The shelf report — what survives, and what actually predicts it
The audit in four charts. The interesting result is not how many games are gone; it is that whether a game survived has almost nothing to do with the game.
18
Entries audited
33%
Still available
7
No publisher established
8
Studios checked
Where the entries stand
Every status confirmed by direct request on 31 July 2026.
The distribution is bimodal and there is very little in the middle. Entries are either fully available from a publisher that never went away, or fully gone. Almost nothing is partially reachable, which is what you would expect if the deciding factor were a single binary condition rather than a gradual decay.
When these entries were first published
Coverage by year, from the date each page first appeared.
Coverage clusters in the middle of the decade and stops. That shape is what makes the audit possible: the list was frozen rather than curated forward, so it records a set of studios as they actually stood, without the survivors being retrospectively favoured.
What predicts survival
Publisher continuity against everything else.
The correlation is total in this sample, which is small enough that the finding should be stated modestly: within these 18 entries, every game still available comes from a publisher that never stopped trading, and no game from a publisher that went dark remains available from any source this audit could find.
Nothing about genre, quality, level count or contemporary reception separates the two groups. The variable that separates them is whether an organisation continued to exist. For anyone wondering what preservation actually depends on, that is a more uncomfortable answer than technology.
How to read these numbers, and how not to
A small sample, an unambiguous pattern, and the limits of both.
The sample is 18 entries. That is large enough for the pattern to be worth stating and far too small to support a claim about shareware in general. What can be said is narrow and solid: within this set, publisher continuity and commercial availability line up completely, and nothing else lines up with either.
The set is also not random. It is what one site chose to cover between 2004 and 2010, which means it is biased towards games that site found interesting and towards publishers willing to supply a demo. If anything that bias should favour survival, because a studio organised enough to run a press relationship is a studio organised enough to renew a domain. The results are worse than that bias would predict, which is the mildly alarming part.
The one thing these figures cannot show is what a broader audit would find. Whether the pattern holds across the thousands of titles sold this way in the same decade is a question this archive is not big enough to answer, and stating it as though it were settled would be the exact kind of overreach that makes a dated audit worth less than an undated guess.
Questions about the report
Findings, limits, and how to cite the figures.
What is the headline finding of the shelf report?
Of 18 entries, 6 — 33 per cent — are still commercially available, and every one of those belongs to a publisher that never stopped trading. Survival tracks the publisher, not the game.
Why are so many entries recorded with no publisher?
7 of 18. The pages those entries are drawn from carried a description, a screenshot strip and system requirements without a studio credit. The gap is shown rather than filled, because a plausible attribution would look like knowledge and would not be any.
Can this report be reused elsewhere?
The charts are hand-coded scalable vector graphics in the page itself, so they can be viewed at any size and read by a screen reader. Anyone citing the figures should quote the check date with them, since availability is exactly the kind of claim that goes stale.