eldervanethe service record

Method revision 3 · in force since 2026-02

How we read a service history

Two dates, four states, and a list of things we cannot see.

Upkeep measures maintenance, not quality. A game can be dormant and wonderful. It can be actively maintained and tedious. This method never touches that question.

The two dates

Everything on this site is built from a comparison of two public figures.

Android date
The "Updated on" line on the Google Play web listing, read with the Canadian storefront (gl=CA).
iOS date
currentVersionReleaseDate from Apple's public lookup endpoint with country=ca. Same field Apple's own store page renders.
First release
releaseDate from the same Apple response, cross-checked against the Play listing where it is shown.
Read date
Printed on every entry. Store pages change without notice, so a date without a read date is worthless.

Both are shown in ISO form, year first, because en-CA readers see both day-first and month-first conventions in daily life and 06-07-2026 is genuinely ambiguous here.

The four states

In service
Both listings updated within the last 12 months, and fewer than 12 months between them.
Split
More than 12 months between the two store dates, regardless of how recent the newer one is. One platform is maintained, the other is not.
Dormant
Neither listing updated in 24 months.
Out of print
A platform has no listing at all. Not for sale there at any price.

The twelve and twenty-four month lines are arbitrary. They are set there because roughly one major OS release passes in a year, and a build that has survived two without a rebuild is being carried by compatibility layers rather than by anyone's attention. Argue with the boundaries if you like; they are printed so that you can.

What else gets recorded, and why

Where this method fails

Four places, and they are not small.

1. Reviews often cannot be read at all

Recent App Store reviews come from a public RSS feed that is inconsistent to the point of uselessness. On 2026-07-17 it returned reviews for one of the six titles in the record and nothing for the rest. Three days later it returned nothing for that one either. So where you see quoted complaints, they were captured on a day the feed worked; everywhere else the entry says not confirmed, and that gap is the source's fault rather than the reader's problem to guess around.

2. An update date does not tell you what changed

A store date moves for a recompile, a privacy label change, a store policy compliance patch, or a full content release. From outside there is no way to tell those apart. A recent date proves someone touched the build. It does not prove they fixed anything.

3. We cannot test at scale

Two phones, one of each platform, both a few years old. When an entry says a game is reported broken, it is reporting other people's evidence, attributed, not a measurement of our own.

4. Some studios never say anything

Nobody announces the end of support for a mobile game. There is no notice, no changelog entry, no farewell. The only visible symptom is a date that stops moving, which is why this record is built out of dates rather than announcements.

What would make us change an entry

A new store date, a corrected figure, a developer statement, or a reader pointing out that a listing has moved. All four have happened. Corrections go to [email protected] and the read date on the entry moves when they land.

See the method applied Send a correction