The Steam Workshop's Set-and-Forget Majority (2026): 5.4 Million Mods Have Never Been Touched Since the Day They Were Published — and They Hold a Third of Every Subscription

Every mod on the Steam Workshop has two dates. One is when the creator published it. The other is when the item page was last changed — a title fixed, a bug squashed, a screenshot swapped, a compatibility patch pushed. On most of the platform, those two dates are the same day.

Of 6,026,179 active Steam Workshop mods carrying at least one subscription — the live, in-use catalogue across 392 tracked games — 5,446,346 of them (90.4%) have a modification timestamp identical to their publish date. Not "not updated in the last 90 days," which an earlier piece in this series measured at 93.4%: never touched at all, between the moment the creator hit publish and today.

That is the single largest cohort on the platform, and if you only read the update-vitality rankings it reads like a failure pile. It is not. Those 5.45 million mods hold 34.5% of every subscription on the Workshop, and they produced 467,966 of the 879,209 mods that have ever crossed 1,000 subscribers — 53.2%.

What the untouched majority actually looks like

The honest place to start is the one detail the word "dormant" hides: a mod with nothing to fix does not need fixing. A static map, a crosshair config, a model swap, a wallpaper — these are finished objects. The median gap between publish and last-touch, for the 1.53 million mods that were touched, is 19 days, with a lower quartile of 1.9 days and an upper quartile of 222. Most updates are not maintenance over years. They are the creator catching a typo in the first week.

Restrict to the properly-collected population — mods whose description field our crawler has actually visited, the discipline the tag cliff and blank-page problem both needed — and the same gap appears with the same magnitude: 33.47% against 17.42%. It is not a collection artefact.

The lever is not whether. It is when.

Split the 1.53 million touched mods by when the touch happened, and the picture stops being about "updating" at all.

A mod whose first post-publish edit lands more than a month after launch carries 3.5x the average subscriptions of one edited in its first month. Read carelessly, that says: wait a month, then patch, and your mod will do better.

It says nothing of the kind. This is a survival bias wearing a cause-and-effect costume. A mod that gets a compatibility patch three years after launch is, by construction, a mod that was still being played three years after launch. The early-touch cohort is dominated by the thousands of mods whose creator fixed a typo the day after uploading and then never opened the page again. The late-touch cohort contains almost every live project on the platform, because a live project is the only kind of mod that survives long enough to be touched late.

The only test that separates a lever from a composition artefact — the same test this series ran on the publishing-hour illusion, the cadence tax and the zero-favorite floor — is to compare the same creator against themselves, inside the same game.

Holding creator and game fixed

Group the catalogue by (creator, game) and keep only the pairs where the same creator published at least one mod they updated and at least one they never touched into the same ecosystem.

• 254,230 paired groups
• the updated side wins on reach-1,000 rate in 47,581, loses in 18,986, ties in the rest
• 71.5% of decided pairs go to the mod that was maintained
• aggregate reach: 28.98% updated against 17.69% never

So the effect is real. It is also 1.6x, not the 12x the pooled magnitude gap implies and not the 3.5x the timing table promises. Compare with the effects this series has already validated at scale: the first tag on a mod survives the paired test too, and so does the debut signal. What does not survive is the tidy version.

That 1.6x holds inside publish-year cohorts as well, which is where a survival bias would be most likely to leak back in:

The gap is stable and the level falls for everyone as the catalogue crowds — the same crowding pressure the attention-dilution piece documented. Roughly three in ten mods, in every year, get touched at all. That constant is itself the finding: the share of creators who ever return to their own upload form has not moved in a decade.

And at the game level the sign test is nearly unanimous. Of the 167 ecosystems with at least 500 mods on each side, the updated cohort wins in 165. The two that reverse are worth naming, because the series publishes its counter-examples:

Two anomalies in 167. Neither is large enough to move the aggregate, and both are small ecosystems where a single prolific uploader's habits can flip a rate.

Where the never-touched cohort actually lives

Here the story turns. If never-updated were a proxy for unsuccessful, the untouched cohort should occupy the bottom of every leaderboard. Instead it owns the top of them.

Wallpaper Engine alone has 105,027 never-touched mods past a thousand subscribers holding 1.25 billion subscriptions — more subscriptions than every touched mod in the entire Garry's Mod ecosystem combined, in a game where only 14.9% of mods have ever been edited. Trailmakers sits at 2.3%; Planet Zoo at 5.1%. These are engines for shipping finished objects, and finished objects do not want maintenance.

The individual entries tell the same story. The largest never-touched mod on the platform is "Four Seasons Warmup [Template]" — an Aimlabs crosshair/warmup config, published 25 January 2021, modified the same day, 6,941,757 subscriptions. The next five never-touched mods are also Aimlabs routines: Oregon (6,395,448), Bind (6,261,932), Ascent (6,131,459), Custom_rA100hz (5,858,483), R6 ClubHouse (5,490,492). Nine of the platform's ten biggest never-touched items are aim-trainer presets that were typed once and never revisited.

The freshest is the sharpest. MECCHA CHAMELEON's "Bikini Bottom" was published on 16 June 2026 and has 4,387,083 subscriptions — never updated, three months old, a hide-and-seek map skin in a game barely older than it is. It did not need cadence. It needed to be the right object in the right place at the right moment.

Across the whole catalogue, 3,526,791 never-touched mods are at least two years old, and 78 of them carry a million subscriptions or more. Not one has crossed ten million — but the ceiling is set by the games these objects live in, not by their immobility.

Two populations, one date field

What the data supports is not "update your mod." It is that the Workshop contains two kinds of object that happen to share a column.

The first is the finished artefact: a preset, a config, a model, a static map, a wallpaper. It is complete at publish. Its time_updated equals its time_created for the rest of its life, and that is the correct outcome. Every one of the platform's top ten never-touched mods is in this group.

The second is the living project: a framework, an overhaul, a game-mode with dependencies. It is never finished, and its audience is a function of whether it is still being worked on. Every one of the 165 games where the touched cohort wins is dominated by this kind.

Both are invisible in the same aggregate. Reading the 90.4% as an abandonment statistic mistakes a completed majority for a neglected one.

What this means if you publish mods

One. The decision that matters is made at publish time, not after. Three in ten mods are ever touched again, in every cohort year since 2015 — and the paired test says the touched side of that split enjoys a 1.6x reach advantage that survives every control. That is a real edge and a modest one. If your mod is a living project, that is the case for keeping it alive. If it is a finished object, the 1.6x does not apply to you and the "dormant" label is simply wrong.

Two. Do not read the 3.5x timing gap as advice to delay your first patch. The late-touch cohort is a list of mods that survived long enough to be patched late. Delaying a fix does not buy you that survival; shipping something worth surviving does.

Three. If you are picking an ecosystem to publish into, the untouched-mod leaderboard is also a map of where finished work wins. Wallpaper Engine, Trailmakers, House Flipper and Planet Zoo reward a complete object delivered once. Garry's Mod, Tabletop Simulator at 54.6% touched, and Cities: Skylines at 16.5% are ecosystems where a mod without follow-through competes against projects that have it.

Method: 7,429,521 active Steam Workshop items and the 6,026,179 of them carrying at least one subscription, across 392 tracked games, from ModSignal's live dataset. "Never touched" means time_updated is within one hour of time_created (or null), which for these items is publish day; 5,446,346 of the subscribed catalogue qualifies. "Touched" means a modification timestamp at least one hour after the publish timestamp. The paired test groups by (creator_steam_id, game_id) and keeps only groups with at least one item on each side; "wins" compares each side's share of items at or above 1,000 subscriptions, with ties excluded from the reported share. Publish-year and per-game cohorts use the same definition; the per-game sign test keeps the 167 ecosystems with at least 500 items on each side and mods whose description field has been collected. This piece measures all-time publish-day lock versus first-touch timing; the update-vitality and freshness-payoff curves (90-day and recency-weighted) are covered separately. All figures are a live snapshot current as of October 5, 2026.