Jewtube is flagging random German music as offensive

youtube.com/watc h?v=awrM-F4z3T0

As you can see, anons, Jewtube is now flagging anything even remotely German and military related as offensive. Like this World War 1 music.

Just in case it doesn't work in embed for (((some reason))), here's the direct link.

But more importantly, since we all know Jewtube is fucked already, how can we exploit this?

Simple. This is an automatic, report-based system. So what do we want to report? Which leftist faggot channel should we report-bomb first?

Other urls found in this thread:

youtube.com/watch?v=-PM9JUe98Ig
hooktube.com/watch?v=-PM9JUe98Ig
hooktube.com/watch?v=SBnXCotdkaE
twitter.com/NSFWRedditVideo

Kike-free first post.

Also, you're going to have to open that link in a seperate tab for it to show the block.

Thanks. Bump.

what the fuck, since when is youtube bluepilled?

No matter how many fucking times Germany apologizes for the Holocaust and lets refugees outbreed indigenous Germans, it's still offensive for some retarded reason. Unbelievable.

Reported for not even trying.

German is the best version.

We should pick a closely related form of music then start mass flagging videos to mess up their machine-learning algorithm.

There is nothing more disgusting to me than the kikes and leftyshits trying to destroy and obscure history. If we don't kill them all, and soon, who knows how much history will be buried. Normie whites who still think this is all some kind of joke and that all that needs to happen is that we should just convince everyone to vote republican deserve the rope too.

As an EECS grad, I agree with this. Machine learning algorithms sound complicated and esoteric, and they can be highly accurate at tasks they are given (such as determining whether or not a given video expresses "approved" ideas), but they're highly vulnerable to malicious tampering with the data they're using to learn.

It's probable that youtube is using user flagging data for offensive content to train their machine learning algorithm. In other words, they feed videos flagged as offensive into the algorithm to extract data points so it will be able to flag such videos automatically in the future. Given the mass quantities of data required to train a machine learning algorithm, I highly doubt this input data is manually screened to any large extent. If we flagged a lot of bluepilled videos as offensive, it would mess up the algorithm.

The only issue is automated filtering of "good" and "bad" reports. Google might not be able to manually filter the reported videos, but they could do automated filtering by ignoring reports made with 8ch.net in the referrer field (or with an empty referrer field, as most normalfaggots don't disable sending referrer data), or a large number of reports from the same IP address. They could also add a cookie to your browser for each report, to filter out people making many reports from different IP addresses. They could even ignore reports where the user agent field is less common (such as Firefox on Linux), as chan users are less likely than average to use Windows 10 and Google. Needless to say, they could also filter reports from ungood youtube accounts, if you have one.

So if you're going to screw with the machine learning algorithm, make sure to look up the video through a regular search engine instead of going there directly from Holla Forums, have the referrer field enabled, have your user-agent set to some common value, either don't log into a youtube account or log into a series of fake accounts, and change your IP address frequently through VPNs, proxies, Tor, etc. This way it will be difficult to filter "good" reports from bad reports.