Perceptle Count: dots, estimates and moving dots

1. Settle your view before starting the flash.
2. Look for a small group you can recognize quickly: five or ten dots, or a manageable patch of the field.
3. Use that group to estimate the whole, allowing for uneven density.
4. Enter one whole-number guess. After submitting, compare the actual count with your guess.

Scanning the entire board sequentially often runs out of time. Grouping gives you a consistent way to estimate even when every dot cannot be counted.

Suppose you see roughly six groups of ten dots: start near 60. If two groups look closer to eight than ten, reduce the estimate to about 56. Another approach is to divide the field into quarters: a quarter containing 15 dots suggests about 60 only if the other quarters are similarly dense.

The picture below is a teaching arrangement of 60 dots, organized into six groups. Daily dots are not necessarily laid out this way. Use grouping as a visual estimate, not as an assumption that every area has the same count.

Friday dots move and bounce. The displayed instruction says about 2½ seconds; the reviewed public generator configures a 2,600 ms exposure. It is still one fixed set of dots. A dot seen on the left and later on the right is not two dots.

Use one broad impression or estimate the density of a fixed region. Do not add counts from successive snapshots. The original Friday tolerance is 32% relative error instead of Normal’s 20%.

Count error is |guess − actual| ÷ actual. With 60 actual dots and a guess of 66, the error is 6/60 = 10%. Normal converts that to about 88 points. With 45 actual dots, the same six-dot miss is about 13.3% and scores about 80.

This is why raw dot differences alone do not compare every puzzle fairly. Change the actual count and estimate in the error calculator to see the relationship.

• Consistently low: check whether sparse-looking gaps make you discount crowded areas too much.
• Large alternating misses: keep the same grouping size for a few attempts instead of switching strategy each time.
• Friday overcount: compare a single population estimate, not every frame of motion.
• Keyboard in the way: finish Count and dismiss the keyboard before starting Track.

Try five fresh counts, record actual and guessed values, then compare the middle relative error. One perfect score is less informative than a smaller typical miss. The companion Count exercise gives repeatable independent Normal practice, rather than the same daily target.