Article · 190 measured outlines
Why you can't draw a country from memory
You can probably recognise every flag in the world and place most countries on a map without hesitating. Ask for the outline, unaided, and the same person draws a wobbly blob. That collapse is not a memory problem — it is the difference between recognition and production, and it is the reason this game feels harder than it looks. What follows is why the failure happens, what our own measurements of 190 country borders say about which shapes are genuinely difficult, and the order of operations that reliably raises a first attempt into a second one.
Recognition is a cue; drawing is a construction
Nearly everything anyone ever did to teach you geography was a recognition exercise. You looked at a coloured map and matched a name to a shape; you were given a blank map and a word list and asked to write the words in; you took a multiple-choice quiz where the wrong answers were still shapes you had seen before. Every one of those tasks hands you the hardest half of the information and asks you to do the easy half, which is pattern-matching. Human pattern-matching on visual material is extremely good, and it makes all of that practice feel like knowledge.
Drawing removes the cue. Nothing is in front of you to match against, so the outline has to be built rather than identified — and what you actually have stored is not an outline. It is a small set of relationships: this country is above that one, that one has a coast on the left, the third is long and thin. Reconstruction from relationships gives you orientation and topology for free and leaves the geometry almost entirely underspecified, which is why the shape that comes out of your hand is roughly the right thing in roughly the right proportions and completely wrong in detail.
This is the single most useful thing to understand about the exercise, because it changes what you practise. The bottleneck is not coastline memory. It is that you are trying to draw detail before you have decided the proportions, and proportions are the one thing your stored relationships can actually supply reliably.
The order of operations that works
On this site the score compares filled areas after your sketch and the real border are both scaled and centred into the same box, so position and size are normalised away and aspect ratio is not. That scoring rule has a direct practical consequence, and the five-step method is built around it:
- 01
Decide the proportion before the pen moves
Is it wider than tall, taller than wide, or square? One guess, made consciously. A correctly-proportioned rough blob beats a carefully detailed shape that is a third too short, because the score is overlap and proportion errors destroy overlap everywhere at once.
- 02
Start from the one feature you remember
Nobody recalls a whole border, but nearly everyone can recall one: the straight edge across the Sahara, the boot, the horn of Brazil, the taper of Chile. Put that down and grow the shape outward in both directions from it.
- 03
Use one continuous stroke
Lifting the pen leaves gaps, and a gap makes the fill behave unpredictably. One stroke that returns to where it started is worth more than four careful segments that never quite join.
- 04
Ignore small islands and fiddly inlets
Tiny pieces contribute almost no area. Real coastlines are fractal; even a perfect hand cannot reproduce Norway's fjords, and you are not being asked to. Approximate the outer envelope.
- 05
Read the sector grid and redraw immediately
The nine-sector readout names the part of the outline you lost. Redrawing in the same sitting is where the correction actually lands — this loop is the difference between practising and playing.
What 190 measured borders say about "hard"
Every outline in the game comes from the same Natural Earth geometry, and each one carries a measured detail score: how much edge the shape actually has. It is the closest thing to an objective answer to "which countries are hard", and it disagrees with intuition in a very consistent direction. The median across all 190 outlines is 2.79, and the split by continent is not about how big the countries are or how well-known they are. It is about water.
| Continent | Countries | Mean detail | Under 2.0 | 5.0+ | Smoothest | Roughest |
|---|---|---|---|---|---|---|
| Africa | 53 | 2.82 | 19 of 53 | 3 of 53 | Eswatini | Cape Verde |
| Europe | 44 | 3.87 | 12 of 44 | 9 of 44 | Andorra | Norway |
| Asia | 45 | 4.24 | 10 of 45 | 12 of 45 | Singapore | Philippines |
| South America | 12 | 4.34 | 2 of 12 | 1 of 12 | Uruguay | Chile |
| North America | 22 | 4.50 | 5 of 22 | 8 of 22 | Barbados | Bahamas |
| Oceania | 14 | 8.21 | 1 of 14 | 10 of 14 | Nauru | Kiribati |
The four failures you can actually fix
Watch enough first attempts and the same four errors come up over and over, in a predictable order. They correspond to the four things a relationship-based mental map is bad at, and three of them are worth more points than any amount of coastline practice.
Detail before proportion
Drawing the fjords before deciding how tall the country is. Costs the most points of any single error, because a proportion mistake penalises the whole shape rather than one edge.
North drift
Pushing everything toward the top of the canvas and leaving the bottom empty. Harmless here — position is normalised away — but it is a tell that you are drawing features rather than a footprint.
Squeezing the width
Drawing every country taller than it really is. Aspect ratio is the one thing scoring does not forgive, and transcontinental states and long, thin shapes get stretched in the wrong direction most often.
Filling in one piece at a time
Stopping to add a peninsula, then a bay, then an island. The shape never closes, the fill misbehaves, and the result reads as several sketches rather than one country.
Start where the outlines forgive you
If you are rebuilding your mental map from scratch, sequence matters more than volume. The ten smoothest outlines in the dataset are the ones where a rough hand still scores well, and they are almost all compact, single-piece shapes — good for learning how the scoring responds. The ten roughest are where everybody caps out early, and they are nearly all islands or deeply indented coasts. Between them sits the useful practice range: big enough to place, simple enough to reward a first attempt.
| Country | Detail | Continent |
|---|---|---|
| Nauru | 1.12 | Oceania |
| Andorra | 1.14 | Europe |
| San Marino | 1.17 | Europe |
| Vatican City | 1.19 | Europe |
| Eswatini | 1.25 | Africa |
| Lesotho | 1.27 | Africa |
| Barbados | 1.32 | North America |
| North Macedonia | 1.34 | Europe |
| Singapore | 1.34 | Asia |
| Monaco | 1.35 | Europe |
| Country | Detail | Continent |
|---|---|---|
| Norway | 23.39 | Europe |
| Chile | 21.47 | South America |
| Philippines | 16.39 | Asia |
| Kiribati | 15.98 | Oceania |
| Solomon Islands | 15.66 | Oceania |
| Indonesia | 15.34 | Asia |
| Bahamas | 15.17 | North America |
| Marshall Islands | 14.96 | Oceania |
| Vanuatu | 14.75 | Oceania |
| Canada | 13.21 | North America |
Two numbers worth holding onto. First, difficulty is not size: Russia carries 17,098,242 km² of land and 7.50 detail, while Vatican City is a fraction of that and only 1.19. Second, concentration is extreme: the 7 largest outlines — Russia, Canada, China, United States, Brazil, Australia and India — hold about 50% of the 130,789,943 km² in the dataset, so learning to place a dozen shapes covers more of the world's sovereign land than learning forty small ones.
A twenty-minute drill
The routine below is the shortest version of what actually moves scores, and it works because it exploits the one advantage this task has over ordinary studying: the feedback is specific and immediate, so you never have to guess what to fix.
Pick five countries in one region, all in the same difficulty tier
Mixing continents wastes the loop — you are learning relationships between neighbours, and those only transfer within a region. The continent index pages list every country in a region sorted by measured detail, so the tier is easy to match. Europe is a sensible first region: 44 outlines, and enough of them are simple that you will not stall.
Draw all five, then stop and look at nothing but the sector grids
Do not check the headline numbers. Read which sector lost on each one and say the fix out loud: 'south-east under-drawn', 'too wide at the top'. Five diagnoses in a row is the information you are actually short of.
Redraw the same five immediately
Same countries, same session, no looking at a map in between. This is the step people skip and the step where the learning happens; gains between the first and second attempt are usually double-digit points on a single outline.
Write down the one relationship you got wrong on each
Not the shape — the relationship. 'Belarus is below Lithuania, not beside it' is a sentence you can carry; a redrawn outline you will not remember tomorrow. Ten of these sentences is roughly a mental map of Europe.
Come back the next day and draw the same five a third time
The overnight gap is what converts the correction into recall. If a shape is still wrong after three attempts, it is a piece of information you have never actually stored — which is useful to know, because it means the fix is looking at that outline once, deliberately, rather than drawing over it again.
What to expect on your first attempts
Set the bar correctly or the exercise discourages you in about four minutes. A first attempt on a country you have never drawn is usually a low score even if you know the country well, and on the roughest outlines that ceiling stays low: expect a hard cap on archipelago states, where the task is really "place fifteen shapes in relation to each other" rather than "draw a border". Chile sits at the 99th percentile of detail in this dataset despite being one of the most recognisable shapes on earth, because it is a thin ribbon with an island chain attached.
The scores that matter are the ones that move. A country going from the forties to the sixties between attempt one and attempt two is the signal that you stored something; a country that stays where it is after four attempts is a signal to look at the real outline on purpose once, then draw it — which is practice, and not cheating, if you then redraw from memory. Everything else on this site is built around that loop: the leaderboard is one shared country per day so attempts are comparable, practice mode covers all 190 outlines, and the country index sorts them by measured difficulty rather than alphabetically, so you can work in the order that suits your current level.
Related
Drawing countries from memory — questions
Why can't I draw a country from memory?
Because drawing asks for a different kind of memory than almost every geography lesson trains. Reading a map, naming a country and clicking it in a quiz are all recognition tasks: the shape is in front of you and you only have to match it. Drawing is a recall task with no prompt, so you are reconstructing an outline from a sparse set of relationships — which side of which neighbour, roughly how wide compared to how tall. Nobody stores a border the way a vector file stores it, so the first attempt at any country looks wrong to almost everybody. On this site the fix is not study, it is order of operations: proportion first, then one anchor feature, then one continuous stroke.
Is a bad memory for maps normal?
Yes, and it is so normal that researchers have a name for the systematic version of it: people consistently place countries nearer the pole they think of as "up", stretch the country they live in, and remember orientation far better than absolute size. The mental map is a rough diagram of relationships rather than a scaled picture, which is exactly why a blank outline quiz feels unfair. Recognition is near-effortless; production is the hard part. The gap between "I know where Albania is" and being able to draw it is normal, not a sign you are bad at geography.
Which country outlines are objectively the hardest to draw?
By measured border detail, Norway is the hardest outline on this site at 23.39, with Chile next at 21.47 — both are shapes made of deep inlets or many separated pieces. The 43 outlines at or above 5.0 are overwhelmingly island states and fjorded coasts, because water multiplies edge detail. Norway and Kiribati sit in that group even though their land area tells you nothing about difficulty.
What is the easiest country to draw?
Algeria, at 1.70 measured detail — the smoothest outline of the 190 in the dataset. That said, the smoothest outline and the most useful practice target are not the same thing: Eswatini, Lesotho and Barbados are also under 2 detail, and several of them are small enough that you learn nothing about their region by drawing them. For practice, pick a country under 2.8 detail that is also large enough to place — 19 of 53 African outlines are, which is why the continent is a good place to start.
How long does it take to get better at drawing countries?
The improvement that matters here is not measured in weeks but in attempts, because feedback is immediate and specific. Most people's second attempt at the same country scores noticeably higher than their first — typically ten to twenty accuracy points — not because anything was consolidated overnight but because the sector grid told you which part of the outline you lost, and you fixed that one thing. Doing ten countries twice in one sitting beats doing twenty countries once, for exactly that reason.
Does size make a country harder to draw?
No. Size and difficulty are separate measurements, and the data disagrees with the intuition every time. Russia is the 1st largest outline of 190 and a genuinely hard shape at 7.50 detail; Brazil is 8,515,767 km² and reads as one clean mass. At the other end, Andorra is the second-smoothest outline in the dataset and almost everybody puts it in the wrong country. What actually predicts difficulty is how much edge detail a shape carries and how many separate pieces it has — not how much land it covers.
Why are African borders easier to draw than European ones?
Because a great many of them were negotiated as straight lines and then drawn on the ground, rather than grown from rivers, mountain spines and millennia of local settlement. Measured on this dataset, Africa averages 2.82 border detail against Europe's 3.87 and Oceania's 8.21: 19 of 53 African outlines sit below 2.0 detail, and only 3 sit at or above 5.0. The exceptions prove the rule — the three African outlines with the most detail are island states and one country that is essentially a river valley.
How is drawing a country from memory different from a map quiz?
A quiz gives you the shape and asks for the name, or gives you the name and a dozen options; either way the answer is cued. Drawing supplies no cue at all, which is why it exposes gaps that a quiz hides. You can pass a European capitals test and still be unable to draw 44 country outlines, because the test never asked you to produce spatial information — only to recognise it. If your goal is to actually know where things are, production practice beats recognition practice, and it is also a much better predictor of whether you can find a country on a real map.
What does the accuracy score on this site actually measure?
Overlap, not penmanship. Your strokes and the real border are both scaled and centred into the same bounding box, filled, and compared by the share of area they agree on — so drawing off-centre or too small costs nothing, while aspect ratio is deliberately preserved because that is the thing being tested. A second figure shows which part of the outline you lost. That nine-sector readout is the useful part of the exercise: "62%" tells you nothing actionable, but "you under-drew the south-east" tells you exactly what to fix on the next attempt.