Review scores are the most consumed and least informative part of games criticism, and they have effects well beyond helping people decide what to buy.
The aggregation problem
Aggregate sites convert varied scales into a common one and average them, sometimes weighted.
Which discards the reasoning entirely and treats a numeric score from one publication as commensurable with another that uses the number differently.
Publications using a ten-point scale where seven means good produce very different aggregate contributions from those where seven means mediocre.
Several publications have abandoned scores for exactly this reason, and some aggregators assign scores to unscored reviews based on their reading of the text.
The commercial consequences
Where this stops being an academic complaint.
Publisher bonuses contingent on aggregate scores have been documented, meaning developer compensation has depended on review outcomes.
Which creates obvious pressure and has been publicly criticised by developers.
Storefront visibility, marketing decisions and in some cases whether a sequel is commissioned have all been linked to aggregate performance.
Review copies and embargoes
The practical arrangements shaping what appears at launch.
Publishers provide advance copies under embargo, which allows reviews to publish at release.
Embargo terms sometimes restrict what may be discussed, which is a legitimate concern when it extends beyond spoilers.
Late or absent review copies are informative in themselves, and publications generally note it.
Games released without pre-release review access have, often enough to be a pattern, turned out to have significant problems.
The launch state problem
Reviews assess a game at a moment, and games change substantially afterwards.
Which means a review of a launch build may not describe the game a player buys two years later, and the score persists unchanged.
Multiplayer games in particular cannot be properly assessed before release, since server populations and matchmaking behaviour only exist at scale.
Some publications now publish reviews in progress or revisit titles, which is better practice and does not affect the aggregate.
User reviews
A different set of problems.
Review bombing — coordinated negative reviews for reasons unrelated to the game — has occurred repeatedly, generally in response to business decisions or unrelated controversy.
Platforms have introduced mechanisms to identify anomalous review activity and to display it separately, which is a partial response.
User reviews also skew toward the enthusiastic and the aggrieved, since neutral experiences generate less motivation to write.
What is actually useful
Reading a review from a critic whose previous judgements you have found reliable, which requires knowing who they are.
Reading for description rather than for verdict — what the game does, how it is structured, what it demands — which lets you judge fit.
And noting what a reviewer disliked, since your tolerance for it may differ entirely from theirs.
The refund route
Most major platforms offer refunds within defined time and playtime limits.
Which makes trying a game the most reliable evaluation method available, and it is underused relative to the amount of energy spent debating scores.
The limits are worth knowing before starting, since exceeding them removes the option.
Access journalism
The structural tension in games coverage.
Publications depend on publishers for review copies, interview access and event invitations.
Which creates pressure that most outlets manage through disclosure and editorial separation, and the incentive exists regardless.
Outlets funded by readers rather than by advertising have fewer of these dependencies, which is an argument for that model rather than a claim about any specific publication.
Sponsored content
Paid coverage exists and must be disclosed under advertising rules in most jurisdictions.
Disclosure practice on video platforms has been inconsistent enough that regulators have intervened in several markets.
Which means checking for disclosure is worth the two seconds it takes.
What a score cannot capture
Fit. A game can be excellent and entirely wrong for a particular player.
Length, pacing, difficulty and tone all determine whether someone will enjoy something, and none reduces to a number.
Which is why reviews describing the experience are more useful than reviews delivering a verdict, and why the number attracts attention that the text deserves.
Timing
Reviews published at midnight on launch were written under time pressure, sometimes without seeing the full game.
Which is disclosed by responsible outlets and is worth noticing, particularly for long games where a reviewer may not have finished.
Performance coverage
Technical analysis of how a game runs is a specialised area, separate from critical reviews.
Which matters because a game reviewed on high-end hardware may perform quite differently on what a reader owns, and reviews rarely specify the configuration used.
Outlets doing dedicated technical testing are the better source for that question.
Accessibility coverage
Dedicated accessibility reviews have emerged as a distinct category, assessing options for players with visual, auditory, motor and cognitive requirements.
Which is information conventional reviews rarely include and which determines whether a game is playable at all for a substantial audience.