The Handover
How We Gave Control of Our Lives to Corporations, States and AIs
David Runciman
Having read it
★★★★★
It’s a really good book with plenty of considered and well explained observations, insights and views that will get you wondering why many of its points are not made clearer in and from the many systems and processes that we need to help support each other – the people and the state – and assist, debate and regulate adaptably and efficiently the organisations that work in and around it all.
The whole book alludes to and a few parts say so, that quite a few attitudes and approaches to the running and maintaining of things don’t help, particularly that modern classic, social (and even our main) media do not really help us, just amplify distress and anxiety at many levels – predominately because of their near-instantaneous answers that are just decisions parading as answers! – and might even look like they help matters.
Instead the algorithms (and advertisers) that run them just ‘drown’ users in misunderstanding, misappreciation and misplaced attention to appear useful and maybe conducive, possibly even helpful, to societal conversation.
Still, this passage sort of hits one of those nails on the head . . .
The young and the old also want different things from the state. How its durability comes across is likely to depend on where you are in your own life cycle. If you are in your twenties, the state is liable to seem elderly and set in its ways, a tie to the past rather than a bridge to the future. Increasing numbers of young people are expressing their frustration over how unresponsive it is to their needs. If you are in your seventies, the state looks like a source of stability in an unstable world, a way of connecting the future to your own lived experiences in the past. In mature democracies older voters are far more likely to cast their ballot than younger voters, which means that the state is more likely to reflect their wishes and needs. So its perceived unresponsiveness to the interests of younger voters grows.
Perhaps there really is a future for us all in which humans and machines work productively together, facilitated by a state that understands the requirements of both. Yet what stands in the way of this nice idea is politics, which remains rooted in the human dimension of the state, with all its presentist biases and divisions. The robots can’t help with that. Or can they?
A good passage
Humans bring the state to life by animating its presence in the world. They provide the thoughts, while the state enacts the consequences. It is a hybrid creature, through which merely human judgements are able to become the judgements of the state. If we are lucky, these will be intelligent judgements. If we are unlucky, they will be unintelligent ones. Humans vary greatly in their ability to process information and come up with smart solutions. This is as true of political leaders and governments as of anyone else. The superpower of the state is not processing power but decision-making power. It takes human decisions and makes them supremely powerful.
But why do we talk about the state acting when it is human beings who are doing the deciding? Why are we so keen to see these as the state’s decisions when the state can do nothing without the thinking power of its human components?
A second good passage
This potential for dangerous confusion extends to the world of machine learning and search engines. Search engines are designed to give us answers, not decisions. Google doesn’t decide what you are looking for, in the way that a state decides how much tax you should pay. It works out what you are looking for, and it tells you where to find it. Nonetheless, the answers that it gives are shaped by the decisions people make. If you Google the question ‘What is beauty?’ and are shown a series of images of white people, it would be a mistake to think this was an answer to your question, rather than a reflection of the choice of many people to equate whiteness with beauty, and the decision of the designers of Google’s algorithm not to correct for that. Treating a decision as an answer risks mistaking a prejudice for a fact.
Equally, treating an answer as a decision risks mistaking a fact for a judgement. If someone pursuing an insurance claim supplies the information asked for by the insurers and is told that the computer says no, it sounds as though the decision to reject the claim has been made by an algorithm. But that is not true. The algorithm has answered the question of whether the information meets the criteria required for a successful claim – that is an answer, not a decision. The decision to reject the claim lies with whoever decided on those criteria. An answer dressed up as a decision can be designed to make it appear as though the decision brooks no alternative. When the computer says no, there is apparently nothing anyone can do about it. But that too is an illusion. If there is nothing anyone can do, that is because someone has decided to leave it up to the algorithm. It is not the computer’s decision to make.
Groups can decide things and groups can know things, just as other kinds of machines can. Neither requires consciousness. All it takes is a procedure, or an algorithm. Still, there is a big difference between deciding and knowing.
A third good passage
All stable societies of even moderate scale need an effective government to organise their affairs. The risk, however, is that any government can become exploitative and corrupt (and that includes democratic ones). Give a small group of people the power to organise others and you also give them the power to take advantage of their position to enrich themselves. At the same time, any prosperous society needs civic dynamism, which depends on independent enterprises being able to organise themselves. The risk here is that their independence undoes social stability by allowing those making the money to make their own rules.12