Scoring

Your target relative to its starting count sets your points. A miss costs 100.

The same rule for both metrics

A post starts at 25,000 views. You predict 100,000 views, or 4×. A hit earns 200 points.

A post starts at 500 likes. You predict 1,500 likes, or 3×. A hit earns 158 points. Either prediction loses 100 points on a miss.

A hit needs an official X count for your selected metric recorded at or above the target within exactly 24 hours of acceptance. Results wait for the deadline and final checks. REACH samples counts; it does not continuously track every view or like.

You choose the target

Any whole-number target or decimal multiplier within 2–64× is valid. These examples are not presets:

Metric	Starting count	Target	On hit
Views	25,000	73,000	+154
Views	25,000	100,000	+200
Likes	500	1,500	+158

Only relative growth of the selected metric affects hit points. Views do not add a bonus to a likes prediction, or vice versa. There are no follower, popularity, or entry-speed bonuses. Potential hit points are fixed on acceptance.

How points are calculated

Hit points = floor(100 × log₂(target ÷ starting count))
Miss = −100 points

A hit earns 100–600 points. Each doubling of your target adds 100 hit points. The result is rounded down to a whole point. New predictions use reach-v3.

Your score

Your season score adds the settled points from predictions accepted into that season, across views and likes. Hits add points; misses subtract 100 each. Void predictions add zero. Scores can be negative.

Example
Views hit at 4×: +200
Likes hit at 3×: +158
One miss:       −100
Season score:    258 points

Only final results add points. Active predictions and potential hit points add none, so a player with only pending predictions has zero settled points. Your lifetime score and prediction history remain separate and are not cleared between seasons. Points determine season rank; they are not a transferable balance. See the published reward plan for prize eligibility and funding.

Season timing

Season 1 starts as an unscheduled draft. Once REACH starts it, there are 24 hours for entries and another 24 hours for accepted predictions to finish: 48 hours in total. Each prediction gets exactly 24 hours. A prediction accepted just before entries close finishes just before the season’s end.

The season remains in Settling after entries close. Finalization waits until the scheduled end plus the existing five-minute final-check allowance, and until every season prediction has a final result. Processing delays or a prediction under review can keep it in Settling longer. Those delays do not extend any target window.

A prediction’s season is fixed at acceptance. An on-time reading processed later still contributes to that same season. Once all results are complete, REACH saves the final rankings as an immutable snapshot.

Season leaderboard

The public leaderboard shows the official season’s Top 25. Ranks are calculated across the full season before limiting the list to 25 players. Search Username can therefore show a player at rank 950 without changing that list. Test seasons are excluded.

Ranking uses settled points from highest to lowest, then hits from highest to lowest, then a stable internal account ID to order any remaining tie. This ordering fixes equal-score positions without adding points. The current board has no Today, 7-day, or All-time tabs.

Search Username looks up the exact handle you submit. It does not sign you in or claim that the account is yours. Sign in with X to see Your rank automatically. X Profile opens X; Prediction History shows the recorded source post and accepted reply. Historical predictions and lifetime results remain stored.

Measurements and exceptions

When counts are checked

Each prediction has at most five read attempts for its selected metric. Times below are measured from acceptance.

Attempt window opens	Window closes
6 hours	Before 12 hours
12 hours	Before 18 hours
18 hours	Before 23 hours 59 minutes
23 hours 59 minutes	Before 24 hours
24 hours · final integrity check	24 hours 5 minutes, inclusive

There is at most one attempt in each window. A failed attempt still uses that slot, and skipped windows are not replayed later. API availability, rate limits, and budgets can prevent a read. The recorded capture time, not the planned checkpoint, determines whether evidence is on time.

A brief target crossing between readings may go unrecorded. REACH does not infer it from a later count or claim continuous observation. Official X counts do not prove organic attention or unique people.

The final check

The final observation must be recorded from the deadline through five minutes afterward. This allowance is for processing and integrity checks; it does not extend the 24-hour prediction window. Evidence captured on time can be processed later without changing its timestamp.

A recorded hit between acceptance and the deadline, inclusive, remains a hit if a later count falls, provided the final check is available and the integrity rules pass.
Without an on-time hit, a final count below the target but at least the starting count is a miss: −100 points.
If a count first meets or exceeds the target after the deadline, its timing is uncertain: void, 0 points.
Without an on-time hit, a final count below the starting count is unusable: void, 0 points.
An unavailable final source or no usable final observation within the allowance makes the prediction void, even if an earlier reading reached the target.

Edited and deleted posts

A recorded edit to the accepted X command incurs −100 points. A recorded source edit voids the prediction for 0 points. Both override a recorded hit; if both were edited, the command penalty takes precedence. Deleting the command does not cancel the prediction.

Timing example

A post starts at 25,000 views and the target is 100,000. An official reading records 120,000 at hour 18. With a valid final check and no edits, the result is +200 even if the final count is 90,000. Without an on-time reading that met the target, a final count of 90,000 is −100. If the first qualifying reading arrives at 24 hours 1 minute, the result is void because it cannot establish when the target was reached. The same timing rules apply to likes.

Earlier scoring versions

Read the version on your receipt. Existing reach-v1 predictions keep their original rule: the first usable official count captured from the deadline through five minutes afterward determines hit or miss. Existing reach-v2 predictions use views and the sampling rules above. New reach-v3 predictions fix a views or likes metric at acceptance. Earlier records are not regraded or changed to another metric.