(THIS ARTICLE IS MACHINE TRANSLATED by Google from Norwegian)
The economic differences are increasing all over the world. One of the most significant scholars of inequality, Thomas Piketty, became world famous for his book The capital of the 21. century. Thanks to him, we now know for sure that the skewed distribution is growing. According to Piketty, this is due to a market economy based on private property, which benefits people who own capital. Piketty writes that “wealth accumulated in the past is growing faster than production and wages. […] The entrepreneur inevitably tends to become an interest rate worker, and more and more overshadows those who own nothing but their labor. " Or, as he states: "The past consumes the future." You are probably familiar with this view due to all the attention Piketty has received over the last couple of years. But you are probably less familiar with the very different conception of the state of things, which is explained in Weapons of Math Destruction.
Learning by doing, or not. The book was written by earlier math professor and computer scientist Cathy O'Neil, and will open our eyes to the flaws of big data analytics. After all, it is man-made algorithms that deliver the warning models big data uses. O'Neil immediately points to the difference between Google – which has millions upon millions of users who can test and improve the data giant's algorithms – and, for example, a computer model for a single Washington DC school with a few thousand students. According to the author, this has serious consequences. "By promising efficiency and justice, they are deforming higher education, raising debt, stimulating mass incarceration, stabbing the poor at virtually any time and undermining democracy." Mathematical models predict the likelihood of someone submitting, defaulting on a student loan, or committing a crime. Occasionally, the model creates a self-fulfilling prophecy when it is initially wrong, and has difficulty learning from its mistakes if it goes in a feedback loop. In some cases, erroneous models cause real damage. “There are many damaging effects. They show when a single mother can't provide childcare quickly enough to fit her work schedule, or when a young person struggling is blacklisted from an hourly job because of a personality test related to the workplace. We see them when a poor minority boy is stopped, beaten up and given a warning by the local police, or when someone who works at a gas station and lives in a poor area is hit by higher insurance premiums, ”according to O'Neil.
The algorithms deform higher education, drive up debt, stimulate mass incarceration.
Maintain competition. The inequality here seems to create more inequality in an endless downward spiral, where the rich get richer and the poor poorer, just like Piketti's book. The consequences of increasing inequality are great, writes Piketty: "[F] or the long-term dynamics of wealth distribution are [the] consequences potentially scary, especially when one adds that capital gains vary directly with the size of the original stake, and that the difference in the distribution of wealth is happening on a global level. ” He continues: “Of course, growth can be stimulated by investment in education, knowledge building and non-polluting technology. But none of these measures will increase growth rates to 4 or 5 percent a year. ” According to Piketty, a growth rate of 4 or 5 percent is needed to improve the economy for everyone, and not just for the wealthy. But reaching this level of growth is not easy. Piketty points out that only countries that are recovering from more advanced economies, such as China, have seen such figures. More likely, annual growth is 1 to 1,5 percent in the long run, "no matter what economic policy is pursued". So economic growth will not help us in the long run. What then can change the development? Piketty proposes a progressive annual tax on capital. "This will allow an endless spiral of inequality to be avoided while preserving competition and incentives for new cases of primitive accumulation."
Anticipate abuse. With regard to the data models, there are also other measures that can be implemented to avoid a downward spiral of inequality. First of all: Not all big data models are evil. Second: Big data has come to stay. With this in mind, what can we do to prevent the models from running? Models must be transparent and display the data used, just as they show the results of the model usage. They also need to be revised and improved when weaknesses emerge. Or as O'Neil puts it: "These are powerful machines after all. We have to keep an eye on them. " O'Neil goes on to point out that some models can have an incredibly positive impact on society. One example is an organization that prevents child abuse in Florida, USA; The model was developed in 2013 in Florida's Hillsborough County, after nine children had died as a result of abuse in the area the previous two years. The researchers identified several indicators that seemed to predict abuse. These were factors such as substance abuse and domestic violence, cohabiting male boyfriend and parent who had been foster children themselves. The families this model highlighted were allocated extra resources, and for the next two years no children in the risk area were killed. With these precautions we can guard data models before they get out of control and it is too late to do anything about the differences.
