Built by Reid M Nelson
I'm a mechanical engineering student who built Flashix as a personal project sitting at the intersection of two things I find genuinely interesting: web development, and how the human brain processes the world around it.
The premise started simple — I wanted to know where my own brain draws the line between knowing a quantity and guessing it. That boundary turns out to be surprisingly sharp, surprisingly personal, and surprisingly trainable. So I built a game around it.
My goal was to make something that's actually fun to play but also quietly reveals something real about your perception. The leaderboard and Daily Challenge came later, once I realized that competing on something this instinctive is its own kind of compelling.
Flashix is entirely self-built — vanilla JS, no frameworks, no third-party analytics. Just a canvas, a timer, and your brain.
What is Flashix?
Flashix is a browser-based estimation game that tests how well your brain counts under pressure. A field of dots flashes on screen for a brief window of time. Your job is to guess how many there were.
It sounds simple. It isn't.
At low counts, your brain handles it effortlessly — you just know. At higher counts, you're working with something messier: a rough, probabilistic sense of quantity that degrades as the numbers grow. Flashix is designed to probe exactly that boundary and measure where it sits for you.
The leaderboard and Daily Challenge add a competitive dimension — everyone sees the same dots, so your score is a genuine comparison, not a fluke of luck.
The Approximate Number System
Your brain has two fundamentally different mechanisms for perceiving quantity:
For small quantities — roughly 1 to 4 items — you don't count. You don't estimate. You just know, instantly and accurately, with no conscious effort. This is subitizing, and it's essentially perfect up to its limit.
Beyond your subitizing threshold, your brain switches to a different system — a fast, parallel estimation engine that gives you a noisy approximation of quantity. The key property of the ANS is that it works on ratios, not absolute differences. Distinguishing 10 from 20 is as easy as distinguishing 100 from 200.
The boundary between these two systems — your subitizing threshold — is typically around 4 or 5 for most people. But it varies, and it can shift with training. The Threshold Finder mode is specifically designed to locate yours.
Your ANS Score (shown on the Stats page) reflects how accurately your estimation system performs across the full range of counts you've seen. Higher is better — and it's trainable.
Game Modes
Flashix has three active modes and three more in development. Each one is designed to test a different aspect of numerical perception.
Classic
The core experience. Ten rounds at a difficulty you choose. Dots flash, you guess, you're scored on accuracy using a curve that rewards precision and punishes large errors progressively.
Use Classic to practice, benchmark yourself, or chase a leaderboard rank. Each difficulty tier changes both the dot count range and the flash duration — harder tiers also add visual noise like clustering, size variation, and overlapping dots.
Your score for each round is calculated as a power-curved percentage of accuracy: a perfect guess scores 100%, and errors compound as the gap grows.
Daily Challenge
One globally-seeded game per day. Every player in the world sees the exact same dot fields — the layout is determined by a daily seed, so there's no luck involved in the comparison.
Five rounds. Difficulty is selected randomly per round with a bell-curve weighting (medium and hard are most common; beginner and inhuman are rare). You won't know the difficulty ahead of time, but a badge shows it during the countdown.
Your result is shareable as a Wordle-style emoji block. Daily Challenge scores feed into a separate leaderboard that resets each day, so the competition is always fresh.
Threshold Finder
A 20-round diagnostic built around one question: where exactly does your brain switch from knowing to guessing?
The sequence probes specific dot counts in a structured order — starting small, ramping up through the likely threshold zone, then revisiting lower counts to confirm. The progression is:
After completing Threshold Finder, your Stats page will show a calibrated subitizing threshold — the highest count at which you consistently score above 90% accuracy.
Difficulty Tiers
Classic mode offers six difficulty levels. Higher tiers increase both the dot count and the visual complexity — not just fewer milliseconds.
The Inhuman tier is the only one with a reduced flash window — 2.5 seconds vs 5 seconds for all others. At 750 dots, even 5 seconds wouldn't be enough to count.