Laplace, prediction, and why we might, everywhere we go, always take the weather with us in care proceedings
By the start of the nineteenth century, scientists had discovered a great many of the principles of physics and particularly how various forces acted on objects in predictable and mathematical ways. This led some scientists to hubristically predict that there was nothing new to be found in the world of physics (obviously not aware that radioactivity, splitting the atom and quantum physics were completely unknown to them at that point).
Anyway, once you discover the various mathematical principles about forces and objects and how forces act upon objects, one starts thinking about whether you could predict something with absolute certainty if you had enough information.
Being a previously sad geeky sciency Suesspiciousminds Junior, I had certainly wondered in my adolescence whether you could, if you had really fast computers and knew everything, no longer be guessing a toin coss, but knowing how it would end up.
That’s something which has also exercised the minds of a great many gamblers, since Roulette is essentially just an exercise in predictable physics (speed of spin of the table, angle and speed at which the ball is dropped) – predictable, but extremely complex, and if you could actually predict which slot the ball would drop into, with certainty, you would be an extraordinarily rich person.
Well, someone else, Pierre-Simon Laplace took that a stage further, and suggested that with a great enough intellect (computers weren’t really around at that stage, other than Babbage’s mechanical one which was more of a theoretical concept than something you could actually boot up and play Farmville on), you could calculate the entire future of the universe and the movement of every particle.
“We may regard the present state of the universe as the effect of its past and the cause of its future. An intellect which at a certain moment would know all forces that set nature in motion, and all positions of all items of which nature is composed, if this intellect were also vast enough to submit these data to analysis, it would embrace in a single formula the movements of the greatest bodies of the universe and those of the tiniest atom; for such an intellect nothing would be uncertain and the future just like the past would be present before its eyes.”
This is really the birth of determinism, the idea that you can, given enough information, accurately predict future outcomes, or more broadly, that given a set of conditions, the outcome which emerges from those conditions is the only one which COULD have emerged.
[Sadly, I learned when doing a bit of quick research, that Laplace’s other claim, that Pope Callixtus had once excommunicated a comet, was fallacious. I have a later essay planned on how the law has treated animals and inanimate objects, and that would have fitted perfectly with the excommunication of beetles and the pig who was put on trial for murder]
I won’t get any further into whether Laplace’s grand conjecture is true or not (if only in a deeply theoretical sense), and it is still debated – Einstein firmly lined up with Laplace on believing that there were firm mathematical laws and principles underpinning all matter and physics and that it would therefore be possible to predict things with certainly, but that there were just things that were yet unknown to us that prevented such predictions being made. Many others think otherwise, and that there’s an element of randomness, particularly at the quantum level that makes that impossible.
Let’s move away from correctly predicting the motion, position and velocity of every particle in the universe and onto a smaller scale, and some predictions which are common to every one of us, and which enter our homes on a daily basis.
And that allows me to yank it back to care proceedings – in one of the dominant cases of the 1990’s, Re H and R 1996, the House of Lords grappled with the issue of what ‘likely’ meant, when considering whether a child was ‘likely to suffer significant harm’ and this is one of the more memorable passages from Lord Nicholls of Birkenhead :-
In everyday usage one meaning of the word likely, perhaps its primary meaning, is probable, in the sense of more likely than not. This is not its only meaning. If I am going walking on Kinder Scout and ask whether it is likely to rain, I am using likely in a different sense. I am enquiring whether there is a real risk of rain, a risk that ought not to be ignored. In which sense is likely being used in this subsection?
And if you know the law, you will grasp that the latter is where we ended up at in terms of likelihood – it does not mean something that is more likely than not to happen, but a risk that cannot sensibly be ignored.
But in a real sense now, I am going to talk about the science of predicting the weather – will it rain on Kinder Scout today or not?
As you will know, the field of predicting the weather has moved beyond hanging up pine-cones or (my standby) looking at whether cows are lying down in a field (a belief I can’t shed, despite knowing how stupid it is, and one which gets me regularly mocked by Ms SuesspiciousMinds)
Meteorology instead uses a combination of :-
- Gathering lots of information about the current situation
- Applying mathematical principles and formula to predict how features in one part of the system will interact with another
- Calculating therefore what a particular part of the system is likely to do at a future point
And thus, is a system that would make Laplace very proud.
The principles that govern whether we get rain, or snow, or a nice bright sunny day, are pretty uncontroversial. There isn’t a band of quarrelling meteorologists bickering about whether isobars are of any significance at all or whether the warm fronts we see so much of on the television are merely illusory. So, the principles are all there. The mathematical models for what these set of conditions will do over the next few hours are there (based largely on thermodynamics and fluid dynamics), and have been refined and improved, the collection of information about those conditions has vastly improved over the last thirty years, as has the quality of computers doing the calculations.
But what is your first answer, quickly, when I ask
“Do you think we can reliably forecast the weather?”
Making my own little forecast, your instant reaction was no, or that we are hopeless. You may, if you are a fair-minded person, have had a momentary recalibration and decided that we are better at it than we used to be, or even that we are not bad at it now.
But let’s go back to Lord Nicholls – it is March, you are about to go up Kinder Scout and the weather forecast says that it is probably not going to rain. Do you take a coat, or not?
Is the risk that the weather forecast will be wrong when it says there won’t be rain, a risk that cannot be sensibly be ignored, if you find yourself up on a mountain without a coat?
You may have had nagging at the back of your mind, or the front of your mind if you are a science geek or liked Jeff Goldlum’s character in JurassicPark, the notion of chaos theory at this point. You may even have recalled the image of a butterfly flapping its wings and causing a hurricane on the other side of the world [incidentally, probably the most misunderstood image in the history of science – it doesn’t CAUSE the hurricane, it is about how small factors can amplify and make things harder to predict]
Essentially, small factors amplify with time, and the way they amplify is hard to predict, so even the very best computer forecasts become more and more unreliable with the passage of time. Forecasts are far more reliable about the next few hours than they are about next week, and break down almost entirely after sixteen days. In numerical models, extremely small errors in initial values double roughly every five days for variables such as temperature and wind velocity
[So every time the newspapers tell you that there are predictions that this is going to be a “barbecue summer” remember that the accuracy beyond 16 days is all to cock]
Okay, so predicting the weather, which is based on inanimate objects, which act under the influence of known forces, in known ways, and which the science of meteorology has been refining and checking against known outcomes to improve the prediction models, isn’t all that accurate and is not very accurate at all after 16 days.
Now, I will pull us back to law.
At the conclusion of a criminal trial, things are simple – did this person do what they were accused of, and has that been proven. It’s similar with any other sort of legal dispute – did one person prove that x happened, and what punishment / compensation should the Court give. The Court doesn’t really have to predict the future – a burglar isn’t convicted of an offence of burglary only if the Court think he will do another burglary next week.
Care proceedings aren’t like that – whilst we may well spend some time arguing about precisely what happened in the past and the Court may have to decide that if we can’t hit on a form of words which everyone can agree, mostly what we are doing is predicting the future.
- Have the improvements seen in the mother’s parenting at a mother and baby placement, or in contact, mean that she can now safely care for the child, or is she going to slip back into her old ways once she stops being watched all the time?
- Is this father, who has been using heroin for 6 years but has been clean for 4 months, going to remain clean, or will he slip back? (What if he was clean for 6 months, but had one lapse?)
- Will the mother, now that she has seen how risky an individual her new boyfriend is, stay away from him when the proceedings are over, or will he be back in her life and have the chance to hurt the child?
- Will the parents who broke their four year old’s leg by handling him far too roughly, ever do anything like that again?
I have probably sledge-hammered this point, rather than making it in a subtle way, but if top scientists with huge computers can’t predict whether it will rain on Kinder Scout tomorrow, how can we possibly predict with certainty whether the mother will succumb to text messages from the dodgy boyfriend and keep seeing him in secret?
Professor Monroe touched on this in her first report – there was for a long time a body of thought in social work, or social work management, that we could avoid the twin pitfalls of social work – being too soft and letting children get hurt, or being too hard and breaking up families who could have stayed together (Baby P at one end, Cleveland and Orkney at the other) by having more information, more accurate models, and getting the decisions just right.
1.43
Professionals can make two types of error: they can over-estimate or underestimate the dangers facing a child or young person. Error cannot be eradicated and this review is conscious of how trying to reduce one type of error increases the other.
1.44
The public tend to learn of cases of abuse after a child or young person has died or suffered serious harm and then, with the benefit of hindsight, make judgments on how it was easy to see that the child or young person was in danger and would have been safer if removed. This is of course not the way the issue looks for the professionals who only have foresight. Removing a child or young person can protect them from immediate risk of significant harm, but is understandably traumatic for them. Maltreated children or young people who come into care often benefit in the long term, but although the outcomes achieved by looked after children have improved, in too many cases, the potential of the care system to compensate for early harm is unrealised for reasons which are well documented.
Our society rightly values the birth family as the primary source of care for children and young people and disrupting that bond is seen as a serious step to take, requiring close scrutiny before the courts will grant the legal authority to do so.
The birth family equally presents a mixture of benefits and dangers. A good assessment involves weighing up these relative risks and benefits and deciding which option, on balance, carries the highest probability of the best outcomes for the child. Neither option carries zero risk of harm.
1.45
In assessing the value of leaving the child in the same situation, professionals have to consider a balance of possibilities: to estimate how harmful it will be, to consider whether it might escalate and cause very serious harm or death. They also need to consider whether resources are locally available so that families can be helped to provide safer care and estimate how effective such interventions are likely to be.
1.46
All of these areas of uncertainty make decisions about children and young people’s safety and well-being very challenging. A well thought out decision may conclude that the probability of significant harm in the birth family is low. However, low probability events happen and sometimes the child left in the birth family is a victim of extreme violence and dies or is seriously injured is therefore very important. Public understanding that the death of a child may follow even when the quality of professional practice is high is therefore very important.
She says, and as you can see, I agree, that you just can’t hope to get every case right, when you predict the future, your predictions have limitations to their accuracy. If you try to move down the safety first side of the scale, you will take children away unnecessarily. If you try to move down the keeping families together side of the scale, some children will be badly harmed at home. The aim to just make the right decisions at the right time, in all case is simply never going to happen.
If the weather forecasters can’t get it right, neither can we.
You are dealing with people, with all their uncertainties, capriciousness and emotions, and you can’t predict exactly what they will do. The cases where you get it ‘just right’ may well end up being few and far between, and may well be more by luck than judgment.
A mother who is utterly resolute about remaining separate from her dangerous ex-boyfriend, who understands what is at stake and how bad he is from her, may on any given day fluctuate about just how resolute she is. Maybe someone handsome smiled at her at a bus stop and she feels good about herself when he sends the text message and she deletes it without reading it. Maybe just before the text message came in, she caught sight of herself in a mirror and felt fat and unloveable. It is utterly impossible to predict that. It seems easier to predict that a mother that tried to separate from ex boyfriend six times and always went back to him, and was caught out two weeks ago, probably won’t stick to her claims that it is all over and she will never see him again. But we can’t be SURE, we can only predict whether the risk is one that cannot be sensibly ignored.
None of that means that we simply give up, and either leave all children at home with their parents, or take away every child where there is a sniff of danger, but we do have to be honest with ourselves, and honest with society as a whole.
And we have to constantly test where we find ourselves on the scale of child rescue and family preservation – are we lurching too far down one end or another? Are we risk averse, fearful of a Baby P headline and ignoring that those actions break up a family which could have stayed together, or running with a rule of optimism that small changes mean a good future prognosis and not seeing the full picture?
We are attempting to predict what human beings, with human emotions, will do in the future – not just in the next few days, or 16 days, but over the course of their children’s childhood. And the very sort of parents that we attempt to do that with tend, not always, but more often than not, to be emotionally fragile, damaged people who are chaotic and unpredictable in their actions.
suesspiciousminds 