During the 1980s, if you wanted a bank loan or a line of credit you needed to see the bank manager who would suss out your ability to pay and decide yes or no.
By the mid-nineties, a simple formula stood in the bank manager’s place. That simple formula is what banks call a credit score. It predicts your credit risk far more accurately than any bank manager, and costs less too. A credit score is an example of an algorithm: a mathematical rule that predicts something.
Algorithms are amazingly accurate. For example, every year, expert wine tasters sample wines before they hit the market and try to predict how they will do. A simple formula based upon temperature and rainfall (hot and dry is best) when the grapes were growing outperformed the experts. Similarly, simple algorithms often outperform doctors at diagnosing medical symptoms and psychiatrists at diagnosing mental illnesses.
Algorithms have the potential to make the justice system fairer and more efficient. The justice system resists using them, however.
For example, judges differ widely in the severity of sentences they impose on a given crime. It is manifestly unfair if a convicted person randomly receives an eight- or twenty-year sentence because it is a matter of chance which judge they appear before. Judges claim this is the result of the differing circumstances involved in the crimes.
However, studies using model cases showed that this is not the case. Judges differ in the severity of their sentences because they differ in temperament or taste for punishment. The circumstances of the case do not appear to play a role.
From 1987 to 2005 the US courts used mandatory sentencing guidelines to limit the range of sentences a judge could impose. That certainly had the effect of reducing the range of sentences given, but justices started insisting they needed to have discretion. Government returned discretionary powers to the judiciary and the unfairness returned too.
Algorithms are also better than people at predicting whether a released prisoner will reoffend.
Political preferences
Fortunately, parole officers still use algorithms for that purpose. The political preferences alone of the Supreme Court bench of the US predict their decisions with 90% accuracy. We cannot replace the Justices with an algorithm because they still need to specify the details of any change in the law and how courts should interpret them.
Algorithms make crime detection easier. There are fraud-detection algorithms at banks that do a sterling job of detecting if someone else is using your bank card. It is impossible for banks to have people watching every card transaction to spot potential fraud, and even if they did, the algorithm would be more accurate.
Algorithmic systems can save resources by indicating where and when certain crimes are likely to happen, and often who is likely to be committing them. The need to respect the civil rights of suspected criminals prevents the use of profiling to guide police, so police departments do not use these systems to full effect.
There are simple algorithms that predict civil wars. Cliometricians produced the formula via statistical studies of the conditions that preceded historical civil wars around the world.
These algorithms are more accurate than human experts because they are based upon variables proven to be reliably associated with whatever the algorithms are predicting. Human experts on the other hand have a variety of cognitive biases and are prone to use factors that they consider relevant, but which are not reliably related to the phenomena they are supposed to be analyzing.
The reliance on taste for punishment by judges, instead of the circumstances of the case in determining sentences, is a case in point.
There is a downside
So, algorithms can really help us improve our lives – but there is a downside. Elites liberally use algorithms to manage the population efficiently and effectively. For advocates of a strong state, like socialists, this is a good thing. For others it is the biggest threat to human freedom.
To the latter, government is at best a means of lowering the transaction costs of keeping peace, protecting against fraud and crime, providing infrastructure and the like. To them, government is meant to be a servant of the people, an aid to help us coordinate to achieve our communal purposes.
When the people can confine government to those functions, greater efficiency is a good thing. When government stops being a servant of the entire nation and serves only the elite, or has itself become an elite with its own independent interests, efficiency is a source of power used for malign purposes. I am assuming here that the elite usually does not have the interests of the rest of society at heart.
For example, governments can easily watch online speech using algorithms, and once identified it is just a matter of political will to squash what they do not like. Governments often have an incentive to suppress opinions they do not like.
The once liberal UK is approaching Orwellian levels of control under Labour. Algorithms make that possible. China has gone full on Big Brother and uses algorithms to give Chinese citizens a ‘social score’ that can have real consequences if they do not measure up.
I realize I sound like someone wearing a tin hat, but you had better believe that multiple algorithm systems from multiple countries (including incompetent states like SA) are watching you, such as, for example, Google, Meta, and intelligence organizations.
To constrain you
It may just take a change in who is in power, before someone uses that information to constrain you to a path you would rather not be on.
There are few options for avoiding it, short of not using any electronic media. It is an argument for strong privacy laws around access to and use of your data. These should apply to state institutions as much as big tech. I encourage you to produce other ideas.
[Image: https://www.rawpixel.com/image/3289770]
The views of the writer are not necessarily the views of the Daily Friend or the IRR.
If you like what you have just read, support the Daily Friend