About yoteki

There is always more news than time. An ordinary feed answers the question of what happened and leaves you the harder one: which of it reaches me. yoteki is being built to answer the second.

The product has one goal: pick out the events that actually touch a person's life, and explain what that effect is. Not important in the abstract, but important because it changes prices, rules, work, safety, money, or the next few months of your plans.

How events are chosen

Every day we read the full stream of reporting from a fixed list of outlets and score each event on three questions: how big it is in itself, how many people it really touches, and what it sets in motion. Only what clears a threshold is shown. Some days that is thirty stories, some days three. When nothing clears it, the feed says so instead of filling the space. The details are in the methodology.

The middle question is the one about you: being affected is not the same as being aware. A broadcast's audience is not its scale, and a statement, a forecast or unfinished negotiations bind almost nobody to anything.

How the effect is explained

Some of this already works. A story carries a gist and a list of what became known since last time. Beside events with a measurable future stands a probability from a prediction market with a link to the market itself: forecasts are quoted, never issued, and we give no advice. Storylines carry a section called “What it means for you”, written by a curator, next to indicators and the events coming up.

Honestly about what is missing. The explanation lives at the level of a storyline, not of every single story. It is the same for every reader and knows nothing about where you live or what you do. Tuning the feed changes the order, not the text. That is what we are working on next.

What it is not

• An infinite feed. Done reading, free until tomorrow.
• Advertising or notifications.
• Scores on display. Order is the only signal: the higher a story sits, the more it matters.
• Hidden stories. Personal tuning reorders and adds, but never hides what cleared the threshold.

How it is made

• Sources are the RSS feeds of named outlets. Every story names the outlets behind it, and its page links to the original articles.
• The summaries are written by a language model from those articles, against a fixed rubric. The scoring arithmetic is computed in code rather than by the model, and the formula is deterministic and versioned.
• Every story has one permanent address that never changes and updates in place as the story develops. Day archives keep the feed exactly as it stood at noon UTC.
• English is the original. The Russian edition translates stories by machine; storyline pages are written by a curator. Until a translation lands, the Russian page shows the English text.
• The feed and every topic are available as RSS, every story and storyline as markdown by adding .md to its address, and a short description of the site for machines lives at /llms.txt.

Tell us what to fix

This is being built, and reader suggestions decide what gets built next. Tell us what a story left unclear, what impact we failed to explain, or what is missing.