Recording and Evaluating Your Own Color Scores

A practical guide to tracking your daily color results, and reading them without over-reading them.

Short answer: the fastest way to find where your color judgement slips is to write the numbers down and look at the trend, not at today's result. A single day tells you almost nothing: attention, light, screen and mood all move a score. Kept over a few weeks, three things become readable: your bias (do you consistently guess too light, too warm, too saturated?), your consistency (how much your results scatter around your own average), and your direction of change (is the scatter narrowing?). None of these is a verdict on your eyes. They are rough, honest snapshots, and they are only as trustworthy as your screen and your lighting allow.

Why one score tells you nothing

Color judgement is noisy by nature. The same person, tested twice on the same task an hour apart, will not produce the same number, and that is true of formal clinical tests as well as games. Published norms for the Farnsworth-Munsell 100 Hue test, one of the most established arrangement tests in use, show a wide spread of error scores among people with entirely ordinary color vision, and a clear pattern of change with age (Kinnear and Sahraie, 2002). If a carefully administered test in a controlled room produces that much variation, a two-minute game on your own laptop will produce more.

The practical consequence is simple. Treat a single day's score the way you would treat a single coin flip. What carries information is the shape of many days together: where the middle of your results sits, how widely they scatter, and whether either of those is moving.

What to record

Keep it to a handful of columns, or you will stop doing it. A note app or a spreadsheet is plenty. The table below sets out what is worth capturing, how to get it, and how not to over-read it.

What to record	How to get it	What to watch for, and the limit
Raw daily result	The score or number of tries the game reports, one row per day.	Meaningless alone. It only becomes useful once you have ten or more rows.
Direction of error (bias)	After a wrong guess, note whether you were too light or too dark, too warm or too cool, too saturated or too dull.	A repeated direction is the single most actionable thing you can learn. One day's direction is noise.
Consistency (spread)	Look at how tightly your last ten results cluster, not at how close they are to a perfect score.	Some scatter is unavoidable. This is the measure least affected by a miscalibrated screen.
Threshold from Huefork	Record the threshold number the game reports at the end of the run.	Expect day-to-day wobble. Compare weekly medians, never yesterday against today.
Lock-score from Huedrift	The proportion of time you held the swatch inside the target zone.	Depends on input device and reaction time as well as vision. Keep the device constant.
Conditions	Device, room light (daylight, lamp, dark), time of day, whether night mode was on.	The column people skip and then regret. Most unexplained jumps live here.

A seventh thing worth noting by hand is the game itself. Comparing a Hueecho memory score with a Hueseam detection score is comparing two different abilities. Keep one row per game, not one row per day.

Reading the numbers without overreading them

Be gentle with the interpretation. A run of ten days is a small sample, and your attention varies more than your eyes do. What is informative is a pattern that repeats across weeks: a reliable lean toward guessing too light, a consistent struggle in the blue-violet part of Huewheel, a Hueecho result that is always biased in the same direction. One week can fool you; a habit that shows up again and again is worth acting on.

Expect memory-based games to be biased rather than merely imprecise. Remembered color drifts systematically toward the prototype of its category, so a color you recall as "that blue" will tend to come back more prototypically blue than it was (Bae and colleagues, 2015). That is a property of human memory, not a mistake you can practise away entirely, and our page on color memory and recall explains why. Reading a consistent pull in one direction as a personal failing is the most common misreading of these scores.

It also helps to separate accuracy from precision. Being consistently a little too dark is a bias, and biases are correctable once you notice them. Being all over the place is imprecision, and that responds to slower, more attentive play rather than to a correction. Our four-week color perception plan is built around exactly that distinction.

The calibration trap

Here is the honest limit. Your screen is not a measuring instrument. Two displays fed identical color codes can produce visibly different colors, and the difference is not small. When researchers evaluated a color vision screening test across ten ordinary commercial displays running in standard sRGB mode rather than calibrated, its sensitivity collapsed: observers with genuine, sometimes severe color deficiencies passed a test they failed on a calibrated screen (Evans and colleagues, 2025). If display variation can undo a purpose-built screening test, it can certainly move your score in a game.

Ambient light matters just as much. The same swatch on the same screen looks different under daylight, under a warm lamp, and in a dark room, because your visual system adapts to the surroundings rather than reading absolute values. This is the practical face of the appearance phenomena Fairchild catalogues: measured color and perceived color are not the same thing.

Practical defences are unglamorous but effective. Use one device throughout. Turn off automatic warm-tint and night-shift modes, which quietly move every color on screen. Keep the brightness setting fixed rather than automatic. Play in similar light at a similar time of day. Record the conditions when you cannot keep them constant. None of that calibrates anything, but it holds the error roughly constant, which is all a trend needs.

What these scores are not

A daily game score is not a color vision test. It is not an Ishihara plate set, it is not the Farnsworth-Munsell arrangement test, it is not administered under controlled illumination, and it cannot diagnose anything. A low score most often means a dim room, a tired evening, or a screen in warm mode. If you have a persistent, specific difficulty, particularly one that shows up in the red-green direction and has been there as long as you can remember, the right next step is an optometrist, not another week of practice. Our page on color vision differences and awareness sets out what is worth knowing and what is worth ignoring.

The same caution applies in a classroom. Educational measurement standards are explicit that a score means nothing without evidence that the instrument measures what it claims to, under the conditions in which it is used. A game score is fine as formative feedback, a prompt for a conversation about what the learner noticed, and useless as a judgement about a learner's vision. That distinction is the backbone of our assessment rubrics, and it follows the formative-assessment tradition that Black and Wiliam set out: feedback that describes what to do next changes learning, while a bare mark mostly does not.

Feeling accurate and being accurate are not the same. In Huefork, five rounds narrow in on how small a lightness difference you can actually see, and hand you a number you can write down. Ten of those numbers is the beginning of a real answer.

Play Huefork →

Frequently asked questions

How do I track my color scores usefully?

Keep one row per game per day with the raw result, the direction of any error, and the conditions (device, light, night mode on or off). Then look at weekly medians rather than daily numbers. Ten or more rows is the point at which the record starts to say something.

How many days before I trust what I see?

More than a week. Color judgement varies a lot from session to session even in people with completely typical vision, so treat a tendency as real only when it repeats across several weeks. A one-off wobble is not a pattern.

Does my monitor ruin the measurement?

It biases it. Uncalibrated displays vary enough that a purpose-built color screening test lost most of its sensitivity when moved onto ordinary commercial screens. Your scores are still useful as a personal trend, provided you keep the device, the brightness and the lighting as constant as you can.

Should I turn off night mode?

Yes, for any session you intend to record. Automatic warm-tint modes shift every color on the screen, usually by more than the differences the harder games ask you to detect.

Why do my memory scores always miss in the same direction?

Because that is how color memory works. Recall blends the specific shade with its category, so remembered colors drift toward a prototype. A consistent direction of miss in a memory game is expected rather than a personal fault.

Is a low score a sign of color blindness?

No. These games are practice tools, not clinical instruments, and a low score is far more often explained by lighting, tiredness or a warm-tinted screen. If you have a persistent difficulty, especially in the red-green direction, see an optometrist rather than reading anything into a game.

Can I use these scores to grade students?

Not as a grade. Educational measurement standards require evidence that a score measures what it claims to under the conditions of use, and a browser game on uncontrolled classroom screens does not meet that bar. Use the results as formative feedback and conversation prompts instead.