The problem isn’t the algorithm. It’s the class system
A string of equations didn’t create our unfair education system, it revealed it.
"Fuck the algorithm!"
A slogan for Gen Z, on whose lives algorithms will have a much greater impact than any before?
But the really important thing in the UK’s exams debacle wasn't the imposition of an algorithm. It was the decision made by those creating the algorithm to protect the 'integrity of the qualification' by aligning results with exam centre performance over the previous 3 years. Which might be what you would do if your only concern was the perceived value of a qualification.
Help us uncover the truth about Covid-19
The Covid-19 public inquiry is a historic chance to find out what really happened.
This is important in itself. And it’s important because it’s going to happen much more as artificial intelligence and machine learning, driven by algorithms, become a shaping force in governance around the world.
The key questions here are not about algorithms, they are about what we as a society – or the institutions administering systems – want to achieve. We see this with ‘racist algorithms’ in the US. The algorithms are not racist, of course. When you apply large data sets to advise on who is likely to be committing a crime, or to what sentence someone should receive on conviction, it can expose the systematic biases, because writing the algorithm means writing them down.
If the police are more likely to arrest Black people, the algorithm will suggest that you should arrest more Black people. Used uncritically, it will amplify existing biases. Looked at critically, it can help to expose those very biases.
And this is important for the exam system because it is also marked by deep systematic biases. In this case against students at schools in poorer areas.
The exam system doesn’t work
When I was at Edinburgh University, the institution did some data crunching and discovered that students from schools in poorer areas were more likely to get a first class degree than students from wealthier schools who had arrived with the same A-Level or Higher grades. Which wasn’t surprising.
Anyone who was paying attention could see that wealthier schools had learned to game the exam system to get less bright students into top universities. But over four years of a degree, it was the bright kids who had got in without that advantage who shone. Ultimately, this became such a problem that the university decided to apply ‘contextual data’ to admissions to correct for this bias.
A-Levels and GCSE’s serve several purposes. The main ones are to gauge progress, for university admission, and for employers. As an employer I have never looked at A-Levels, nevermind GCSE’s, though others might. And universities find that, especially since the reforms imposed by Michael Gove and Dominic Cummings, A-Levels do not tell us much about what a school student will achieve at university.
If one of the main reasons why we have school exams is to help with university admission, and if universities have to interpret the grades, it suggests the exam authorities could have done something different. They could have learnt from the Edinburgh University approach and applied contextual data, boosting students from schools in poorer areas: recognising that getting a 70% from a college in one of Britain’s poorest areas is more of an achievement than getting 70% if you’re at Eton.
This would have given results that actually reflected the performance of individual students – a much more important outcome than the abstract aim of ‘protecting the integrity’ of qualifications which, as we discovered at Edinburgh, didn’t have much integrity in the first place. It would help to correct the systematic biases that amplify the existing social divisions. And from Edinburgh’s experience, it wouldn’t dumb down, but sharpen up: since the university adopted this approach, it has become a much stronger and more desirable institution.
Doing this would have been simple in data terms. But it would have been hard politically because it would have reduced the advantages of the wealthy. And wealthy parents have more power.
Every year that Edinburgh University admitted students on their potential rather than just their grades from a broken exam system, there was a systematic attack by private school headteachers and the right-wing media. They claimed that offering places to those most likely to succeed was ‘dumbing down’ and ‘reverse discrimination’. It is with this in mind that the exam authorities no doubt chose to increase the grades of students at private and selective schools while reducing the grades of those at state schools in poorer areas.
The answer to this problem is not less use of algorithms. It is for us to understand how algorithms work and to make sure that they work for the right outcomes. It is the system that is the problem, not the algorithm.
And perhaps the lesson is that the slogan should be “Fuck the System” rather than “Fuck the Algorithm”.
Why should you care about freedom of information?
From coronation budgets to secretive government units, journalists have used the Freedom of Information Act to expose corruption and incompetence in high places. Tony Blair regrets ever giving us this right. Today's UK government is giving fewer and fewer transparency responses, and doing it more slowly. But would better transparency give us better government? And how can we get it?
Join our experts for a free live discussion at 5pm UK time on 15 June.
Claire Miller Data journalism and FOI expert
Martin Rosenbaum Author of ‘Freedom of Information: A Practical Guidebook’; former BBC political journalist
Jenna Corderoy Investigative reporter at openDemocracy and visiting lecturer at City University, London
Chair: Ramzy Alwakeel Head of news at openDemocracy
We’ve got a newsletter for everyone
Get our weekly email
CommentsWe encourage anyone to comment, please consult the oD commenting guidelines if you have any questions.