Reported players
Current player count, scaled logarithmically and capped at the candidate set’s 95th percentile.
Methodology v1.0
A useful ranking should
show its work.
ServersRank combines independent signals from AnyServer only when they are comparable across nearly the entire candidate set.
Votes
AnyServer votes, scaled the same way so a very large raw count cannot dominate every other signal.
Review quality
Average rating pulled toward the cohort average using 20 prior reviews, reducing the effect of tiny samples.
Historical uptime
The upstream uptime proportion, used only when AnyServer identifies one consistent measurement period.
Eligibility
Comparable data first
A component turns on only when valid values cover at least 95% of the full candidate set. Reviews require both an average and a review count. Uptime also requires one known measurement period shared across the set.
Enabled weights are rescaled to total 100%. A server missing one enabled input receives an “insufficient data” label rather than a misleading score.
Fallback
When the score steps aside
Balanced scoring requires at least two eligible components. If fewer are available, ServersRank orders servers by reported players and says so above the list.
Score ties break by votes, then reported players, then the stable AnyServer ID.
Boundaries
What is excluded
Paid placement, bumps, backlinks, and listing completeness never affect the score. Filtering and pagination happen after the score is calculated.
ServersRank requests up to 5,000 Java and 5,000 Bedrock listings, then merges cross-play duplicates by AnyServer ID.
Limitations
A discovery signal
Every input comes from AnyServer. ServersRank does not ping servers independently and cannot verify whether reported players are genuine. The score helps compare available listings; it is not a safety, fairness, or quality certification.
Historical charts appear only when a server detail response contains timestamped upstream observations.