loader
Game Reviews

Upvotes, Downvotes and What 60 Games Actually Get Rated

Oct 18, 2026 Game Reviews
Upvotes, Downvotes and What 60 Games Actually Get Rated

Almost every Roblox statistics page that mentions votes shows the raw counts, which is nearly useless — big games collect more of everything. The interesting number is the ratio, and this directory now has enough games with a complete record that the ratio finally says something. There are 60 titles carrying both vote figures and all three statistics after the 18 October 2026 sync, up from 41 a week and a half ago.

Anime Dice cover art, the highest upvote ratio in the directory

The aggregate picture

Across all 60: 199,413,564 upvotes against 23,025,603 downvotes, which is 89.6% positive. Every single game in the set has more upvotes than downvotes. That headline is misleading in an instructive way, and understanding why is most of the value in this article.

Vote buttons on Roblox sit next to a game someone has already decided to open. Reaching the point of casting a vote requires installing and playing, and people who dislike a game usually leave instead of registering an opinion. The denominator of this ratio is therefore self-selected towards people who stayed long enough to care, and any interpretation that treats 89.6% as " players approve this much" is misreading what was measured. It is better understood as: among people who played long enough to vote at all, roughly nine in ten were positive.

The most positively rated games

#GameUpvotesDownvotesPositive
1anime-dice666,2475,33499.2%
2slayers-2942,69022,73197.6%
3speed-keyboard-escape9,972,414328,01696.8%
4loot-to-forge162,9465,61896.7%
5ride-a-pet450,20717,05796.3%
6build-an-ant-empire54,6352,13896.2%
7bee-swarm-simulator3,039,336139,95495.6%
8build-the-pyramid123,0156,78594.8%

The top of this list is populated almost entirely by small, focused games. Anime Dice carries 666,247 upvotes against 5,334 downvotes — 99.2% positive — on 45,379 concurrent players and 108,306,477 lifetime visits. Loot to Forge shows 162,946 against 5,618. Neither is a top-twenty title by population, and that is the pattern rather than a coincidence.

Speed Keyboard Escape is the exception worth studying, because it is not small: 78,188 concurrent players, 6,493,249,598 lifetime visits and 7,106,485 favourites, with 9,972,414 upvotes to 328,016 downvotes. It is one of only two titles the obby genre page files, and it manages a top-three positive ratio at genuine scale — 96.8%. A large audience that agrees with itself is unusual and worth noting on its own terms.

Where the middle of the distribution sits

Between the extremes, the overwhelming majority of these 60 games land in a narrow band. The total vote pool is 222,439,167 votes across the set, and once you exclude the top and bottom handful almost every remaining title sits somewhere between roughly 85% and 95% positive. That clustering is itself the finding: with a self-selecting voter population, a vote ratio carries very little signal, and the differences between two games five places apart in the middle of this table are not meaningful differences at all.

It also means the ordering within the middle is unstable. Adding nineteen games to the sample this month moved the aggregate positive rate and reshuffled the middle places without changing anything about the games themselves — a reminder that this is a measurement of who voted, not a measurement of what was voted on.

The least positively rated

GameUpvotesDownvotesPositive
adopt-me3827.3%
barrys-prison-run-obby406,678264,01460.6%
tower-of-hell4,375,9941,568,45473.6%
fling-things-and-people955,349233,26980.4%
ski-alpha16,0183,87780.5%
bedwars-rerelease2,484,171534,68082.3%

Two different things produce a low ratio, and the table mixes them. Adopt Me's figure comes from a sample of 3 upvotes and 8 downvotes — eleven votes in total, which is not a measurement of anything, attached to a game whose visit and concurrency figures are internally inconsistent anyway. We flag it every time we publish this list so nobody quotes it as evidence.

The others are real. Barry's Prison Run Obby has the most favourites in this entire directory and one of the lowest positive ratios, at 60.6% from 406,678 upvotes against 264,014 downvotes. Tower of Hell carries 4,375,994 upvotes to 1,568,454 downvotes. Both are old, both are heavily favourited, and both carry more disapproval than the directory average — which is consistent with the idea that games people arrive at with high expectations and long memories collect more argument about them than games nobody has feelings about either way.

What this does and does not predict

We wrote about the same ratio when the sample was 41 games, and the conclusion has not changed with nineteen more titles added: vote ratio does not predict popularity, and popularity does not predict vote ratio. The games with the most favourites and the games with the best ratios are almost disjoint sets.

It follows from this that a low ratio is not a warning sign for choosing what to play. It describes how a population voted, not the quality of an evening you might have. The genuinely actionable reading is narrower than people hope: a small game with a very high ratio and a few hundred players online is a reasonable bet if you want something uncontroversial and you have no prior expectation, and nothing else on this axis should influence a choice.

There is also a structural limit worth stating. This directory holds no timestamp on votes, so we cannot tell whether a ratio accumulated over eight years or eight days. A title that changed design mid-life carries both eras in one number, inseparably. The same limitation applies to every accumulation metric here, including favourites — discussed further in the favourites versus live players comparison — and it is why no ranking on this site is presented as a quality judgement.

Where the gaps are

60 of the 93 games in this directory carry any vote record at all. The remaining 33 have titles and slugs and nothing else, so they contribute nothing to this analysis and no claim is made about them. Among the 60 that do, the record is complete, but "complete" describes three statistics and a vote pair — not genres, not release dates, not update cadence, none of which this directory's game endpoint returns.

For a concrete sense of what a full per-game record does contain, Speed Keyboard Escape's entry at the Speed Keyboard Escape codes page shows everything held for it in one place, and how dead code lists happen explains why the code section on those pages sometimes reads empty for games that are otherwise well documented.

What to take away

89.6% positive across 60 games is a fact about who votes, not about how good anything is. The per-game spread is real but small, it barely overlaps with success, and the only honest use for it is narrow: uncontroversial small games cluster at the top, and games with long histories and strong expectations cluster at the bottom. Anything stronger than that is reading taste into arithmetic.