Tomorrow’s metrology, the science of measurement for the digital economy | APMP 2019 Symposium

By
Dr Bronwyn Evans, Engineers Australia CEO

For those who don’t know us, Engineers Australia is the peak body for the engineering profession in this country. We have around 100,000 members – including some metrologists [specialists in the science of measurement] who are also engineers, so hello to our members who are in the room!

Engineers Australia acts as the voice of the profession, and we exist to advance the science and practice of engineering for the benefit of society.

We were founded 100 years ago in 1919, so it’s our Centenary year this year!

Before I get started, I’d like to congratulate the global metrology community on the recent elimination of physical objects in the SI measurements – a once-in-a-generation advance that means we can have accurate units ‘for all times, for all people’ around the world… without flying lumps of metal to Paris. A remarkable achievement.

As engineers, we rely on the work of metrologists, and collaborate with you, for example through shared developments of standards on subjects such as nanotechnology.

Engineers Australia recently held the World Engineers Convention in Melbourne, a once-in-a-lifetime event and the first time it’s ever been held in Australia.

One of the things that has come out of that Convention is the announcement that, from next year on into perpetuity, 4 of March will be World Engineering Day for Sustainable Development – so put that in your diaries, as I’ve put World Metrology Day on 20 May in mine.

At the convention, there were some excellent presentations from staff of the National Measurement Institute of Australia, and other speakers showing the power of measurement. 

For example, the Beijing Water Cube, where the swimming was held at the Olympics, they managed to cut energy use by 30 per cent without changing the physical systems – just by better using the sensor data they were collecting to inform a smart algorithm that controlled the existing chillers.

There was much talk about driverless cars but, it seems there are other ways to improve safety on our roads. One of the speakers had measured the accuracy of existing lane assist technology and how it depended on the width of the lines painted on roads, and the contrast of the paint against the road surface. His conclusion? You could cut fatalities on rural roads in Australia by half, just by painting better road markings and using existing lane assist technology!

When people think of metrology and the digital economy, many immediately turn to manufacturing, where sensors and 3D scanners could enable components to be measured and tested at multiple points in the process without probing, reducing errors and cutting costs. Measurement, so the thinking goes, would move from the quality control laboratory to the factory floor. Equipment would need to be more portable, simple and flexible to account for shorter product life cycles.

However, manufacturing is only one element of Australia’s economy. Our economy has changed from one that used to be dominated by physical objects – crops, minerals, manufactured goods – to one dominated by services. And yet, the precision with which we can measure physical objects far exceeds our ability to measure services. Could this change?

I think it’s interesting to reflect on metrology’s history in facilitating human economic activities – trade and exchange – introducing greater fairness and transparency, so you could go to your butcher shop or petrol station and not be overcharged.

The National Measurement Institute was in the media recently with an inspection blitz in shops to ensure Australian families are getting what they pay for in their Christmas grocery shopping – so there’s still that link between accurate measurement and everyday life.

However, in today’s digital economy, we are seeing a situation where there is not equal access to information in trade.

For businesses, it’s all-important to rank on Google searches or to have your content seen on social media, but the algorithms used to create those results are secret, and constantly changing. Google generates 60% of web traffic and 56% of web revenue.

That enables Google and other sites to charge for advertising. In the most recent quarter, businesses spent almost US$34 billion on Google advertising – in three months alone.

Different social media channels have “likes”, “impressions” and “shares” – and these do correlate to purchasing behavior.

Do we need a unit that measures human online attention consistently across all channels – Facebook, LinkedIn, Twitter, TV, radio, Google search, push notifications, signage on the street – to have a fair and informed information market? What do we call the unit? An attention-let, a micro-thought?

Community ratings are currently an inexact science. Here we are in the International Convention Centre Sydney, which is currently a fabulous place according to Google, variously three and a half and four stars on Tripadvisor and an okay-ish place to work according to glassdoor. A single disgruntled customer, staff member or competitor can do lasting damage.

Do we need an internationally consistent unit that measures the likely trustworthiness of information online? Or the carbon emissions per kilometer of travel? Or the quality of datasets? Or healthiness of food?

I’m not a metrologist, but my powerful friend Dr Google who we’ve already met tells me there’s an emerging field of so-called “soft metrology” which explores the powerful domain of human perception and interpretation.

That has to be an area of opportunity in future, and it’s one that will require multidisciplinary input – from metrologists, medical technicians, psychologists… and, potentially, engineers.

Engineering, as a profession, is increasingly multidisciplinary – engineers collaborating not just with engineers from other specialisations but people who are not engineers at all.

  • Could metrology work towards us all better understanding:
  • Psychometric or perceived feeling – touch, taste, odour, colour, as we’re currently further along the track towards artificial vision than an artificial “nose”
  • Qualitative measurement – customer satisfaction etc
  • Econometrics and sociometry (opinion) and 
  • Measurements related to human sciences: biometry, behavior, intelligence etc.

The amount of data already “out there” about each and every one of us is huge, and the implications are similarly enormous – with the potential for immense benefit and immense harm.

They say the punishment has got to fit the crime. Well, the accuracy and reliability of measurement needs to fit the purpose for which the data will be used – and, the problem is, we don’t always know what that ultimate use will be.

For example, people enjoy finding out more about their ancestry by sending saliva or a cheek swab to sites like Ancestry.com and 23andMe. “Hooray, I’m one 16th Scandinavian!”

Now, pharmaceutical giant GlaxoSmithKline has acquired a $300 million stake in 23andMe – and, with it, bought exclusive access to the genetic code, the very DNA, of more than four million people.

There are some problems with this. 

  • Ethical questions – did we provide full consent for medical research when all we wanted to know was our ancestry?
  • Scientific problems – is this a biased data set dominated by people from developed nations?
  • Health questions – is the accuracy of the data sufficiently robust to base a health solution on rather than just an “Oh, that’s interesting!”?

Should a health provider or insurer make decisions on the basis of the output of my fitbit or its more sophisticated successor?

Accuracy of information is one issue. Another is bias. Because the brave new world of artificial intelligence can perpetuate the biases of its human developers. 

For example, couple of years ago, it emerged that a facial recognition program developed by IBM could correctly identify a person’s gender from a photograph 99 per cent of the time – for white men. For dark-skinned women, not so much. The system got it wrong 34% of the time.

If we don’t use the right data training sets, whole cohorts of people risk being left behind.

As an engineer, I know that measurements matter. As a human being and a CEO, I know emotions and perceptions also matter.

Facts like measurements can make people think, but emotion makes people act.

A graph on plastic consumption doesn’t sway behavior the way that it does when we watch Blue Planet 2 and see a mother pilot whale carrying her dead newborn pilot whale for days, and hear David Attenborough telling us the reason is likely to be contamination of its mother’s milk from plastic in the ocean.

And, if we want action – climate action, more votes, more sales, more hugs… whatever it is we want – then we need better ways of measuring and predicting human perceptions, moods, emotions and behavior.

In closing, I’d like to note that accuracy alone should not be the sole goal of any scientist, in my view. We have to consider impact and potential impact on people, and on the environment.

American academic and author Virginia Eubanks is one of many who have drafted a digital equivalent of the Hippocratic oath sworn by doctors. I won’t read the whole thing, but her “oath of digital non-harm” includes some points particularly relevant to those of us in this room as they relate to data.

“I will not collect data for data’s sake, nor keep it just because I can,” Eubanks wrote.

“When informed consent and design convenience come into conflict, informed consent will always prevail.

“I will remember the technologies I design are not aimed at data points, probabilities or patterns, but at human beings.”