We write algorithms. But are we ready to be responsible for their consequences?

Each of us at Codeforces is used to thinking about how to optimize code, how to pass all the tests, how to get a rating. But the reality is that the "tests" of algorithms in life are different: it is what children watch on YouTube, what news the society reads and what decisions companies make. An error in these algorithms costs more than a Wrong Answer.

Let's take Spotify as an example. Its recommendation algorithms are built on collaborative filtering, audio analysis, and language models. The goal is simple: to keep the user in the app longer. In practice, this leads to a narrow feed — we hear the same thing, and young artists are drowned in noise. Unfair? Yes. But the risk is limited to music and cultural diversity.

Now YouTube. The algorithm is the same principle: maximize "engagement". But here the price of error is higher. Imagine a child watching a scientific video. The algorithm quickly finds similar "scientific" videos, only about UFOs or "secret experiments". An hour later, the child is sure that this is true. The result is not just a narrowing of tastes, but an undermining of trust in science, cognitive distortions, and the mass dissemination of misinformation.

Same algorithm. Two platforms. Two realities. And if we're losing new artists on Spotify, we're risking losing an entire generation of critical thinkers on YouTube.