Many outlets have replaced numerical scores with a recommendation. The simpler output hides the same problem the number had: several separate judgements are being merged into one.
A verdict combines quality and suitability
A reviewer is assessing both whether a game is well made and whether it will be enjoyed. These are different questions with different answers.
A precisely built game in a genre a reader dislikes is well made and unsuitable. A recommendation collapses that into a single word.
Readers who share the reviewer's tastes find the compression harmless. Readers who do not receive a verdict that answers a question they did not ask.
Price is silently part of the judgement
Whether something is worth buying depends on what it costs, and most verdicts are implicitly made at full price. A game marginal at launch is straightforward at half that.
Subscription availability changes it again, since the marginal cost of trying a game already included is close to nothing.
None of this appears in the verdict, which is why a recommendation ages badly. The judgement was made against a price that no longer applies.
The comparison set is invisible
Reviewers judge against what else exists, and their reference points come from having played a great deal within a genre. That context is not shared with the reader.
A strategy game that is unremarkable to someone who has played thirty of them can be excellent to someone playing their first. Both reactions are consistent.
This is the most common source of disagreement about reviews. The reviewer's baseline is higher, and the verdict is expressed relative to it.
Severity is flattened
A recommendation cannot express that a game is outstanding except for one flaw that will ruin it for a particular group. Motion sickness, difficulty floors and platform performance all work this way.
Reviewers handle this in the text, which is why the body of a review is more useful than its conclusion. The nuance exists but is not carried by the verdict.
Aggregators then discard the text entirely. What survives into the average is the part that lost the detail.
Why outlets keep using them anyway
A verdict makes a review scannable, and most readers do not finish long pieces. It also gives the outlet a consistent output across many writers.
The alternatives are worse in practice. Reviews with no summary judgement leave readers guessing, and long scales invite disputes about single points.
The compression is therefore deliberate rather than lazy. Reading the reasoning rather than the label is the only way to recover what was removed.