We know rankings work
Let's be honest first: a leaderboard is an excellent growth device. Rank ignites competition, top players screenshot their scores, everyone else retries. It's gamification 101, and of course we considered it.
We didn't skip it to be contrarian. We skipped it because a browser-based reaction-time ranking forces you to lie about your own measurement. Two reasons.
Reason 1: Tens of milliseconds of device variance — fair competition is physically impossible
Browser measurements carry latency that has nothing to do with your reflexes: input-device lag, display refresh rate (one frame at 60Hz weighs ~4× one frame at 240Hz), browser paint timing, OS load. Summed up, environment-driven offsets can reach tens of milliseconds.
Individual and state-dependent differences in reaction time live in exactly that range. A 20ms gap on a leaderboard cannot be attributed to the person or to their monitor — they are indistinguishable. Publishing that gap as a ranking means faking precision you don't have. A board where gaming-monitor owners cluster at the top is hardware bragging, not measurement.
Reason 2: Cheating can't be prevented — and trying costs something worse
Anything that runs in a browser can be manipulated from the developer console: spoof the timer, script the clicks, post a theoretical score. None of it is hard. Our test even ships a public verification hook (we use it for automated testing; hiding it would achieve nothing).
"Validate on the server, then." But there is no fundamental way for a server to prove the authenticity of numbers a browser sends. The stronger countermeasures — behavioral logging, fingerprinting, constant surveillance — collide head-on with our local-first principle of not collecting your data. Building a score-collection pipeline to police a free test would have been the tail wagging the dog.
What we built instead: self-reference, not competition
- 1-on-1 challenge links — your score travels in the URL you send a friend; our server never stores it. Yes, the number can be tampered with — but with no prizes and no rankings, there is little reason to bother. The fun of competition, kept; the false precision claim, dropped.
- Comparison with your previous self — history is stored only in your browser (localStorage), and the test shows your delta against last time and your recent average. Same device, same environment — the variance problem largely cancels out. What a ranking would fake, self-comparison keeps honest.
You may notice this mirrors the design of AXIOM itself. Absolute reaction speed compared against strangers tells you little. What matters — and what stays statistically honest — is how far you are from your own usual. The free test is a miniature of the product's philosophy.
Honesty as a spec, not a constraint
A reaction test without a leaderboard is objectively worse at spreading. But a measurement tool that fakes the limits of its own measurement has nothing left to offer. Our test page states up front what it cannot know — the exact on-screen moment of the stimulus, dropped frames it failed to catch. No leaderboard is just one more decision on that same line.
If you want to race someone, go ahead — just know what the number contains: your reflexes, your monitor, and your browser's mood. The most honest opponent we can offer you is yesterday's you.
Author: PRO ORDER
Developer of AXIOM, a cognitive performance measurement tool (sole proprietor). Building objective measurement tools for traders and poker players with Tauri + Rust.