SCORING
The board is worth 100 points.
The moments the internet remembers most are worth the most. Here is how each answer gets its value.
How board points are assigned
First, the eight eligible moments are ranked by Internet Prominence. We then divide exactly 100 board points among them based on their relative prominence. A moment that stands out much more strongly receives a larger share of the board. Close moments receive similar values.
Board values are rounded to whole numbers. Any leftover points are given to the answers with the largest decimal remainders, which guarantees that the eight boxes always add up to exactly 100.
Board points are relative to today’s category
Board points compare the eight answers in one puzzle. They are not the same as a moment’s separate Internet Prominence value out of 100. The prominence value helps establish the ranking and distance between moments; the board converts those differences into the points you can earn today.
What is Internet Prominence?
It is our evidence-based estimate of how strongly a specific moment lives on across the internet. Model v3 looks at four signals:
Each signal receives a score from 0 to 100. We multiply it by the percentage shown above, then add the four results. The result is the moment’s raw Model v3 score.
30% Video reach
This asks: how widely was the specific moment watched online? We collect the recorded view count and publication date from an approved official or verified video through the YouTube Data API. We save the source URL and the date the evidence was captured so the result can be checked later.
We use one representative upload instead of adding duplicate clips together. A clip centered on the exact play receives full credit. A performance compilation receives reduced credit, while a full-game or series recap is treated only as context. That prevents the popularity of an entire game from being assigned to one play.
Views are compared with other moments from a similar media era. Recent moments can be compared using their rate of views rather than a lifetime total. This keeps an older moment from losing simply because YouTube or modern social media did not exist when it happened.
30% Long-term recall
This asks: does the moment continue to resurface after its original news cycle? Evidence includes a later official retrospective, a dedicated historical page, or another saved source that revisits the specific moment years afterward.
The score rewards the length of time between the event and verified resurfacing, up to a fixed limit. A dedicated reference adds additional credit. Very recent moments have not had enough time to prove longevity, so they receive a disclosed score range instead of being assumed memorable or forgotten.
25% Category fit
This asks: how directly does the moment answer today’s prompt? Category fit is curator reviewed because no view count can determine whether an answer truly belongs in a category. The review uses a consistent ladder: an inseparable or defining association scores highest; a direct and widely recognized connection scores strongly; a relevant but shared or broad connection scores lower; and a weak connection is excluded.
Every answer must describe a distinct event. Two labels for the same underlying play cannot occupy separate board spaces, and a famous event cannot qualify merely because a well-known player or team was involved.
15% Event interest
This asks: how much lasting attention surrounded the larger game, series, contest, or season? We collect monthly English Wikipedia pageviews through the Wikimedia API, use the median month to reduce the effect of brief spikes, and compare that value with the other eligible events in the evidence set.
Event interest is deliberately the smallest component. A popular series can provide context, but it cannot by itself prove that one particular play was remembered.
A simple example
Suppose a moment scores 70 for video reach, 80 for long-term recall, 90 for category fit, and 60 for event interest. Its raw score is: 70 × 30% + 80 × 30% + 90 × 25% + 60 × 15% = 76.5.
The player-facing prominence value then compares that raw result with one fixed benchmark used across every category and caps the display at 100. The category leader is not automatically assigned 100.
How we verify the evidence
Each evidence run stores the source link, collection date, signal it supports, and whether it was verified. Official league, team, statistics, video-platform, and Wikimedia sources are preferred. Search results, unsourced claims, and a curator’s memory are not treated as measured reach or interest data.
A confidence label reflects how much of the weighted formula has verified evidence: complete coverage is high confidence, at least 80% is medium confidence, and anything below that is low confidence. Missing evidence stays visible as missing or provisional; it is not silently converted to zero.
Score 60 to pass
Correct answers add their board points. Wrong or off-board answers add one strike. Score 60 before your third strike to pass, then continue in Heat Check for a perfect 100. Broad, ambiguous, duplicate, and empty answers do not add strikes.
Performance tiers
0–39 Forgotten · 40–59 Almost Remembered · 60–74 Remembered · 75–89 Internet Historian · 90–99 Unforgettable · 100 Perfect Memory.
Evidence first, curator reviewed
The measured signals come from saved sources; the model does not invent viewership. Curator review is limited to category eligibility, category fit, overlap between answers, source quality, and unusual or incomplete results. Development puzzles remain clearly marked until their selected board answers have complete Model v3 evidence.