Home › Video Game Preservation Statistics

Video Game Preservation Statistics 2026: The 13% Left, in 120+ Sourced Data Points

On Monday 27 March 2023 Nintendo switched off the 3DS and Wii U eShops, and 121 Game Boy family games stopped being sold in any form. Only 110 were left. Nintendo was mid-way through an active programme of reissuing that exact library — which is the finding: the best case never clears the average. The most commercially active ecosystem the researchers could find, the PlayStation 2, reaches 12.00% availability against the abandoned Commodore 64’s 4.50% — more than double, a real and statistically solid gain — and still lands no better than the 13.27% average for every US game released before 2010, a comparison the study itself calls roughly statistically even. Caring about your back catalogue doubles your odds and still leaves about seven-eighths of the library unbuyable, because a game is not an artefact on a shelf. It is a bundle of separately licensed works on separate expiry clocks, bolted to hardware nobody manufactures. When the Video Game History Foundation and the Software Preservation Network ranked platform ecosystems by how much of each library you can still buy, the striking thing was not the gap between best and worst. It was that maximum effort buys you average.

Which console libraries are best and worst preserved?

The PlayStation 2 ranks best at 12.00% of its US library still in release, the Commodore 64 worst at 4.50% — and the whole point of the ranking is that 7.5 percentage points is all that separates an actively re-commercialised library from an abandoned one. The Video Game History Foundation, working with the Software Preservation Network and the University of Washington Information School, checked availability by hand for four separate samples between 31 January and 15 April 2023, then sorted the platforms by how much commercial attention each one gets.

All four figures are from the same study, use the same definition of “in release”, and cover US releases only. Platform samples are limited to titles published during each platform’s commercially active lifespan, which is why the PS2 population runs to 2013 while the benchmark population stops at 2009 — the two are not the same universe, and the benchmark is listed as a reference line rather than ranked against the platforms. The Game Boy row is a complete census of all 1,873 known US Game Boy family titles (509 Game Boy, 442 Game Boy Color, 922 Game Boy Advance), not a sample, which is why the study reports it at ±0.0%. Other confidence intervals are 95%. Source: VGHF and SPN, Survey of the Video Game Reissue Market in the United States, Phil Salvador, July 2023 (tables 1–2).

The Game Boy result is the one that should worry anyone who assumes an attentive rightsholder solves this. Nintendo has an active, ongoing programme of reissuing Game Boy software. Its library still lands at 5.87%, which sits inside the Commodore 64 sample’s margin of error — the study says so plainly. A platform holder trying and a platform holder that no longer exists produce statistically indistinguishable outcomes.

One caveat belongs here rather than in the small print, because it is the strongest argument against everything above, and the study makes it against itself. In a footnote, the authors observe that the near-parity between the PS2 and the all-games rate implies there are ecosystems with higher levels of commercial activity that this study did not identify. In other words, the researchers do not claim the PS2 is the ceiling — only that it is the best case they measured. Read the ranking as three data points and a benchmark, not as a proof that no rightsholder anywhere does better. The claim that survives is narrower and still uncomfortable: across an abandoned platform, a neglected one, an actively re-commercialised one and a 1,500-game cross-section of everything, the answer never leaves the 4.5% to 13.3% band.

There is a second way to attack the ranking, and it is worth running because it is the one correction that would most flatter the industry. The report discloses, in its own section on service degradation, that a login fault on the PlayStation 3 and Vita stores made some titles unreachable that were technically still for sale — and that the researchers counted those as unavailable. They quantify it: 4 of the 400 PlayStation 2 entries (1.0%) and 12 of the 1,500 general-sample entries (0.8%). Hand those back and the PS2 rate rises from 12.00% to 13.00%, while the all-games rate rises from 13.27% to 14.07%. The gap barely moves — 1.27 points becomes 1.07 — and the conclusion survives its own most generous stress-test intact (16Best analysis of the study’s service-degradation counts; 52 of 400 and 211 of 1,500 respectively).

Which era of games is hardest to buy today?

Everything before 1985, by a distance: the study puts the pre-1985 reissue rate at 2.59% against 16.49% for 1985–2009, and only one of the 132 sampled games released before 1980 has ever been brought back. Ranked by release period, the curve rises sharply after the 1983–85 North American crash, peaks in the early 1990s, and then wobbles inside a five-point band — it never once clears 20%.

Ranked by availability, not chronology — the ordering is the finding. The three earliest buckets are merged here because they rest on 1, 2 and 17 sampled games respectively and are too thin to rank separately; all three returned zero. Percentages and counts alike are the study’s own published figures, and the percentages are shares of each period’s sampled games, not of the full population of that period. What is 16Best analysis is the total row and the check behind it: the published rows sum to 199 games in release out of 1,500 sampled, which is 13.27% — the headline figure, rebuilt from the period table and landing on it exactly. Published stat tables often fail that test. This one passes. One small note for anyone recomputing: the report states the pre-1985 average as 2.59% and the 1985–2009 average as 16.49%, while the same table rebuilt from counts gives 2.56% and 16.54%. The difference is a rounding artefact and does not move any conclusion. Source: VGHF and SPN, table 3, July 2023.

How many individual games does 87% actually mean?

Roughly 23,961 titles, against about 3,666 you can still buy. The study’s own population — every video game released in the United States before 2010, counting each game once regardless of how many platforms it appeared on — is 27,627 titles, exported from the MobyGames database in May 2022 and current as of 28 June 2022. Apply the measured 86.73% and the abstraction becomes a number you can hold.

16Best calculation. The 27,627 population figure and the 13.27% ±2.5% availability rate are both the study’s. This multiplies a sample-derived rate by a population count, so treat every figure as an order of magnitude with a stated band, not a census. Rows are rounded to whole titles, and each pair reconciles exactly to 27,627. It is deliberately built on the unique-title basis: the study notes that counting every platform-specific release separately raises the population to 46,251, but says that basis does not match how it defined release status, so we do not use it. Two different scopes, two very different answers — which is exactly why published “games lost” totals vary so widely.

What happens to a game library when a storefront closes?

It can lose more than half its commercial availability overnight, which is exactly what happened to the Game Boy when Nintendo closed the 3DS and Wii U eShops on 27 March 2023. This is the single cleanest natural experiment in the entire preservation literature, because the VGHF was mid-study when it happened and had a complete census of all 1,873 US Game Boy family titles rather than a sample.

The 121 and 110 figures are the study’s; the 231 and 12.33% rows are 16Best additions of the two, and they assume nothing else in the library changed status inside the same window. The study is explicit that the 121 titles are “no longer available in any form”, so they are disjoint from the 110 that remain — which is what makes the addition safe. The report gives the underlying arithmetic too: 155 Game Boy family titles were on the 3DS and Wii U Virtual Console storefronts in North America, and 121 of those existed nowhere else, leaving 34 that were also sold elsewhere and therefore survive inside the 110. The 121 breaks down as 47 Game Boy, 25 Game Boy Color and 49 Game Boy Advance titles, and the 110 as 29, 12 and 69; both sets sum correctly against the 1,873 census (509 Game Boy, 442 Game Boy Color, 922 Game Boy Advance), a useful internal check on the data. One caution on the source: the report gives this loss as 6.46% in its results section, 6.56% in its Summary of Findings and “6.5 percent” in its executive summary. 121 divided by 1,873 is 6.46%, so we use that.

The pattern is not confined to Nintendo, and the dates matter more than the anecdotes.

These counts come from different analysts using different bases and must not be added together. The 2021 PlayStation figure was an estimate of exposure before Sony reversed the PS3 and Vita closures; those two stores are now scheduled to close again on the timetable in the last row. Previously purchased content remains downloadable in each case — these are commerce closures, not deletions from customers’ accounts, which is precisely the distinction that makes them a preservation problem rather than a consumer one. The 674 figure is the study’s own arithmetic on fan-compiled platform lists current to April 2023, not an audited catalogue — it states “as many as 674 PlayStation games” in the body and rounds it to “over 600” in its summary, and it is a count of titles at risk from a service fault rather than a delisting. Sources: Nintendo Support; PlayStation.Blog, 1 July 2026; VGC analyses of the 2021 PlayStation and 2024 Xbox 360 closures; TechRadar on the Xbox 360 closure; VGHF, 2023.

Why do games go out of print when books do not?

Because a game is not one work — it is a stack of separately licensed works that expire on separate clocks, and when any one of them lapses the whole product has to come off sale. A novel is one copyright held by one party. A racing game is engine code, a soundtrack licensed per-territory for a fixed term, car models licensed per-manufacturer, brand marks, sometimes athlete or actor likenesses, and middleware, all bound to hardware that stopped being manufactured.

Three documented cases show the mechanism at three different points:

• The whole game is withdrawn. Remedy’s Alan Wake was pulled from digital stores on 15 May 2017 when its seven-year music licences ran out. Microsoft later renegotiated and the game returned — which is the exception that proves the rule, since it required a platform holder with lawyers to want it back.
• The game stays, the artefact changes. On 26 April 2018, ten years after release, Rockstar removed a batch of licensed tracks from Grand Theft Auto IV as ten-year agreements lapsed; the Russian-language station Vladivostok FM was hit hardest, losing about 15 tracks across the base game and expansions. Community trackers put the overall total at 53 to 54 songs across GTA IV and its two expansions — a fan tally built against Rockstar’s own support notice, not a publisher-issued figure, which is why the count varies by one. The product on sale in 2026 is not the product shipped in 2008.
• Delisting is scheduled in advance. Playground Games removed Forza Horizon 4 from all digital stores on 15 December 2024, citing licensing and agreements with third-party partners; Forza Horizon 3 went the same way in 2020, four years after release. The game remains playable for owners. It is simply no longer for sale.

Is it legal for a library to preserve a video game?

Yes to preserve it, no to lend it remotely — and the US Copyright Office reaffirmed that split on 25 October 2024 when it declined to grant an exemption allowing libraries and archives to offer off-site access to preserved out-of-print games. This is the load-bearing legal fact of the whole subject, so the specifics matter.

Sources: US Copyright Office, Ninth Triennial Section 1201 Proceeding (2024 cycle) and Tenth Triennial Section 1201 Proceeding (2027 cycle); VGHF statement on the 2024 ruling. The VGHF has said the industry trade body would not support remote research access under any conditions. Nothing in this section is legal advice.

Europe tested the other route and got a similar answer. The Stop Destroying Videogames European Citizens’ Initiative collected 1,294,188 verified statements of support — 29% above the one-million threshold — and was submitted on 26 January 2026. Parliament held a public hearing on 16 April 2026 and debated it in plenary on 21 May. On 16 June 2026 the Commission declined to propose a legal obligation to keep games playable after commercial end-of-life, judging such a requirement disproportionate and citing intellectual-property rights, publisher costs and cybersecurity risk. It committed instead to three things by the end of 2026: a voluntary industry code of conduct on end-of-life management, consumer-awareness work with consumer organisations, and a report on how the Digital Content Directive applies to discontinued digital content. The largest organised consumer push in the medium’s history produced a process, not a rule.

Are games worse preserved than films and recorded music?

Games are roughly level with recorded music on a like-for-like measure, and the famous comparison with silent film is not like-for-like at all — it compares a commercial-availability rate against a physical-survival rate. This is the most-repeated line in preservation coverage and the one most often stated wrongly, so here is the ranking with the scope of each figure attached.

The last row measures something different from the four above it: whether a physical print still exists anywhere, not whether the work is for sale. Pierce counted 1,575 of 10,919 American silent features surviving as complete domestic-release copies in their original format — and a further batch surviving as foreign versions or on lesser gauges such as 16mm, which takes survival in some form to 25% (2,749 of 10,919). So “only 14% of silent films survive” is itself a misquotation of the film study, before anyone compares it to games. Two scope notes on the recordings rows: Brooks measured reissue by the rights owners who control the recordings, and he split his eras at the war rather than at a stated calendar year, so we do not put a boundary year on them. The VGHF cites both comparisons accurately in its own report — against the pre-WWII recordings figure and the silent-film survival figure, each correctly labelled; a great deal of the coverage built on it does not. Sources: Tim Brooks, Survey of Reissues of U.S. Recordings, CLIR pub133, August 2005; David Pierce, The Survival of American Silent Feature Films 1912–1929, CLIR pub158 / Library of Congress, 2013 (pp. 21, viii); VGHF and SPN, 2023.

Who is actually preserving games, and at what rate?

Commercial re-release programmes, subscription catalogues and non-profit archives all operate at a scale two to three orders of magnitude below the problem. Rank the visible efforts by catalogue size and the gap is immediate.

These are not comparable catalogues and are not additive: Antstream streams licensed titles across many platforms, GOG’s programme is a maintenance commitment on PC games it already sells, Nintendo’s figure is its own advertised floor for a curated subscription shelf and community trackers count more, and the VGHF library holds documentation and press material rather than playable games. They are ranked here only to show order of magnitude against a population of roughly 23,961 unavailable titles. Sources: Digital Preservation Coalition 2026 Awards (GOG); Nintendo official Classic Games page; VGHF library launch and 2025 impact report; Antstream.

Which leaves libraries, universities and non-profits carrying the archival load under the legal constraint described above — they may preserve, but not lend at distance. The Council on Library and Information Resources, the same body that published both the recordings and silent-film surveys, published an overview of emulation as a preservation method in July 2025, and it cites the 87% finding directly. Emulation is the standard technical method by which memory institutions keep obsolete software runnable; it is also the reason the §1201 question is unavoidable, because running a protected old game generally means circumventing something. In July 2026, after Sony confirmed it would end PS5 disc production in 2028, VGHF director Frank Cifaldi said of the industry’s trade body: they refuse to offer a meaningful alternative. That is the deadlock in one line: the legal route is closed, the commercial route covers 13%, and the archival route is not allowed to reach anyone remotely.

Is the modern catalogue being archived any better?

No — the modern catalogue is being created far faster than any archive is growing, and it is being created almost entirely on storefronts, which is the failure mode this page has just documented. Set the two populations side by side, with their scopes stated, and the arithmetic is unkind.

These populations are not comparable like for like and are not meant to be: 27,627 counts US releases across every platform of the pre-2010 era, while the Steam rows count one global PC storefront. The comparison is a rate comparison, not a share calculation, and every scope is stated so it cannot be misread as one.

Subscription catalogues are often offered as the answer, and they are genuinely the delivery mechanism for a large share of what remains available — the VGHF found PlayStation Plus Premium among the sources keeping PS2 titles in release. But a subscription is a curated shelf with a rotation policy, not an archive: our game subscription services statistics show these products are built for retention, which means the catalogue is a marketing budget and shrinks when the budget does. Nintendo advertises “over 150 classic games” across eight systems for a hardware family that has sold 175.78 million units per our console market share analysis. Against about $205 billion of global consumer spending on games in 2026 and a 3.6 billion total-reach player count in our video game industry statistics, and the 20,017-release-a-year supply pressure in our indie game statistics, the preserved shelf is a rounding error by design.

Why do video game preservation statistics disagree?

Because “commercially available” is a definition with at least six switches in it, and flipping any one of them moves the headline by several percentage points. This section is where the argument lives, not the housekeeping. The 87% is stable; the numbers built on top of it are not, and knowing why is what separates a usable figure from a repeated one.

• What counts as “available”? The VGHF used a deliberately broad test: a game is in release if it, or a version derived from it, is reasonably, readily and legally available from the rightsholder for a currently supported platform, in any format. That means remasters, emulated re-releases, compilation sets and subscription catalogues all count as availability. A narrower definition — original version only, purchase only, no subscriptions — would push the 13.27% down, not up. The famous number is the generous reading.
• Does a remake count? Not in this study. A rebuilt-from-scratch remake is treated as a separate work, while a faithful reconstruction using the original assets counts as a reissue of the original. The study explicitly declines to define “substantially different” and left borderline calls to researchers, instructing them to count versions as the same release when unsure — again biasing towards higher availability.
• Which universe of games? This is the biggest source of divergent “games lost” totals. The same dataset yields 27,627 pre-2010 US titles counting each game once, or 46,251 counting every platform-specific release separately. A 70% swing in the denominator, before anyone measures anything. Regional releases are also counted differently across databases, and games with an inherently international release, such as browser games, sit outside the US-release framing entirely.
• Availability rate versus survival rate. The comparison to silent film measures whether a complete original-format print physically exists — and even inside that one study the number moves, from 14% for complete domestic-release copies to 25% for survival in any form. The games and recordings figures measure something else entirely: whether you can buy the thing. A game can be perfectly preserved in an archive and still count as unavailable, and most of the 87% are in exactly that position. Any sentence claiming games are “more lost than silent films” is comparing two different measurements.
• Does unauthorised availability count? No. Commercial availability means available from the rightsholder. Emulation communities and unofficial archives change what is technically playable; they do not change the commercial-availability number by a single title, and the study explicitly excludes illegally reproduced versions. This is also why the 87% has not fallen despite decades of enthusiast archiving.
• Sampling versus census. Three of the four ecosystem figures are samples with margins of error of 2.5–4.3 points; only the Game Boy figure is a complete census. That is why a 1.37-point gap between the Game Boy and Commodore 64 rates is not a real difference, while the 7.5-point PS2-to-C64 gap is.
• Everything has a timestamp. The study measured availability between 31 January and 15 April 2023 and nothing after. Since then the Xbox 360 store has closed and the PS3 and Vita stores have been scheduled to close. Every number on this page describes a market that only moves in one direction between measurements, which means 13.27% should be read as a ceiling for 2026, not a current reading.

Our house basis for this page: the VGHF and SPN survey as the single source for availability rates, its own populations as denominators, ranges shown wherever a margin of error exists, and every cross-media comparison labelled with the measurement it actually uses. On that basis the ranking is remarkably robust — and the robustness is the story. Change the platform, change the era, change the rightsholder’s level of interest, and the answer stays between 4.5% and 19.33%.

Who paid for these numbers, and which way does that bend them?

One side of this argument has published a dataset; the other has published none. That asymmetry deserves stating out loud before anyone cites the 87%.

The study is not neutral research. It was produced by the Video Game History Foundation and the Software Preservation Network — the same organisations petitioning the US Copyright Office for a preservation exemption — and its own executive summary says why it exists: industry lobbyists opposed new exemptions “on the grounds that there is already a thriving reissue market,” and the researchers set out to test that claim empirically. An advocacy group measuring the thing it is advocating about is a conflict worth knowing.

Read the design against that motive and the bias runs the wrong way for a critic. Every discretionary call in the methodology inflates availability, not deflates it: emulated re-releases count, compilations count, subscription catalogues count, faithful reconstructions count as the original, and volunteers were told to resolve borderline cases in favour of “still available.” A group trying to make the market look bad had the levers to reach a lower number and did not pull them. The 13.27% is the friendly reading.

The instructive part is what the other side publishes: nothing comparable. The Entertainment Software Association’s position in the 2024 rulemaking rested on the assertion that a healthy reissue market exists, and no industry-funded census of that market has been released to check it against. So the honest description of this vertical’s evidence base is not “two studies disagree.” It is one measurement, from an interested party, using assumptions that flatter its opponent, against an opposing claim with no measurement attached at all. Cite the 87% with that on the record — and note that at three years old, it is also the newest figure anyone has.

Key takeaways

• The ranking barely ranks, and that is the finding. PS2 at 12.00%, Game Boy at 5.87%, Commodore 64 at 4.50% — a 2.7× spread and only 7.5 percentage points between an actively re-commercialised library and an abandoned one. Maximum commercial effort buys a real, statistically solid gain over neglect, and still does not beat the 13.27% all-games average; it matches it (16Best analysis).
• Digital distribution did not help — and that is the claim, not that it hurt. Games from 2005–2009 sit at 17.89% against the 1990–1994 cartridge era’s 19.33%, a difference far too small for the sample to resolve. No improvement is the finding. The sharper evidence is a single day: 27 March 2023 removed 121 Game Boy titles against the 110 that survived, halving that library’s availability, because those reissues existed in exactly one place.
• About 23,961 titles are the scale of it (range 23,270–24,652), out of the 27,627 US titles released before 2010. For every one you can still buy, roughly 6.5 you cannot (16Best analysis of VGHF data).
• The law is the bottleneck, and it moves on a three-year clock. The Copyright Office declined the remote-access exemption on 25 October 2024; the next possible effective date is October 2027. In Europe, 1,294,188 verified signatures — 29% above the threshold — produced a voluntary code-of-conduct process on 16 June 2026, not an obligation.
• Games and recorded music are level; the film comparison is not like-for-like. 13.27% against 14%, both measured on 1,500-work samples, both measuring commercial availability. The silent-film 14% counts complete original-format survivals among 10,919 films — a different question entirely, and one where survival in some form is 25%.
• Rescue capacity is roughly 1% of the gap. The most systematic commercial programme, GOG’s, went from 100 to 267 titles between November 2024 and February 2026 — about 1.1% of the unavailable population, at a pace that implies roughly 179 years to close it (16Best analysis).
• The problem compounds faster than the solution. Steam published 20,017 titles in 2025, about 36× the pre-2010 catalogue’s average annual rate and 72.5% of that entire 50-year catalogue in one year, growing 15.7% annually. No archive grows at 15.7%.
• So the spine, landed: commercial interest does move availability — it more than doubles it — but it only moves it around inside a ceiling nobody in the study got past. An industry taking roughly $205 billion a year in consumer spending cannot sell you about seven-eighths of the American catalogue it built before 2010, and the ranking shows why. Effort sets your floor. Live licences and a working storefront set your ceiling. Records ran into a version of the same wall: Brooks found 54% of historical blues legally on sale abroad against 10% in the US, which is a rights-geography problem, not a taste problem. Games bolted two more expiry clocks on top of it — hardware that stops being manufactured, and a store with an off switch.

Frequently asked questions

What percentage of video games are no longer available to buy?

About 87%. The Video Game History Foundation and the Software Preservation Network found that 13.27% of video games released in the United States before 2010 are still in release, meaning 86.73% are not, based on a random sample of 1,500 titles with a 2.5 percentage point margin of error at 95% confidence. Availability was checked by hand between 31 January and 15 April 2023 and published on 10 July 2023. The definition is deliberately broad: remasters, emulated re-releases, compilation sets and subscription catalogues all count as a game being available, so 13.27% is the generous reading rather than the strict one.

How many old video games are out of print?

Roughly 23,961 titles, on the study's own population of 27,627 US games released before 2010, with a range of about 23,270 to 24,652 at the sample's margin of error. That leaves around 3,666 titles still commercially available. This is an extrapolation from a sample rather than a title-by-title count. The figure changes substantially with the denominator: the same dataset yields 46,251 titles if every platform-specific release is counted separately, which is why published totals for how many games are lost vary so widely.

Which console library is best preserved?

The PlayStation 2, at 12.00% of its US library still in release, ahead of the Game Boy family at 5.87% and the Commodore 64 at 4.50%. The more useful reading is how small that range is. Only 7.5 percentage points separate an actively re-commercialised library from one whose manufacturer dissolved in 1994, and the PS2 does not beat the 13.27% average for all pre-2010 US games; the 1.27-point gap sits inside both samples' margins of error, and the study calls the two roughly statistically even. Note the scopes differ slightly: the PS2 sample covers 2000 to 2013 releases, the benchmark covers all US releases before 2010. The Game Boy figure is a complete census of 1,873 titles rather than a sample, and it lands inside the Commodore 64 sample's margin of error, so an attentive rightsholder and an absent one produced statistically indistinguishable outcomes.

Is video game preservation legal in the United States?

Libraries and archives may preserve games, but they may not offer remote access to them. On 25 October 2024 the US Copyright Office declined to grant a DMCA Section 1201 exemption sought by the Software Preservation Network, supported by the Video Game History Foundation, that would have let institutions share preserved out-of-print games with off-site researchers. The Final Rule was published on 28 October 2024 and left the preservation exemption on-premises only, with one narrow clarification allowing simultaneous access by as many users as an institution owns lawfully circumvented copies. The tenth triennial proceeding opened on 9 June 2026 with petitions due 24 August 2026, and any resulting exemption would run from October 2027 to October 2030.

Are video games worse preserved than films and recorded music?

They are level with recorded music, and the film comparison measures a different thing. Games sit at 13.27% commercially available for pre-2010 US releases; Tim Brooks found 14% of US recordings from 1890 to 1964 had been reissued on CD by their rights holders, using a sample of exactly the same size, 1,500 works. Those two are directly comparable and land 0.73 percentage points apart, despite the recordings being on average decades older. The often-quoted silent film figure, 1,575 of 10,919 American silent features from 1912 to 1929, measures survival as a complete domestic-release 35mm copy rather than commercial availability, so it answers a different question; counting foreign versions and smaller gauges, 25% of those films survive in some form.

Do digital storefronts make game preservation better or worse?

Not better, and there is no evidence they are worse. Games released from 2005 to 2009, the first digital-distribution era, are 17.89% available against 19.33% for 1990 to 1994 cartridge-era releases, but on samples of 246 and 269 games that 1.44-point difference is far too small to be meaningful, and the study's own conclusion is that the digital era showed no major increase over earlier periods. The real risk digital adds is concentration, not decline. When Nintendo closed the 3DS and Wii U eShops on 27 March 2023, 121 of 1,873 Game Boy family titles left sale against 110 that remained, cutting the library's availability from about 12.3% to 5.87% in one day. A physical release puts thousands of copies into the world. A digital-only release exists in one place, and a store closure removes it from the only place it existed.

Why do video game preservation statistics disagree?

Six reasons. The definition of available differs, and the widely quoted study uses a broad one that counts remasters, compilations and subscription catalogues. Remakes are treated as separate works while faithful reconstructions count as reissues, and the boundary is judgment-based. The population differs: 27,627 unique US pre-2010 titles or 46,251 platform-specific releases from the same dataset. Availability rates and survival rates get compared as though they measure the same thing, which they do not. Unauthorised availability never counts, since commercial availability means available from the rightsholder. And every figure is timestamped: the survey measured availability up to 15 April 2023, and the Xbox 360 store has closed since, so 13.27% should be read as a ceiling for 2026 rather than a current reading.

Sources

Not all of these carry the same weight, so here is the tiering. Primary research: the VGHF and SPN survey, Brooks 2005 and Pierce 2013 — hand-checked samples or censuses with published methodologies and stated margins of error. Primary regulatory and official records: the US Copyright Office proceedings, the European Citizens’ Initiative register, the Commission press release, Nintendo Support and PlayStation.Blog notices. Third-party analyst counts: the two Video Games Chronicle store-closure analyses and SteamDB — credible, widely corroborated, not peer-reviewed or audited. Self-reported vendor figures: Antstream’s catalogue claim, Nintendo’s “over 150” advertised floor and GOG’s numbers as submitted to an awards jury — none independently audited, and each is flagged where it appears. Community tallies: the GTA IV track count and the PS3/Vita reissue lists the VGHF itself relied on — useful, unaudited, and labelled as such in the tables.

• Video Game History Foundation & Software Preservation Network — Survey of the Video Game Reissue Market in the United States (Phil Salvador, version 1.1, 10 July 2023; 13.27% overall, ecosystem and period tables, 27,627 and 46,251 populations, Game Boy census, definitions appendix)
• Video Game History Foundation — 87% Missing: the Disappearance of Classic Video Games (10 July 2023; headline findings, sample sizes, comparison framing)
• Video Game History Foundation — The Game Availability Study, Explained (platform rates, era summary, definition of in print)
• Video Game History Foundation — Statement on the DMCA 2024 triennial review ruling (25 October 2024; exemption declined, petition scope, trade-body position)
• US Copyright Office — Ninth Triennial Section 1201 Proceeding, 2024 Cycle (final rule issued by the Librarian of Congress 25 October 2024, effective 28 October 2024, exemptions running to 28 October 2027; the Register’s Recommendation is the primary document for the denial and its reasoning)
• US Copyright Office — Tenth Triennial Section 1201 Proceeding, 2027 Cycle (notice published 9 June 2026; petitions due 24 August 2026; exemptions would run October 2027 to October 2030)
• European Citizens’ Initiative register — ‘Stop Destroying Videogames’ initiative record, ECI(2024)000007 (1,294,188 verified statements of support, thresholds met in 24 member states, submitted to the Commission 26 January 2026)
• European Commission — Commission will engage with industry following the European Citizens’ Initiative on the disabling of videogames by publishers (press release IP/26/1369, 16 June 2026; no legal obligation proposed — a voluntary industry code of conduct, consumer-awareness work with consumer organisations, and a report on the Digital Content Directive, all by end-2026)
• European Commission — Hearing in the European Parliament on the ‘Stop Destroying Videogames’ European citizens’ initiative (IMCO, JURI and PETI joint public hearing, 16 April 2026; plenary debate 21 May 2026; Commission answer 16 June 2026)
• Tim Brooks, CLIR / Library of Congress — Survey of Reissues of U.S. Recordings (August 2005; 14% of pre-1965 recordings reissued, 10% or less pre-WWII, 25%+ post-WWII, genre and territory splits, n = 1,500)
• David Pierce, Library of Congress / CLIR — The Survival of American Silent Feature Films: 1912–1929 (CLIR publication 158, 4 December 2013; 1,575 of 10,919 films — 14% — survive as complete domestic-release copies in their original format, and 2,749 of 10,919 — 25% — survive in some form, pp. 21 and viii)
• CLIR — An Overview of Emulation as a Preservation Method (Kaltman et al., CLIR publication 194, July 2025; cites the 87% finding in the library-science literature)
• PlayStation.Blog — An update on PlayStation Store for PS3 and PS Vita (Sid Shuman, 1 July 2026; phased closure from August 2026 to July 2027)
• Nintendo Support — Wii Shop Channel Discontinuation (points purchases ended 26 March 2018; software purchases ended 30 January 2019)
• Nintendo — Nintendo Switch Online classic games (over 150 classic games across eight systems, accessed July 2026)
• Digital Preservation Coalition — Digital Preservation Awards 2026, industry finalist: GOG Preservation Program (100 titles November 2024 to 267 by February 2026; 1,460+ preservation improvements)
• Video Games Chronicle — Analysis: 2,000 digital-only games will disappear when PlayStation closes its stores (April 2021; roughly 2,200 digital-only titles across PS3, PSP and Vita, about 138 exclusive)
• Video Games Chronicle — Analysis: more than 220 digital games will disappear when the Xbox 360 Store closes (2024; store closed 29 July 2024)
• Game Developer — Remedy’s Alan Wake to be delisted in wake of expiring music licenses (May 2017; seven-year music licences, delisted 15 May 2017)
• Game Informer — Here are the Grand Theft Auto IV songs removed today (26 April 2018; licensed tracks removed as agreements lapsed)
• Forza — Changes in Forza Horizon 4’s Festival Playlist and delisting from digital stores (delisted 15 December 2024, citing licensing and agreements with third-party partners)
• Video Game History Foundation — The VGHF Library opens in early access (30 January 2025; 1,500+ out-of-print magazines) and Our impact in 2025 (20,000 pieces of digital game art)
• SteamDB — Steam Game Release Summary by Year (20,017 releases in 2025)
• Nintendo Support — Service Discontinuation: Nintendo DSi Shop (last day to purchase software, 31 March 2017)
• PlayStation — Important notice: PlayStation Store service changes (PlayStation Store on PSP hardware closed 31 March 2016)
• Antstream — Antstream Arcade games list (service’s own catalogue claim of 1,300+ licensed retro games, accessed July 2026; self-reported, not independently audited)
• Time Extension — The VGHF wants meaningful solutions from the ESA for preserving digital-only content (July 2026; Frank Cifaldi quotation, following Sony’s confirmation it will end PS5 disc production in 2028)
• Rockstar Games Support — Music updates in Grand Theft Auto IV, The Lost and Damned and The Ballad of Gay Tony (publisher notice of licensed-track removals; the 53–54 track total is a GTAForums community tally, not Rockstar’s, and Vladivostok FM was the hardest-hit station at about 15 tracks)
• TechRadar — Over 200 titles will become unavailable when the Xbox 360 store closes down (corroborating count for the 29 July 2024 closure)