IMPORTANT NOTE: in the time since the videos discussing this issue went live on the 29th November, and writing the below (30th/1st), YouTube appears to have removed all traces of the 'secret demonetisation codes' from source. They can no longer be found using the search parameters outlined below. This literally (lidderally) happened whilst this post was being written (occurred some time between 29th Nov and 1st December). The fact the codes were removed without any notification or comment (at time of writing at least) could be considered odd to say the least.• • •
In trying to get to the
bottom of the YouTube video demonetisation #adpocolypse, the controversy seems to have taken a step or two into darker murkier waters, or so
at least some commentators on the subject are saying, the TL:DR of which suggests YouTube makes use of super-secret hidden codes to "censor" YouTube Creators and their video content (
this does not relate to/is separate from videos marked as "not eligible for monetisation"). Naturally this means doing some fact-checking.
So, do the 'super-secret codes' exit?. Yes (
although the numbers are not exactly super-secret - see below, what they mean is (or was - see note above)).
Whilst all the alleged codes are time-consuming to verify a number of them can be found right-clicking a page in which a YouTube video appears and searching the html source for two markers, "
excluded_ads"
and "
excludedAds", which can be found in
two locations, both part of a script function,
not just the one as
described by the authors linked above.
With that said, not all videos carry the numbered tags; those that do, may not have been specifically content or ToS flagged; those that have been, don't necessarily continue to carry any if at all the tags once checked and/or remonetised (
DMCA flagged videos, and those with other specific Copyright violations, don't seems to carry the tags once cleared).
The lack of transparency on this, at least in regards to the codes being 'behavioural markers' like the red-sticky spots teachers might place on a child's report card/progress report, is more than a little confusing and concerning. For example, the following video, "
Easily export a map to Unity3D", has never been flagged or demonetised and yet it carries a number of the numerical tags being reported on, the critical values precede the "=";
Tag 1: "excluded_ads":"46=14_14;59=14_14;64=14_14;76=2_2_1,2_2_4",
Tag 2: "excludedAds":"46=14_14;59=14_14;64=14_14;76=2_2_1,2_2_4",
That's "
46", "
59", "
64" and "
76". These particular numbers appear common to almost all KatsBits videos, few of which have ever been flagged. So rather than being behavioural red marks, this fact perhaps hints at their likely being '
advert category identifiers'; "46" might be "
personal care", or "
Holidays and travel", both of which might be less valuable to Advertisers placed on education videos about Blender or content creation, so videos being tagged to identifying this incongruity makes sense.
In addition to this likely reason for some of the codes, the other half of the equation is a little more troubling because they do only appear on videos that have been flagged as "
not suitable for most advertisers". For example the number "
102" appears in code for the following video "
Monday #imvufbx shenanigans w/ IMVU's JinxyWithMoji & CodeRodent" in addition to the aforementioned common codes;
"excluded_ads":"46=14_14;59=14_14;76=2_2_1,2_2_4;102=1_1,1_2_1,1_3,2_1,2_2_1,2_2_4,2_3,14_14,17_2_1,17_2_4",
"excludedAds":"46=14_14;59=14_14;76=2_2_1,2_2_4;102=1_1,1_2_1,1_3,2_1,2_2_1,2_2_4,2_3,14_14,17_2_1,17_2_4",
This is where things get a little more serious because the codes, it's being alleged, appear to be policy violation identifiers, content deemed to contain "
Controversial issues and sensitive events", "
Drugs and dangerous products or substances", "
Harmful or dangerous acts", "
Hateful content", "
Inappropriate language", "
Inappropriate use of family entertainment characters", "
Incendiary and demeaning", "
Sexually suggestive content" and/or "
Violence". Each 'category' is assigned a unique tag that can be innocuously dropped into page source such that the KatsBits video linked above can be flagged with "
102" identifying it as containing "
Inappropriate language" (at least according to the research linked to up-top).
As
discussed in a follow-up chat (also demonetised at time of writing), unless the word "but" is considered inappropriate it's difficult to imagine exactly how YouTubes system works aside from it automatically flagging 'live' videos, obligating Tubers to challenge, just so the process teaches the AI something or other about content flagging/making course corrections. Even if this were true, the implications for 'marking' certain content, and users, as a matter of course regardless, doesn't bode well in any sense.
Who knows at this point...