A while back, we were discussing getting around image censors.
Check this out.
>https://tineye.com/search/acf135ced2d67f21b883876f956d438386da7413/?sort=score&order=desc
>tineye doesn't see much a difference between that attached OBEY image and the much more common BETO image
I think that by taking known liberal images/logos and simply rearranging letters to make them virtually identical to the statistical machine learning, we can easily get through censors. This is easy enough to do with any default image program like MSPaint.