loader
Game Reviews

Six Numbers From the October 23 Directory Sync, and What Each One Actually Measures

Oct 22, 2026 Game Reviews
Six Numbers From the October 23 Directory Sync, and What Each One Actually Measures

A directory is only as useful as the numbers it can actually stand behind. This one syncs statistics from the platform on a schedule, and after the most recent pass 68 of the 93 tracked titles carry a complete three-statistic record — lifetime visits, concurrent players and favourites, all non-zero. The remaining 25 are listed but not yet synced, so they are excluded from every figure below rather than counted as zero. That exclusion is the first thing to understand about any total you see on this site.

What follows are six numbers drawn from those 68 games, chosen because each one is either an extreme or a total, and because each one is routinely misread.

1. 655,738,659,441 lifetime visits

That is the sum of every visit ever recorded across the 68 synced games. It is the largest number on the site and the least useful one, because it has no time dimension and no ceiling. It will be larger next month no matter what anyone does.

Its one real use is as a denominator. Knowing that brookhaven-rp accounts for 88,766,331,279 of it — 13.5% of the directory's entire lifetime traffic from a single title — tells you something about concentration that the raw figure cannot. One game out of 68 carries more than an eighth of all the footfall recorded here.

Cover art for the directory game with the highest concurrent players

2. 3,140,447 concurrent players

This is the number people assume is "how many people are playing Roblox", and it is nothing of the kind. It is how many people were playing these 68 specific games at the instant of the sync. It says nothing about the other 25 tracked titles, nothing about the platform as a whole, and nothing about any other moment in the day.

What it is good for is comparison between games inside the set, because every count was taken the same way at the same time. The spread is severe: steal-an-egg holds 776,312, which is 24.7% of the entire directory's live population on its own, and it is 191 times the count on jailbreak, the quietest game in the set at 4,062. A single distribution that wide is a warning: averages computed across this directory describe almost nothing real.

3. 786,518,738 favourites

Favourites are the most sentimental of the three headline statistics and the easiest to misinterpret, because they accumulate and never decay. A favourite added five years ago by someone who has not logged in since is still in this total.

The consequence shows up at the top. barrys-prison-run-obby leads with 105,251,144 favourites — 13.4% of everything in the directory — while sitting at 8,531 concurrent players. Compare rivals, second on favourites with 95,512,940, which has 88,973 people playing: more than ten times the live audience on a slightly smaller bookmark count. Two games, near-identical on one measure, an order of magnitude apart on the other.

4. 210,961,563 upvotes against 24,513,907 downvotes

Across the 68 synced games the directory holds 235,475,470 votes in total, of which 89.59% are positive. That aggregate is nearly meaningless on its own — players who dislike a game generally stop playing rather than vote — but the per-game spread around it is informative.

At the top, anime-dice scores 99.16% on 699,047 votes and speed-keyboard-escape scores 96.80% on 10,317,592 votes. At the bottom, adopt-me scores 27.27% — but on 11 votes in total, which is far too small a sample to conclude anything. Always check the vote count before believing the percentage.

5. A maximum visits-per-player ratio of 159597772

Divide lifetime visits by concurrent players and you get a rough measure of how much history sits behind each person currently playing. meepcity tops this at 159597772: 16,278,972,724 visits against 102 people. The second-highest, tower-of-hell, sits at 2716991.

A very high ratio is usually a signal about the data rather than the game. It means the lifetime counter and the concurrency counter are not describing the same population — which happens when a game is played in many short sessions, when much of its traffic is not session-based, or when its statistics are collected differently. Use it to spot games whose numbers need care, not to rank quality.

6. 379 tracked update entries across 68 games

The directory also keeps a changelog per game, and after this sync there are 379 entries in total, ranging from one game with 15 entries down to 15 games with a single entry. This is a measure of how much the site has managed to record, not of how much the games have actually changed — a game that ships constantly but publishes no readable log will still show one entry.

It is worth reading anyway, because the distribution is genuinely uneven and tells you where public changelogs exist at all. Games with a dozen or more recorded entries are the ones where you can find out what changed before you reinstall.

Why the totals move every time the sync runs

It is worth being explicit about the mechanism, because it explains a pattern that otherwise looks like instability. The directory does not poll continuously. It runs a scheduled pass that walks the tracked titles and refreshes their figures, and everything you read here comes from the most recent completed pass. Each pass can also change which games are eligible: a title whose statistics were missing last time may be synced this time, and it then enters every total and every ranking at once.

That is what happened recently. The synced set has grown repeatedly over the past fortnight, and each growth step moved 655,738,659,441-style totals and reshuffled rankings that had nothing to do with any game changing. When you see a headline figure move sharply, check whether the number of eligible games moved too before concluding that player behaviour did.

The practical upshot for reading this page: totals are comparable only between snapshots with the same eligibility, and per-game figures are always comparable within a single snapshot because they were collected under identical conditions. That asymmetry — trustworthy within a pass, fragile across passes — is the single most important thing to carry away.

What none of these numbers tell you

None of them tells you whether a game is good, appropriate, or worth your evening. They describe scale, timing and recorded sentiment, and the gaps between them are more informative than the values. The 25 unsynced titles are absent from every figure above, so any total you see here is a floor, not a complete count. And every concurrency figure is a single instant, which is why the same game can look busy on Monday and empty on Wednesday without anything having happened.

For the other views of the same data, the earlier directory-numbers piece works through the totals from a smaller synced set, and the visits-per-player explainer covers the derived ratio in more depth. The upvote and downvote breakdown takes the fourth number apart game by game. Roleplay is the largest genre in the set — start from the roleplay genre page — and the Brookhaven code page is the busy end of it.