Question to the expert: Is it true that social networks are watching us
Dmitry Kurkin
RESPONSES TO THE MAJORITY OF US QUESTIONS we used to search online. In the new series of materials we ask such questions: burning, unexpected or widespread - to professionals in various fields.
The 10 Year Challenge flash mob, launched in social networks at the beginning of the year, not only spawned conspiracy theories that the campaign’s goal was to collect photos of users and train them to recognize the face recognition system, but also once again made them think about how much they know about us. social networks and third parties working with them (from commercial companies to government agencies).
The fact that technology giants are collecting and analyzing the so-called digital footprints left by billions of users daily is no secret to anyone. And the awareness of this gives rise to a new kind of fear of the “big brother”: social networks know a lot about us, but what if they know too much about us? Can big data be used to find out all the connections, tastes, habits of a person, his past and present? And if so, what harm can our desire to socialize online, for the sake of which we voluntarily share information about ourselves, cause us?
We asked experts about how user data is processed by large companies and how great the danger is to inherit on social networks.
Liliya Zemnukhova
Researcher at the Center for Science and Technology Research at the European University at St. Petersburg
A digital footprint contains all the possible types of data — texts, images, audio and video recordings, geolocation, and a lot of metadata (for example, gadget model, mobile operator, operating system, dynamics and duration of visits, etc.). And it’s not just us who replenish our digital footprint. Social networks form us as users with the help of three data sources: the fact that we ourselves report about ourselves; that others report about us; and what is going to most often without our knowledge. Especially opaque last. We, as a rule, do not read user agreements and policies for the collection and use of personal data. We only note that this “black box” somehow influences our user experience: targeted advertising, suggestions from friends, recommendations for music, the procedure for launching news ... We construct a small part of this experience ourselves, when we manually build the news feed, but mainly algorithms perform the functions embedded in the default profiles. That is why we will never get rid of contextual advertising or intrusive suggestions of groups or (not) friends. Social networks as corporations use data about their users for commercial purposes, offering their platform for selling targeted content. And along the way, they continue to collect data about us: for example, if you have paid for advertising at least once, then the bank card and transaction data also remains with the company. Data can also be provided to government agencies when there is a great need: for example, Facebook regularly collaborates with US government agencies, in accordance with its policy of transparency.
In addition to the internal policy of social networks, there is one more important detail: accounts can be associated with hundreds of thousands of other applications and functions. This, for example, was the reason for big discussions last year about third-party access to user data. An important attempt to regulate the freedom of developers was made in the European Union - the General Data Protection Regulations (GDPR) came into force last year. He decided not to transfer data problems, but drew the attention of users to this question. This does not oblige us to read all user agreements, but it makes us think and at least be more responsible for our digital footprints and follow the elementary rules of digital hygiene.
Valeria Karavaeva
data scientist at Spiking
We sometimes do not think about how many tracks we leave on the Web and how much later it helps companies, not only social networks - although social networks as well. Social networks collect data not only for themselves, they can sell them - I know about it, because I worked at an advertising agency, and we bought data from Facebook. And most often we, the users, give consent to this without noticing it. People spend half their lives on social networks and give a lot of information about themselves.
But it was possible to collect data before - so why have you started talking about big data only recently? First of all, because computing power grows and, accordingly, becomes cheaper. The main issue of big data is not how to collect data - in principle, each of us today can collect and store terabytes of information - but how to work with them. Most of the data obtained from social networks (text, voice, pictures, video) are not structured in any way, therefore without machine learning big data is useless. Now, due to the fact that power and memory have become cheaper, the demand for neural networks and deep learning has increased - we finally learned to process large data arrays.
Take, for example, pictures - and this is really big data, they can give a lot of information. There are millions of pictures, but what to do with them? What benefits can be extracted from them? What patterns do they let you know? Machine learning, in fact, is not so far gone. This is not such a simple process as it seems: there is no such thing that you press a button and in a week receive full calculations.
Directly machine learning is preceded by more complex tasks. The same pictures must first be properly processed (for example, cropped, centered photos; this is important for learning) - this is the first stage, which usually takes a long time. The second stage is to choose a network architecture suitable for solving the problem. Roughly speaking, you build ten different neural networks, and they give ten different results. Then you need to somehow evaluate the results. And after that you, with high probability, return to the first stage. It is impossible to build one universal network for any task: you either build it from scratch or modify an existing one. Face recognition is one task, the recognition of cats is another.
In the process of machine learning, we also participate, without knowing it. For example, introducing captcha on sites: using captcha, Google trained neural networks to digitize books.
We must understand that companies that collect big data are not interested in our personal profiles. They need data about a lot of different people who are interested in something specific. As for the special services, I think they can collect data without resorting to social networks. I think that our fears that we are being watched will soon pass. This is the new world: it is possible not to inherit the web, but it is difficult. It's easier not to appear on the Web at all.
PHOTO: antonsov85 - stock.adobe.com