![\"\"](\"http:\/\/www.alainveuve.ch\/wp-content\/uploads\/Fotolia_107214712_S.jpg\")
<\/p>\nThe longer I work with machine learning concepts and big data, the more I come to the conclusion that we need to fundamentally rethink our idea of data protection. This may sound strange or disturbing at first, but like many things in today’s world, it is a question of perspective. <\/p>\n
<\/p>\n
(Reading time 6 minutes)<\/p>\n
Data is experience. The more data a machine learning-based system can incorporate into its decisions and mode of operation, the better these decisions will be. In order for this to work better and better, we need the appropriate performance to process these experiences. The more of them that can be combined into a specific topic, the better. This is how we function as humans and the same logic applies to intelligent machine systems. <\/p>\n
Aggregating these experiences is the basis for learning. Learning happens when we combine the aggregated experiences and trigger an action based on them and evaluate the success of this action. <\/p>\n
<\/p>\n
The reason why we are not good at something is that we either have too little experience, too little intelligence to be able to combine experiences or a reduced ability to evaluate the triggered actions and incorporate them as a changed experience for the next actions. Or, as in today’s machine learning, all three. <\/p>\n
We can observe how these roughly outlined processes and dependencies play out very well in growing toddlers, if we leave out the “J\u00f6 factor”. It is fascinating to see how children “program” themselves accordingly in small steps – in other words, how they learn things. I’m not talking about conscious learning such as reading and writing, but unconscious learning. <\/p>\n
For example, I often criticized my wife when she gave our very young children yoghurt to spoon themselves. In the beginning, this endeavor always ended in total disorder because the little ones were unable to understand the interplay between moving the cup, bringing the spoon to their mouth, the correct amount of yogurt on the spoon, etc. and to act accordingly. In addition, motor skills are only rudimentary at the beginning. <\/p>\n
The result was that the amount of yogurt that actually found its way into the child’s stomach was roughly equal to the amount of yogurt that was spread over the table, chair, sweater, etc. (much to my annoyance).<\/p>\n
Eating yoghurt neatly is quite a difficult task, which we or most adults today can do casually. Over time, however, I realized that these little people learn incredibly quickly. Two-year-old children need between 15 and 35 “yoghurt-eating” experiences to master the task more or less well. <\/p>\n
I’m not going to go into detail, but my rough estimate is that these are pretty much all the obvious combinations of things you can do wrong when eating yogurt (e.g. holding the spoon wrong, not holding the cup, too much yogurt on the spoon, etc.). It is fascinating that most people don’t seem to make mistakes twice at the beginning of their lives (considering that experiences are sometimes simply forgotten). What I find more fascinating is that without these experiences, learning the task of eating yoghurt would not be possible at all. <\/p>\n
This is a major difference between human systems and computers. It is very difficult to transfer external experience to humans. We can certainly memorize information. But the experience that helps us to master complex tasks is something we generally always have to gain ourselves. This is a lengthy and exhausting process that most people become less interested in as they get older. <\/p>\n
Computers, on the other hand, can import and process data and experiences one-to-one. And they do not forget. Admittedly, the complexity of these computer tasks and their processing is still at an extremely low level. But the progress in this area is remarkable. And so are our expectations of computer technology, btw. <\/p>\n
George Orwell’s 1984<\/a> is directly associated with the ubiquitous availability of data and, logically, with data protection. I think it is an important novel because it has sensitized many people to data protection issues. <\/p>\n On the other hand, the dystopian scenario of the book has not become reality at all. Of course there is bugging, of course data is misused. But the fact that this is done by a centralized, all-controlling force is not the case at all. On the contrary. The Internet has ensured that every individual has more diverse information at their disposal than ever before. As a society, we still find it a little difficult to handle and evaluate this information<\/a>. <\/p>\n However, the nature of human beings is that we share our knowledge. This does not always happen per se, but rather according to the situation. We withhold knowledge if we hope to gain a certain advantage from it. We release it when we realize that we can make a difference with it. And sooner or later, all knowledge becomes available. Like so much else, we have the Internet to thank for the fact that this development has gathered pace. <\/p>\n But it is also simply in our nature to share knowledge – the more explosive the better. We can’t do otherwise. I believe, and I say this as a representative of open culture, that all knowledge should be freely available. <\/p>\n In order to advance machine learning systems, we therefore need as much data as possible. In Switzerland, for example, data protection was discussed in the early 1970s and then passed into law in 1992. While the misuse of data by the state was discussed during these deliberations, probably also to some extent due to “1984”, the focus today is more on the digital industry giants “GAFKA”. The longread from 1975<\/a>, in which the significance of data protection and its legal classification was discussed, is interesting. <\/p>\n In recent years, however, I have observed an oversensitivity when it comes to data protection. On the one hand, this has to do with the sensitization of the topic. On the other hand, it has to do with technological developments that legislation cannot keep up with. Efforts<\/a> are underway in this area. And that is a good thing. <\/p>\n A central question in this debate is who owns data. Intuitively, we believe that the experiences we have had belong to us. This also applies to experiences we have had with other people. We are happy to pass on these experiences, whether they are good or bad, and usually do so carelessly. If I, as the person with whom the experience was made, come off well (e.g. as a recommendation), I think that’s good. If not, I consider the dissemination of this information to be unfair. <\/p>\n It’s exactly the same with data. Except that we personally see it completely differently. When companies or institutions collect data about us, most people don’t think that this data belongs to these companies. If they pass on the data and it is to our advantage, the vast majority of people have no concerns at all. As soon as this data leads to personal disadvantages, there is great consternation and data protection concerns. <\/p>\n Individuals must be protected from systematic abuse, quite simply because large institutions can take advantage of individuals. However, situational assessment is neither logical nor beneficial to society in the long term. <\/p>\n I believe we need to develop a more relaxed and laid-back approach to data, including personal data. Data protection should not be confused with personal privacy. But the boundaries are fine and fragile and are subject to the ravages of time. <\/p>\n The manifesto of the Swiss OpenData association,<\/a> which is a section of the Open Knowledge Foundation<\/a>, sets out the following principles:<\/p>\n Transparency is a minimum standard. Only compelling reasons such as the protection of privacy justify exceptions. <\/p><\/blockquote>\n The information society cannot afford to let innovation potential lie dormant.<\/p><\/blockquote>\n While much of the work of representatives of the OpenData movement relates to the state, I think we need to apply the same principles to the private sector. Companies need to start publishing more of their data via APIs. This is how the whole of Western society will move forward, because we can suddenly learn from all experiences. And this is the only way we will be able to fundamentally advance machine learning technology without certain people or groups being taken advantage of. <\/p>\n Artikel auf Social Media teilen:<\/p>\n<\/div>Knowledge is shared intuitively<\/h3>\n
And now data protection?<\/h3>\n
Who owns the data?<\/h3>\n
OpenData<\/h3>\n