A screen does not simply show what exists. It orders, selects and repeats content through technical and commercial choices that are rarely visible. Failing to look at how recommendations, data collection and advertising work leaves families and young people with an incomplete explanation: every video, news item or product appears to have arrived by chance, when it often answers calculated incentives. Critical literacy does not require distrust of everything or specialist knowledge. It requires enough questions to keep judgement, protect privacy and decide more freely.
What is visible is not all that decides#
Platforms present a simple interface: a list of videos, a search, a message or a news item. Behind it are systems that prioritise some options over others. They may consider earlier interactions, watch time, contacts, approximate location, device type and inferred signals. This does not mean a system knows a person completely or that every recommendation is malicious. It means the experience is not neutral. Someone who sees only the screen may attribute to individual taste choices that have been shaped by design.
Platforms present a simple interface: a list of videos, a search, a message or a news item.
This distinction matters especially when content repeats. Repetition can create a misleading sense of consensus or urgency: if something appears many times, it seems important; if others react, it seems confirmed. Yet visibility may reflect retention logic, an account's capacity to pay for reach, or a temporary trend. Before treating repetition as evidence, it is worth separating popularity, evidence and public relevance. They are different things, and mixing them makes error easier.
Recommendations pursue objectives#
A recommender needs a measurable objective. Clicks, time spent, reactions, purchases or the likelihood of returning are often optimised. Those indicators can help organise a large catalogue, but they do not equal quality, health, learning or social value. A video provoking outrage may retain attention better than a qualified explanation; an impulsive purchase may be more profitable than an informed decision. Choosing the indicator is a design decision with consequences, not a natural law of the internet.
Understanding this avoids two extremes. One is to think that people retain no ability to choose. That ability exists, but it operates in an environment that makes some routes easier and others harder. The other is to blame an individual for every click. Companies design alerts, autoplay, counters and sequences to reduce pauses. Shared responsibility means recognising the architecture and creating practices that return time for decisions: turning off needless alerts, actively seeking different sources and pausing before buying, forwarding or following a recommendation.
Data are not an administrative detail#
Data can provide services, detect faults and personalise options, but they also build profiles and predictions. An app can combine declared information with observed activity: schedules, interests, contacts, searches or reactions. Even where a complete identity is not known, sets of signals can reveal habits or vulnerabilities. Accepting a privacy policy without reading anything is therefore not mere formality. The practical question is what information is requested, why it is used, how long it is kept, who receives it and what controls exist.
Protection does not depend on a single tick box. It helps to review permissions, limit location where it is unnecessary, separate accounts, use strong passwords and avoid publishing information that can enable impersonation or pressure. A private setting does not remove all data uses. Data-protection rights provide tools to access, correct, erase or object to certain processing, but those tools work better when the service is understood and clear information is consulted. The Spanish Data Protection Agency offers guidance for minors and families.
The business model explains part of the message#
Asking who pays for a service is not cynicism; it is context. A product that appears free may be financed by targeted advertising, subscriptions, complementary services, commissions or data processing. Each model creates different incentives. Advertising may reward sustained attention; a subscription may reward continuity; a commission may promote certain transactions. No model automatically makes a tool harmful, but hiding it prevents a realistic assessment of its limits.
Advertising must also be recognised when it dresses as a recommendation, review or entertainment. Sponsored-content labels are a signal, although they do not always explain the commercial relationship. A family can learn to ask what is being sold, what information is missing, and whether a claim about wellbeing, health or performance comes from an independent source. In matters affecting bodies, money or rights, an attention-grabbing promise deserves an additional check through public bodies, qualified professionals or original documents.
Critical reading can be practised#
Critical literacy works better as a routine than as a lecture. Five questions can be asked of relevant content: who published it, what they seek to achieve, what data they use, which primary source supports the claim, and what would be absent if the algorithm showed only this. Answers will not always be immediate. That is why tolerating uncertainty and delaying a decision is valuable. Saying “there is not enough information yet” is more responsible than filling a gap with a quick conclusion.
Lateral comparison helps. Instead of remaining inside the same app, it is useful to open an institution's site, locate the cited report, check its date and see whether there has been a later correction. If a video summarises a study, it is worth finding who carried it out, how many people were included and which limitations it recognises. This does not ask every home to conduct academic research. It prevents persuasive presentation from replacing evidence. Schools, libraries and responsible media can support this learning, but should not replace family conversation.
When young people are involved, adult duty increases#
Adolescence is a period of exploration, belonging and identity-building. That does not make young people incapable, but it does make it unfair to delegate to them the entire burden of understanding complex persuasive systems. Adults can discuss recommendations without making an accusation: a sequence does not define a person's worth, a trend does not compel participation, and asking for help does not mean losing all privacy. An explained, reviewable boundary protects better than opaque surveillance.
Some signs call for additional attention: pressure to share images, messages demanding secrecy, unexpected spending, sharp mood changes, content normalising self-harm or violence, or contacts trying to isolate a young person from support. It is not appropriate to diagnose from a screen or dramatise every use. It is appropriate to listen, preserve evidence where risk exists and use suitable resources. INCIBE, education, health and protection services can guide action; immediate danger requires urgent help.
Autonomy needs conditions#
Digital autonomy does not mean leaving everyone alone in an attention market. It means having understandable information, time to think, reasonable settings and support to repair an error. Companies must provide transparency, meaningful controls and enhanced protection for minors. Institutions must oversee abusive practices and enable media education. Families can create conversations in which questions are not mocked. No single part is sufficient alone.
Looking at how a platform works does not remove the pleasure of using it. It makes use possible without confusing what is convenient with what is inevitable. A person who identifies a recommendation, recognises an incentive and seeks an independent source is better placed to decide. This competence is modest: it does not promise absolute certainty, but it reduces the power of manipulation and supports a freer relationship with technology.
Explaining the system without turning it into a threat#
Conversation works better when it is grounded in observable facts rather than labels. It is possible to examine why a post appears, which permission an app requests, or where a promotion is disclosed without portraying technology as an unavoidable force. That approach avoids both uncritical enthusiasm and paralysing fear. It also makes room for valuable uses: a platform may support learning, connection or creativity while still embedding commercial choices that deserve scrutiny. Digital maturity is the ability to hold both ideas at the same time.
At home or in a classroom, a short check can be more useful than a general lecture. When an app is installed, it is reasonable to review which permissions are essential to its function and which can be refused. When a purchase recommendation arrives, it is possible to see whether the profile issuing it discloses a collaboration. When a striking number is shown, the institution that produced it and the publication date can be located. Repeating these small acts turns caution into a practical, shared skill rather than an abstract warning.
There should also be room for error and repair. No one notices every incentive or interprets every privacy notice correctly on the first attempt. If someone shares something without checking it or accepts an unnecessary permission, the helpful response is to review what happened, correct what can be corrected and learn a pattern for next time. Shame and blame encourage people to conceal problems; a calm response makes it easier to ask for help and verification when it truly matters.
Sources for further checking#
Spanish Data Protection Agency: minors and privacy; INCIBE: minors and Internet Segura for Kids; European Commission: Digital Services Act; UNESCO: media and information literacy.
