Your Company’s Soul, In An Algorithm

Geoff Smith's Blog

Posted On January 07, 2020


I have this reactive stubbornness against people trying to run my life with their algorithms. My bias has always been that if you’re a robot, I’d like you to stay away from the important decisions facing me and our company. I’m all in with ‘data’, and machine learning too. It’s the ‘judgement’ piece, the values part, that I worry about. That, plus how will we keep learning when the machines do all the thinking?

But it now occurs to me that there’s an opportunity here I’ve been missing. Maybe we can organize the situation so that we enslave the algorithms we use, rather than the other way around. Maybe we can force the algorithms to entrench our individuality, rather than blending us all into some great gray mass.

Stop me if you’ve heard this one. It’s a fun story, so I’ve dined out on it a bit. It’s possible I should have been thinking harder, and laughing not so much.

About eight years ago, EllisDon was receiving too many resumes to professionally analyze and respond to them all. So our HR team found – bingo! – an algorithm. Each applicant could do a quick test which would prescreen all these people for us. I hated the idea. I don’t have much faith in a Vice President’s ability (or my own) to predict a candidate’s success in real life, so the idea of having a formula do the job seemed lazy and dangerous.

The HR team persisted. I persisted back. Until finally we agreed to see if the little digital thingee could predict what we already knew. That is, we would put some of our key people through the test, and see if the machine accurately ‘predicted’ what was already known about their individual strengths and weaknesses. And on a whim, I added my own name to the list of crash test dummies.

You know how this ends, of course. The algorithm instructed EllisDon not, under any circumstances, to hire their current CEO. The individual in question would never work effectively as part of a team, and perhaps worse, seemed to have no clear goals, no idea what he wanted to be when he grew up. Stay away at all costs.

So we had a laugh at my expense, and the idea went out the window, at least for the time being. Except for something that took a few days to comprehend: The algorithm had it right. I really don’t like teams (and especially detest team building exercises). Independent thought and intense argument is better than polite consensus building all day long. And on the second point, our Board is continually frustrated by my refusal to have a disciplined strategic plan, I’d much rather keep our options open. The point here isn’t what’s right or best on these fronts, the point is that the algorithm’s analysis of the candidate was dead on, and the algorithm was not impressed.

Which is why I thought then, and think now, that there is real danger here and one needs to be very careful. The algorithm wanted to make our most important decisions – whom to hire - based on someone else’s criteria. These algorithms will employ the best universally accepted conventional wisdom to choose your employees, select the best strategy to employ, tell you what music to listen to, what stocks to buy, and what news to read. (And their vendors will tell you that resistance is foolish.) Crazy ideas and intellectual risk taking not allowed.

Until this fairly obvious idea finally emerged, just recently: Ignore the vendors, pay what it costs, and create your own. Algorithms are inevitable, so we need to make sure that the ones we employ reflect precisely whom we want to become, the people we want to go on that journey with, and the values that will define our life (and company). We need to do the hard work of deciding exactly what the characteristics of all that are, and find the courage to commit to it.

Do you want a company full of argumentative individualists or cooperative team builders? Do you like a higher risk/reward approach or the opposite? Whom do you want to marry? What kind of music do you want to listen to? Write your own algorithm, then hire some math genius whose coding skills will bring all of that uniqueness to your digitized doorstep, and the world will unfold as it should.

Buy someone else’s algorithm to run your business (and your life), and you will end up with someone else’s business, (and someone else’s life). These little formulae may be the most important element of your digitized future; shouldn’t they reflect the unique existence you want to lead? Whatever it costs, surely it’s worth all of that to create your own.

That’s it. Thanks for reading. (Megan and I have some close friends coming over tonight. We haven’t seen them in a while, it should be really terrific. I probably won’t pick out any music though, too much work, we’ll just download a Spotify playlist. I really hate Journey, but it’s a sure bet I’ll be plugging my ears to ‘Don’t Stop Believin’ in a few hours.)