pull up your track represents the problem that we have in IT when we rely on outside manufacturers, developers, and integrators to provide us components for a new IT solution. We know what we want, and we think that we’re being completely clear in what we ask for, but what we get is often an unwelcome deviation from what we wanted … and we have to make the best of whatever we get.

From the executive’s perspective, deploying new tech solutions should be easy; just choose the right pieces and put them together … From the IT team’s perspective, it’s nowhere near that straightforward since we can’t predict the future, and we can’t always know what we’re getting when we deploy something that doesn’t already exist in the enterprise.

Further, IT folks are never quite sure how a deployed solution will change over time as different elements of the solution change to suit evolving tastes. There’s a very good chance that changes to the original, known-good, approved solution may or may not get the job done in the future. Every variable that IT tries to control for in a solution represents a new cost that might or might not turn out to be necessary, and might or not turn out to be cost-effective. Further, the product’s developers (who are analogous to an artists that performs a song) are constantly reinterpreting their products to reflect evolving tastes. Just look at how the Microsoft Office team randomly changes their entire user interface every few years for no good reason. Or how the Bee Gees changed from psychedelic folk rock to disco.

I realize that this iPod example isn’t a perfect metaphor; it’s not meant to be perfect. It’s just meant to get an executive’s heads wrapped around the idea that tech solutions are largely unpredictable, and that IT is hard pressed to guess how a given technical solution will perform (and how it’ll likely evolve unpredictably after it’s implemented).

Just like people, technologies often start out adorably cute and grow up to be shockingly different from what you expected.

Just like people, technologies often start out adorably cute and grow up to be shockingly different from what you expected.

To be sure, we can isolate a specific, working version of a solution in the lab and prove that it works right here, right now, under current conditions. We cannot, however, guarantee how long our recommended solution will keep working … and the longer that we keep it in production, the more likely it is that elements of our solution will change until it the whole simply becomes non-viable. [5]

The core takeaway of this message is that we IT folks do our utmost to satisfy business needs and we strive earnestly to control for future environmental changes, but we’re not infallible oracles.

If that metaphor isn’t enough to get the point across, make it clear that if we could accurately predict the future, we never would have wasted company resources deploying solutions like MySpace pages, BlackBerry handsets, or OS/2 ‘Warp.’ Just like disco, many reasonable IT solutions seemed great when they were new and original, but failed to hold up over time … and now we don’t want to see or hear any trace of them around the office. Those were ‘bad’ decisions only in retrospect; at the time, everyone seemed to think that there were a great idea.

[1] I’ve never been comfortable being in large crowds. Perhaps that’s why I took so quickly to the science of riot suppression.

[2] Most people can relate to this idea, even if they haven’t run competitively since kindergarten. It’s a human universal.

[3] As Clark over at put it, ‘There’s an old joke about prog rock: “it’s the only musical genre where 23:17 could be either the time signature or the track length.”’ I never get tired of that joke.

[4] Yes, I’ve had this happen while running. If you can imagine a person rhythmically pounding down a quiet street in the pre-dawn darkness, try to visualize that same runner suddenly spasming as if they’d stepped on a downed electrical line. It looked (and felt) like that.

[5] The problem is magnified when you venture into custom or proprietary technical solutions. In terms of music, a soulless pop single that was cynically engineered in a major studio to have shallow, widespread appeal is likely to play nearly everywhere, and remain popular for may years (think Windows XP). It may not be technically impressive, but it’s good enough for enough people to keep in regular rotation for a long, long time (see previous, re: Windows XP).

On the other hand, a prog rock epic that’s only performed live by a randomly-changing line-up of musicians is likely to be different every time it’s played. That’s what it’s like deploying a bespoke tech solution (or any LINUX distro) in a live, ever-evolving production environment. You may have loved your favourite love song when it first came out as a rock opera; unfortunately, as the artist experiments with new formats like dubstep country or harpsichord punk, everything you may have loved in the original inexorably mutates beyond all recognition until it’s intolerable.

POC is Keil Hubert,

Keil Hubert is a business, security and technology operations consultant in Texas. He’s built dot-com start-ups for KPMG Consulting, created an in-house consulting practice for Yahoo! Broadcast, and helped launch four small businesses (including his own).


His experience creating and leading IT teams in the defence, healthcare, media, government and non-profit sectors has afforded him an eclectic perspective on the integration of business needs, technical services and creative employees. He currently commands a small IT support organization for a military agency, where his current focus is mentoring technical specialists into becoming credible, corporate team leaders.

Tags: , , , , , , , , , , , , ,