advertisement


To Parents of 16 and 18 year olds

This is the key point, there's only a certain number of top grades to go around. I think people have only begun to grasp that particular nettle this year. By allowing parents essentially to purchase an unfair advantage (smalller class sizes, more equipment, private coaching etc), kids in the state system have been losing out for years. It's quite simple really we stop all this nonsense of fitting a Gaussian to the distribution and have straightforward grade boundaries that eveyone understands. Moderation would then only be used in special cases, i.e. where something has gone wrong.

Do you think that would require more time from teachers than was required by the 'current' process?
 
Although it has still to be published, it is a very lengthy thing ("150 pages" if this article is to be believed: https://edtechnology.co.uk/he-and-fe/36-of-a-level-grades-in-england-downgraded-by-ofqual-algorithm/) and juggles determines the score from a number of factors (previously discussed) but also seems to be is subject to one overriding principle - no grade inflation (https://www.hepi.ac.uk/2020/08/10/a-levels-2020-what-students-and-parents-need-to-know/).

Just have a think about what's happened here for a minute. There's been a mad panic to write an algorithm to generate a fair and equitable exam score that has, rightly, tried to us the minimum necessary input from teachers to allow them to get on with other pressing issues. Yes, a politician will have signed it off but on the principles of how the algorithm works not the detail. The algorithm seems to have taken 3 months to write, test, adjust before being used, which is not bad going when you consider this is all new ('unprecedented' is way to overused at the moment). And apologies if I'm teaching granny to suck eggs but the test runs would generate score distributions, not individual scores so it was always the case that until the moderation was run in anger that these issues would come out.

Welcome to the brave new world of algorithms. They are difficult to write and, as they increase in complexity (150 pages is quite complex), often difficult to predict how behaviour will change over time (https://analyticsindiamag.com/8-rea...thms-turned-rogue-causing-disastrous-results/). Am I using this to excuse the current situation? To a degree, yes, but I suggest that it's likely that this is just an unfortunate result of a set of rules and it is unlikely class or race bias was baked in; that would be all too difficult, especially in three months. What is more likely is that, as has been seen with previous problems with algorithms, is that use of historical data which is derived from processes that themselves contains biases and discrimination, will generate biased output (and often amplify them). It will take more than three months to sort out those issues - they are long standing problems, subject to deep debate and not going to be fixed by a bunch of mathematicians and coders.

TL;DR - Too complex to engineer in /out bias in the code. Too much emphasis on avoiding grade inflation.

Biggest mistake is, however, political. Not recognising that there will be losers in this and not allowing proper appeals process based on individuals performance.
Agree with all this, especially “welcome to the brave new world of algorithms” - this is the future of governance as Cummings sees it, and that will be a big obstacle to any u-turn. Data - collected and processed opaquely, used to develop opaque and unaccountable algorithms - is going to determine the distribution of rights, privileges and resources if Cummings has his way.

“Not recognising there will be losers in all this” - this is key: Conservative thinking refuses to recognise the existence of class, racial or gender bias, or structural inequality, so it can’t see how these things might be embedded in code. Can’t adjust for it if you don’t believe in it.

It will be interesting to see how they account for glaring injustices and absurdities. In conservative thinking responsibility always rests with the individual. So this is all somehow going to have to be the teachers’/pupils’/parents’/experts’ fault.
 
For what it’s worth, I as a teacher and a parent agree that schools must open, and that if there’s a cost in other things being closed that’s a necessary evil. In future if we have to lockdown again they should be the first thing to come out, not the last thing.

I agree, just about all teachers I know agree. Everyone agrees. Which is why the guff about teacher unions trying to prevent schools reopening and the language of ‘moral duty’ should be challenged.
 
it will also produce anomalies at the top with pupils getting A* instead of, for example, B’s.

not always, algorithms of this type are complicated beasts and cannot be engineered quickly without unforeseen circumtances

Ww are capped here. So even if we wanted to, we can't take all the students on their predicted grades.

Indeed although we did get some extra numbers in the mad scramble of May 2020. But subjects like medicine, dentistry and many others have resource restrictions.

There's been a mad panic to write an algorithm to generate a fair and equitable exam score that has, rightly, tried to us the minimum necessary input from teachers to allow them to get on with other pressing issues

Astonishing they managed to pop one out in that time.

Welcome to the brave new world of algorithms. They are difficult to write and, as they increase in complexity (150 pages is quite complex), often difficult to predict how behaviour will change over time (https://analyticsindiamag.com/8-rea...thms-turned-rogue-causing-disastrous-results/). Am I using this to excuse the current situation? To a degree, yes, but I suggest that it's likely that this is just an unfortunate result of a set of rules and it is unlikely class or race bias was baked in; that would be all too difficult, especially in three months. What is more likely is that, as has been seen with previous problems with algorithms, is that use of historical data which is derived from processes that themselves contains biases and discrimination, will generate biased output (and often amplify them). It will take more than three months to sort out those issues - they are long standing problems, subject to deep debate and not going to be fixed by a bunch of mathematicians and coders.

yes

Biggest mistake is, however, political. Not recognising that there will be losers in this and not allowing proper appeals process based on individuals performance.

there is no excuse for not having a properly prepared appeals process.
 
There's been a mad panic to write an algorithm to generate a fair and equitable exam score that has, rightly, tried to us the minimum necessary input from teachers to allow them to get on with other pressing issues.

A five month 'mad panic'.

Part of the issue is this Government's insistence on secrecy and a 'they know best' attitude.

Read Cummings' blog. It's all in there.

Stephen
 
Do you think that would require more time from teachers than was required by the 'current' process?

Well to predict grades teachers have a have an understanding of the grade boundaries, but if they are fluid, and indeed change according the centre, the excercise has been pointless. The class only needs to be ranked at all to fudge a preconceived outcome. As I understand things the predicted grades are used in normal years to help to spot potential marking anomalies but perhaps that's not the case and they've been used in the moderation process.
 
Yes, it is the key point. I can't find it now but there is an Ofqual report about the algorithm design where the various criteria it must meet are described. Avoidance of grade inflation is clearly prioritised, at the request of a government minister.
Replying to myself is sad, but:

https://twitter.com/queenofswords6/status/1294471488351985667

Looks like grade inflation isn't such a big deal after all - as long as it's in private schools.
 
Agree with all this, especially “welcome to the brave new world of algorithms” - this is the future of governance as Cummings sees it, and that will be a big obstacle to any u-turn. Data - collected and processed opaquely, used to develop opaque and unaccountable algorithms - is going to determine the distribution of rights, privileges and resources if Cummings has his way.
The GDPR should, in theory, scupper this. It has specific provisions to prevent opaque collection and processing of personal data, and an explicit prohibition on ‘automated’ (ie, algorithmic) decision-making. The circumstances in which these provisions can be circumvented are few, and quite tightly ringfenced and safeguarded.

Having left the EU, of course, we are free to diverge from these provisions, but there will be consequences if we do, to any significant extent. This could, in theory, culminate in the cessation of data flows from EU states to the UK. At the moment, much midnight oil is being burnt to prevent just such a thing happening to the US by ensuring that there are suitably robust safeguards in place.
 
The GDPR should, in theory, scupper this. It has specific provisions to prevent opaque collection and processing of personal data, and an explicit prohibition on ‘automated’ (ie, algorithmic) decision-making. The circumstances in which these provisions can be circumvented are few, and quite tightly ringfenced and safeguarded.

Having left the EU, of course, we are free to diverge from these provisions, but there will be consequences if we do, to any significant extent. This could, in theory, culminate in the cessation of data flows from EU states to the UK. At the moment, much midnight oil is being burnt to prevent just such a thing happening to the US by ensuring that there are suitably robust safeguards in place.
Do you reckon I could request for all my son's gcse data to be deleted and thus get his teacher's predicted grades? ;)
 
The GDPR should, in theory, scupper this.

I predict the following.

Much noise and right-wing press saying "we will no longer be bound by the EU's GDPR regulations."

Much cock ups, noise, financial and job losses and gross displays of incompetence follows.

The Government announce the UKDPR—a much better system that strangely mimics the GDPR but is more 'red, white and blue.'

Stephen
 
Agree with all this, especially “welcome to the brave new world of algorithms” - this is the future of governance as Cummings sees it, and that will be a big obstacle to any u-turn. Data - collected and processed opaquely, used to develop opaque and unaccountable algorithms - is going to determine the distribution of rights, privileges and resources if Cummings has his way.

This is a recognised issue long before Dominic Cummings came on the scene.

https://hbr.org/2018/11/why-we-need-to-audit-algorithms
https://www.computerweekly.com/feature/Auditing-for-algorithmic-discrimination

It's whether the UK is prepared to legislate for transparency / auditing of algorithms or not that is the fight from this point forward.
 
Yes, it is the key point. I can't find it now but there is an Ofqual report about the algorithm design where the various criteria it must meet are described. Avoidance of grade inflation is clearly prioritised, at the request of a government minister.
Yes, an Ofqual consultative committee person admitted as much on Radio 4’s The World at One, today. She also said that Ofqual had done a test run on last year’s cohort and it was accurate for about 2/3 of the cohort, which, she said, was about as good as you’d get in the circumstances. Ofqual seems to have deemed this acceptable, but probably under considerable political pressure, and also pressure to get results out on time.

Grade inflation avoidance seems to have been a priority, this seems to be acknowledged. As other commentators are asking, ‘why on Earth?’ - it’s not like this year’s results will be used as a comparison against anything else. 2021’s results will be compared against 2019’s, not this year’s, as everybody can see these will be anomalous.
 
This is a recognised issue long before Dominic Cummings came on the scene.

https://hbr.org/2018/11/why-we-need-to-audit-algorithms
https://www.computerweekly.com/feature/Auditing-for-algorithmic-discrimination

It's whether the UK is prepared to legislate for transparency / auditing of algorithms or not that is the fight from this point forward.
Not saying Cummings invented algorithmic governance but as far as I’m aware he’s the first to consciously place it at the heart of a project of total government transformation. PM’s clearly on board, or Cummings would be gone. Whether we get legislation for transparency will depend on the extent to which it will get in his way. It’s not clear who might be in a position to insist on it.

I do think the current scandal is going to be good for consciousness-raising on the issue - especially amongst this cohort of youngsters.
 
Yes, an Ofqual consultative committee person admitted as much on Radio 4’s The World at One, today. She also said that Ofqual had done a test run on last year’s cohort and it was accurate for about 2/3 of the cohort, which, she said, was about as good as you’d get in the circumstances. Ofqual seems to have deemed this acceptable, but probably under considerable political pressure, and also pressure to get results out on time.

Grade inflation avoidance seems to have been a priority, this seems to be acknowledged. As other commentators are asking, ‘why on Earth?’ - it’s not like this year’s results will be used as a comparison against anything else. 2021’s results will be compared against 2019’s, not this year’s, as everybody can see these will be anomalous.
Tories don’t govern to fix this or that problem or meet this or that need, so they’re not constrained by the logic of these things. They’re only constrained by their own sh-tty media narratives and “grade inflation” is an important one. Really riles their audience up, and feeds into the larger narrative of young people having it too easy these days and needing a good war or something to toughen em up.
 


advertisement


Back
Top