How A Score Gets Made Here, And What Each Number Means

We are a small desk with a specific and narrow interest: whether an online game's servers are capable of describing the game its designers made. That is the whole remit. We do not score art, writing, balance or value, and a score here should never be read as a judgement of a game overall.

The four connection profiles

Every game is played from all four. A game that only works on the first is not a game that works.

The tether profile is synthetic and we say so. It exists because it separates games that degrade from games that collapse, and that distinction has turned out to predict long-term player sentiment better than any other measurement we take.

How we derive tick rate

From packet timing on our own connection, over a minimum of forty sessions per game, discarding the first and last thirty seconds of each. We never use a figure from a studio announcement, a patch note or a community wiki.

This matters more than it sounds. Eleven games in the archive have a measured rate different from their publicly stated one, in nine cases lower. Two games change rate under load without announcing it, of which Dock Nine is the extreme: fifteen hertz nominal, six measured during large engagements.

A stated tick rate is a marketing figure. A measured one is a fact about the game you are playing tonight.

What we count as a correction

A visible correction is any moment where the server overrules the client in a way a player can perceive: a position snap, a rewound death, a broken lock, a damage batch arriving late. We count them per match from recorded footage rather than from telemetry, because the question is what the player sees.

We record separately whether the game tells you a correction happened. This has become the single most predictive variable in our archive for whether players describe a game as cheating, and it is a small piece of interface work rather than an engineering problem.

Peek advantage

Measured in games where an event resolves instantly and sight lines exist. Two testers, one static and one approaching, on recorded footage, comparing the frame at which each becomes visible on the other's screen. Sixty repetitions per configuration, median reported.

Where a game has no instant resolution — trading, construction, most cooperative survival — the figure is meaningless and we leave the cell empty rather than filling it. Bazaar Run and Spore Depths both have blank peek columns for this reason.

How the score is composed

Thirty per cent measured tick rate against the requirements of the game's own shortest decision. Twenty-five per cent correction behaviour, counting both frequency and whether it is disclosed. Twenty-five per cent region coverage relative to where the audience lives. Twenty per cent what the game tells the player about their own connection.

A game with a modest tick rate can score well if its design does not need more, which is why Spore Depths sits at 8.2 on thirty hertz. A game with a good rate can score poorly if it is hosted in one place, which is why Frost Relay sits at 6.8.

Purchasing, revisions and errors

Every game is bought at retail with our own funds. We accept no review access, no early builds and no travel. We do not run affiliate links.

Scores are revised when a server change is detected, and the revision is stamped at the top of the review with the old figure beside the new one. Verge Squad carries such a revision, from 6.1 to 8.9, following a tick-rate change.

The short version

• Four connection profiles including one deliberately ruined.
• Six hosting regions, two of them chosen for being far from everything.
• Minimum three hundred logged matches before a score is published.
• Tick rates derived from packet timing, never from announcements.
• Corrections counted from footage, and disclosure recorded separately.
• Retail purchases, no review access, no affiliate arrangements.