Cognitive crash dummies

by Paul Curzon, Queen Mary University of London

The world is heading for catastrophe. We’re hooked on power hungry devices: our mobile phones and iPods, our Playstations and laptops. Wherever you turn people are using gadgets, and those gadgets are guzzling energy – energy that we desperately need to save. We are all doomed, doomed…unless of course a hero rides in on a white charger to save us from ourselves.

Don’t worry, the cognitive crash dummies are coming!

Actually the saviours may be people like professor of human-computer interaction, Bonnie John, and her then grad student, Annie Lu Luo: people who design cognitive crash dummies. When working at Carnegie Mellon University it was their job to figure out ways for deciding how well gadgets are designed.

If you’re designing a bridge you don’t want to have to build it before finding out if it stays up in an earthquake. If you’re designing a car, you don’t want to find out it isn’t safe by having people die in crashes. Engineers use models – sometimes physical ones, sometimes mathematical ones – that show in advance what will happen. How big an earthquake can the bridge cope with? The mathematical model tells you. How slow must the car go to avoid killing the baby in the back? A crash test dummy will show you.

Even when safety isn’t the issue, engineers want models that can predict how well their designs perform. So what about designers of computer gadgets? Do they have any models to do predictions with? As it happens, they do. Their models are called ‘human behavioural models’, but think of them as ‘cognitive crash dummies’. They are mathematical models of the way people behave, and the idea is you can use them to predict how easy computer interfaces are to use.

There are lots of different kind of human behavioural model. One such ‘cognitive crash dummies’ is called ‘GOMS’. When designers want to predict which of a few suggested interfaces will be the quickest to use, they can use GOMS to do it.

Send in the GOMS

Suppose you are designing a new phone interface. There are loads of little decisions you’ll have to make that affect how easy the phone is to use. You can fit a certain number of buttons on the phone or touch screen, but what should you make the buttons do? How big should they be? Should you use gestures? You can use menus, but how many levels of menus should a user have to navigate before they actually get to the thing they are trying to do? More to the point, with the different variations you have thought up, how quickly will the person be able to do things like send a text message or reply to a missed call? These are questions GOMS answers.

To do a GOMS prediction you first think up a task you want to know about – sending a text message perhaps. You then write a list of all the steps that are needed to do it. Not just the button presses, but hand movements from one button to another, thinking time, time for the machine to react, and so on. In GOMS, your imaginary user already knows how to do the task, so you don’t have to worry about spending time fiddling around or making mistakes. That means that once you’ve listed all your separate actions GOMS can work out how long the task will take just by adding up the times for all the separate actions. Those basic times have been worked out from lots and lots of experiments on a wide range of devices. The have shown, on average, how long it takes to press a button and how long users are likely to think about it first.

GOMS in 60 seconds?

GOMS has been around since the 1980s, but wasn’t being used much by industrial designers. The problem is that it is very frustrating and time-consuming to work out all those steps for all the different tasks for a new gadget. Bonnie John’s team developed a tool called CogTool to help. You make a mock-up of your phone design in it, and tell it which buttons to press to do each task. CogTool then worked out where the other actions, like hand movements and thinking time, are needed and makes predictions.

Bonnie John came up with an easier way to figure out how much human time and effort a new design uses, but what about the device itself? How about predicting which interface design uses less energy? That is where Annie Lu Luo, came in. She had the great idea that you could take a GOMS list of actions and instead of linking actions to times you could work out how much energy the device uses for each action instead. By using GOMS together with a tool like CogTools, a designer can find out whether their design is the most energy efficient too.

So it turns out you don’t need a white knight to help your battery usage, just Annie Lu Luo and her version of GOMS. Mobile phone makers saw the benefit of course. That’s why Annie walked straight into a great job on finishing university.

This article was originally published on the CS4FN website and appears on pages 12 and 13 of issue 9 (‘Programmed to save the world‘) of the CS4FN magazine, which you can download (free) here along with all of our other free material.

Related Magazine …

This blog is funded through EPSRC grant EP/W033615/1.

Cold hard complexity: learning to talk in nature’s language

by Paul Curzon, Queen Mary University of London

A gentoo penguin slumps belly-first on a nest at Damoy, on the Antarctic Peninsula. Nearby some lichen grows across a rock, and schools of krill float through the Southern Ocean. Every one of these organisms is a part of life in the Antarctic, and scientitsts study each of them. But what happens to one species affects all the others too. To help make sure that they all survive, scientists have to understand how penguins, plants, krill and everything else in the Antarctic interact with one another. They need to figure out the rules of the ecosystem.

Working together

When you’re trying to understand a system that includes everything from plants to penguins, things get a bit complicated. Fortunately, ecology has a new tool to help, called complexity theory. Anje-Margriet Neutel is a Biosphere Complexity Analyst for the British Antarctic Survey. It’s her job to take a big puzzle like the Antarctic ecosystem, and work out where each plant and animal fits in. She explains that ‘complexity is sort of a new brand of science’. Lots of science is about isolating something – say, a particular chemical – from its surroundings so you can learn about it, but when you isolate all the parts of a system you miss how they work together. What complexity tries to do is build a model that can show all the important interactions in an ecosystem at the same time.

Energy hunt

So for a system as big as a continent full of species, where do you start? Anje’s got a sensible answer: you start with what you can measure. Energy’s a good candidate. After all, every organism needs energy to stay alive, and staying alive is pretty much the first thing any plant or animal needs to do. So if you can track energy and watch it move through the ecosystem, you’ll learn a lot about how things work. You’ll find out what comes into the system, what goes out and what gets recycled.

Playing with models

Once you’ve got an idea of how everything fits together you’ve got what scientists call a model. The really clever thing you can do with models is start to mess around with them. As an example Anje says, ‘What would happen if you took one group of organisms and put in twice as much of them?’ If you had a system with, say, twice as many penguins, the krill would have to be worried because more penguins are going to want to eat them. If they all run out what happens to the penguins? Or the seals that like eating krill too? It gets complicated pretty quickly, and those complicated reactions are just what scientists want to predict.

The language of nature

Figuring out how an ecosystem works is all about rules and structure. Ecosystems are huge complicated things, but they’re not random – whether they work or not depends on having the right organisms doing jobs in the right places, and on having the right connections between all the different parts. It’s like a computer program that way. Weirdly, it’s also a bit like language. In fact, Anje’s background is in studying linguistics, not ecology. Think of an ecosystem like a sentence – there are thousands of words in the English language but in order to make a sentence you have to put them together in the right way. If you don’t have the right grammar your sentence just won’t make sense, and if an ecosystem doesn’t have the right structure it’ll collapse. Anje says that’s what she wants to discover in the ecosystems she studies. ‘I’m interested in the grammar of it, in the grammar of nature.’

Surviving Antarctica

Since models can help you predict how an ecosystem reacts to strange conditions, Anje’s work could help Antarctica survive climate change. ‘The first thing is to understand how the models work, how the models behave, and then translate that back to the biology that it’s based on,’ she explains. ‘Then say OK, this means we expect there may be vulnerable areas or vulnerable climate regions where you can expect something to happen if you take the model seriously.’ If scientists like Anje can figure out how Antarctica’s ecosystems are set up to work, they’ll get clues about which areas of the continent are most at risk and what they can do to protect them.

Surviving on a continent where the temperature hardly ever gets above freezing is tough, and climate change is probably going to make it even tougher. If we can figure out how Antarctic ecosystems work, though, we’ll know what the essential elements for survival are, and we’ll have clues about how to make things better. Extracting the secret grammar of survival isn’t going to be a simple job, but that’s no surprise to the people working on it. After all, they’re not called complexity scientists for nothing.