From Demos to Signal

AlgoArena has always had a simple promise: skill should be visible in the work itself. This release cycle was about making that promise easier to see from the outside.

What changed

We tightened the public product surfaces around live demos, clearer product pages, and assessment artifacts that explain what happened instead of just showing a final score. The Compete surface now does a better job showing the full loop: read the prompt, write code, run tests, submit, and compare progress in a live 1v1 IDE.

On the assessment side, we continued improving fuller judging, special-output checks, and evidence views that help reviewers understand the path behind a candidate's result. That matters because hiring teams should not have to guess whether a score came from strong reasoning, lucky tests, or shallow automation.

Why it matters

Public product pages are not just marketing polish. They are a trust surface. If the demo looks vague, the product feels vague. If the demo shows real constraints, real submissions, and real feedback, the story becomes much easier to believe.

That is why the homepage now points people toward specific lanes: [Compete](/product/compete), [Classroom](/product/classroom), [Assessments](/product/assessments), Builder, and [Vibecoding](/product/vibecoding). Each lane has a different audience, but the same underlying goal: make the process observable.

Where it points

The next step is evidence that travels cleanly across the product. Practice should teach. Classrooms should reveal where a group is stuck. Assessments should explain why a candidate's work is trustworthy. The [benchmark](/oa/benchmark) direction is part of that same arc: not just prettier demos, but measurable signal.