@Mapzimus
Infographic maps and spatial data visualizations.
Published Work
Maps, infographics, and data visualizations. More coming soon.
Content is on the way. Follow @mapzimus to see it first.
Viral Map Idea Generation
A custom scoring engine that ranks 19,852 candidate map ideas by predicted virality. The V-Score is the centerpiece — a weighted composite of seven dimensions, calibrated across five iterations against what actually performs on Reddit and Twitter.
01 Why a score at all
A database of 19,852 candidate map ideas is useless without ranking. Gut-picking doesn't scale, and "interesting" is too lossy a signal. The V-Score is a 0–100 virality estimate computed from seven weighted dimensions, with a data-readiness penalty and a format bonus — designed to surface the ideas that are both worth making and ready to make.
02 The formula (v5)
Each idea is hand-scored 0–100 on seven dimensions. The raw score is a weighted sum, normalized by the sum of weights (11.75) back to a 0–100 range, then multiplied by a data-readiness penalty and offset by a format bonus.
raw = emotional·2 + relatability·2
+ clarity·1.25 + surprise·1.5
+ tension·1.5 + visual·2.0
+ originality·1.5
vs = ⌊raw / 11.75 · penalty⌋ + bonus
penalty = 1 − 0.5·(1 − data_ready/100)
bonus = +3 choropleth, +2 bivariate/dot
03 Calibrating the weights
The first four versions of the V-Score treated all dimensions roughly equally. They were wrong. v5 reweighted against observed engagement:
- Visual → 2.0 — highest correlation with engagement, the #1 scroll-stopper.
- Tension → 1.5 — controversy drives shares and comments.
- Originality → 1.5 — uniqueness prevents scroll-past.
- Clarity → 1.25 — knocked down: low variance across ideas means it barely differentiates winners.
04 The data-readiness penalty
Earlier versions let half-baked ideas outrank shovel-ready ones just because they scored high on emotion or surprise. v5 steepened the penalty: an idea with data_ready = 60 only earns 80% of its raw score; data_ready = 20 cuts it to 60%. The result is a leaderboard biased toward what can actually ship this week, not what would be cool if you had three months and an unreleased dataset.
05 The format bonus
Two visual formats have a proven track record on Reddit and Twitter map subs: choropleths (+3) and bivariate / dot maps (+2). v5 added these as explicit additive bonuses on top of the normalized score — recognizing that "how it's drawn" can be as decisive as "what it shows." A mediocre choropleth often outperforms a brilliant scatter plot in the wild.
06 What the board does with it
Beyond the score itself, the board filters 19,852 ideas by geography, topic, output format, type (MAP / XREF / CHART / RANK), and bonus category. The correlation engine surfaces ideas that share datasets or joinable variables. Status tracks each idea from concept through In Progress, Built, and Published. The whole system is one human (me) trying to be honest with myself about which idea is actually worth making next.