Several times last year Australian games ratings have been reported for arbitrary ratings assigned under the Australian Classification Board's IARC automated game and app rating tool.
Variants of the same game on different platforms appeared in
the classification database with wildly different outcomes. One game achieved being 15 rated, 18 rated and banned. Inevitably when the shit hit the fan and the incompetent ratings gained the attention of publicity, human censors stepped in and sorted out
the rating (down to 15), and expunged all the embarrassing misfires from the database.
Well it seems that the shoddy system has been discussed for a while and a damning report from 2016 has just been published as a result of a Freedom of
The report reveals that a selection of ratings from the tool were audited by compared them with an assessment from a human censor.
Results were particularly atrocious fro the higher ratings. A table on page 13 reveals
- 56% of M (PG-15) ratings assigned by the tool were wrong
- 72% of MA 15+ ratings were wrong
- 100% of R 18+ ratings were wrong
- 99% of RC (banned) ratings were wrong
In all of these categories the automated ratings were nearly always lowered by the audit.
The failure of the system was attributed to the inaccuracy of data input but surely this is a systemic failure to define tight enough definitions of date