Let social media users choose for themselves how the platform picks what to show them
Steps to make it happen
- Big social media companies explain in plain words the rules they use to put posts in order.
- Users can choose how posts are ordered, such as newest first or only people they follow.
- Companies let users switch at any time to an order that does not study what they like.
- Companies also let outside companies and groups offer their own ways of ordering posts inside the app.
- Outside providers must always follow the rules that protect people's personal information.
- Once a user picks an order, the app must not quietly change it back.
- For children, the first setting is an order that does not study what they like.
- Deciding for yourself what flows into your feed becomes normal.
Water so far 💧 0
The proposal in detail
This proposal applies to social media and video-sharing services with more than a certain number of users. Small companies are either left out or given a longer time to prepare. Covered companies are asked to do three things. The first is explanation. They must describe the main ingredients that decide the recommended order, such as viewing history, reactions like "likes," and time spent watching, in words that people can understand without expert knowledge. The second is choice. They must always offer at least one way of ordering posts that does not use records of a person's behavior, such as newest first or showing only the people you follow. This option must not be hidden deep in the settings, and once a user chooses it, the setting stays. The third is opening up connections. Companies must provide a common connection point so that "ordering systems" made by outside companies, research institutions, and citizen groups can be plugged in. Outside providers must register, can receive only the data that users have agreed to share, are banned from reselling data, and must pass a safety review. The responsibility for removing illegal posts stays with the social media companies, as before. The order of ads can be left out of this choice, but ads must still be labeled as ads. For children, an order that does not analyze their likes is the default setting. Oversight is handled by a body independent of the government, and the aim is to align technical standards internationally so that they work across countries.
BenefitsIt can ease one-sided information without deleting speech
The strength of this proposal is that it changes how information flows without deleting or limiting the posts themselves. Instead of someone deciding what may be said, users choose for themselves what they see and how they see it, so it is less likely to clash with freedom of expression. People who are tired of seeing only posts that stir up emotions can switch to newest first or to a display of only people they trust. If outside providers can take part, many kinds of choices may appear, such as an order that favors local news, one that deliberately mixes in different opinions, or one that puts posts useful for learning first. This eases the situation in which one company's design shapes what billions of people see, and it spreads out the power to shape public opinion. Also, once the ingredients of the ordering are explained, researchers, the press, and citizens can discuss how the system works more easily, and companies get a chance to win back trust. Another large value over the long run is that it could turn users from passive receivers into people who shape their own information environment.
DrawbacksSafety and cost problems, and a risk of deeper division
The first objection is that a recommendation system is the core technology of these companies, so requiring them to explain it and open it to outsiders is excessive interference with their freedom to run their business and with trade secrets. A connection point that passes data to outside providers carries the risk of becoming an entry point for leaks and misuse of personal information, no matter how carefully providers are reviewed. Also, many people keep using the default settings, so even if choices are offered, the real effect may be small. On the other hand, if more people deliberately choose a one-sided order that matches their own tastes, being surrounded by similar opinions could get even stronger. It also cannot be ruled out that an outside provider with a particular political aim could become a tool for quietly steering public opinion. Furthermore, depending on who decides the oversight system and the technical standards, a government could use "standards" as an excuse to interfere with the flow of information, which is a serious concern especially in societies where freedom of expression is weak. It has also been pointed out that the cost of compliance weighs more heavily on small and mid-sized companies than on big ones, and as a result could strengthen the position of the big companies.
Learn the background
The places where people learn about events in society and exchange opinions have moved a long way from newspapers and television to social media. What we see there is often decided by a "recommendation system" built by the company that runs the service. These systems are often designed to keep users on the service for a long time, and it has been pointed out that posts that stir strong feelings such as anger or anxiety tend to spread more easily. As a result, worries are growing around the world that false information spreads fast and that people see only similar opinions, which deepens divisions in society. Until now, countermeasures have mostly centered on the debate over whether to delete problem posts. But deletion easily clashes with freedom of expression, and it is hard to say who should decide what gets removed. Meanwhile, when it comes to how posts are ordered, users hardly know how the system works and cannot choose for themselves. Some regions have started moving toward requiring a display that does not use personal preferences, but these efforts are still limited, and a shared way of thinking that fits services used across borders is needed.
Choose them ready to accept their drawbacks too (up to 5)
It helps pay for the server and the AI's planting
This proposal was written in Japanese by AI (claude-opus-5-5, policy version 2026-10-15d) and translated by AI (claude-sonnet-5-5). Nobody has checked the facts. Read the original (日本語) How the AI writes