Thanks Bokke – Just How Wrong or Right Can an Algorithm Be?

12 Nov

Caps off to the Springboks who all but blew the English Roses right out of the proverbial water with their convincing win on Saturday the 2nd of November! As the country erupted into a roar of pride and excitement, few people gave much thought to the predictions that had been made before the game. No one cared! All that mattered was that the Boks had won and that their win unified a nation!

Now that the excitement has lulled into a warm, fuzzy feeling for all South African’s, and we cringe as we consider our outspoken criticisms of a team who are now our heroes, we take a breather to look back at some of the pre-World Cup predictions.

There’s a certain prediction software created by Dr David Scott, a statistician from New Zealand. Scott developed an interesting formula that was used not only to predict who will win each rugby match but also what the likely difference in points would be. He created this software with an algorithm.

What is an algorithm?

The Oxford English Dictionary defines an algorithm as follows: “a process or set of rules to be followed in calculations or other problem-solving operations, especially by a computer”. In a nutshell, by making use of algorithms, AI can sort through large quantities of data to determine patterns.

Did the Algorithms Get it Right?

If you have followed the rugby and have an interest in statistics (possibly for betting purposes), you might have found it intriguing when the prediction software got it right. Think about Australia vs Wales games – the prediction was dead on target. For those scared that robots and computers are coming for their jobs and lives, that could have been disturbing news. But don’t worry – the algorithms didn’t get it entirely right, which means your job and lifestyle are safe!

The Devil is in the Details

The one thing that brings to light the unreliability of algorithms is that out of 39 predictions, only 4 were completely wrong, only 1 was spot on, and the rest were mostly right. That’s not too convincing is it? How do you rely on an algorithm if the outcomes aren’t exact? That’s the whole point; you can’t when it comes to sports and well, betting on sports.

Businesses collect data so that algorithms can get to work and predict some pretty important information for them. This type of data and the outcomes are based on past behaviour and is potentially more useful in the business scenario than when betting on a sports match.

Why Does it Matter?

Algorithms should matter to all of us because they are only possible when data is collected (which includes your data too). This data is then used to predict an outcome.

Algorithms can also be used to manipulate an outcome. For example; a government or political party publicly predict a certain outcome backed by viable sounding reasons. Does the outcome perpetuate because it is true, or do people behave in the predicted manner to get the predicted outcome, because psychologically, everyone else seems to be doing it? After all, the government told them it would be so!

Also, with algorithms being used to predict every possible outcome, life becomes a little less interesting. It’s a little mundane to think that humans are working on algorithms that will be able to predict every possible outcome in future moments. Scary even!

Algorithms or No Algorithms – the Boks Won!

At the end of the day, what the algorithm said before the clash of the Springboks and the English roses really doesn’t matter. What matters is that the Boks won and that’s something that no algorithm can take away!

Comments are closed.