There be treasure in mapping software

I like Google Maps. Having largely shunned sat-nag for traditional paper-based mapping in my car (although it’s becoming difficult to find a decent map book these days), I like the step-by-step directions Google offers at journey end points. It can be confusing at times because it lists road names to travel along, which are not always obviously signposted under the UK’s nebulous road marking system. But on the whole it’s pretty good.

One thing I don’t do though is trust the route it gives. It’s only software and is thus only as clever as the people who wrote it. It plots my route to work, for instance, by the main roads because it’s “shorter”, but in the mornings a combination of back streets and motorway are way faster than the more direct route it proposes. Conversely in the evenings, the A roads, a smidgeon of a different motorway and back streets are way faster than its recommended route.

The facility to get directions and give it a departure time is a vast improvement over just being able to see journey times as a whole. Go crowd sourcing! But even then, it doesn’t always pick the best route, especially if you project it to rush hour.

An example: I plotted a route a few hundred miles away, telling the software I would be leaving at lunchtime, which would mean by the time I reached the destination it would be rush hour. It dutifully told me the time for my journey would be between 3hrs 20mins and 6hrs. Traffic, huh?

But I noticed that along the route it chose — although both shorter and usually quicker in non-traffic situations — there were a lot of red and amber sections: slowdowns.

So I dragged the cursor to change the route to the M1, which although is a longer path, claims a journey time of between 3hrs 30mins and 5hrs 20mins. For the extra ten minutes minimum, I’ll take narrower variation any day.

And this is where software misses out. It’s always looking for a shortcut. For the quickest or shortest route from A to B, regardless that in this case quickest is only under ideal conditions. It’s better to use statistics to your advantage and be the tortoise not the hare. Narrower variance — less deviation from the mean — pulls me closer to the centre of the bell curve, which increases the likelihood of me getting to where I’m going in a reasonable time frame instead of running the risk of encountering outliers. Being a fairly risk-averse person, that suits me better.

Sat-nag systems use the same kind of logic as Google Maps so, by definition, will also try and take you the shortest route, usually via trunk roads which become gummed up because everyone is following the route supplied by the damn machines.

So take heed: next time your mapping software or satellite navigation device suggests a route, don’t blindly follow its advice. Try a few things out first. If you can reduce the variance, you’ll get to your destination on time more often, and more consistently than taking the software-supplied recommendation.