I believe the parent is a proponent of the theory of racial relations that defines racism as "prejudice + power". I don't think I'm eloquent enough to sum up the idea, so here's a link to a paper that analyzes this proposed meaning among others: http://www.andover.edu/About/Newsroom/TheMagazine/Documents/...

It's not a useful redefinition. Certainly, degree of power affects the impact of racism, but there is always some power involved in any prejudice, and even prejudice by members of a generally disadvantaged group against members of a generally advantaged group will end up manifesting in contexts where the individuals involved are in different power relationships than the group power dynamics in the broader society would suggest.

In fact, the redefinition is often invoked to justify as not-racist acts that are racist even in terms of the redefinition when the context of the action is considered, because it's usually coupled with a misdirection about power relations in a different, usually broader, context than is relevant to the action in question.

Avoiding the redefinition entirely and talking about racial bigotry and the power relations which magnify it's impacts as distinct and interacting things, rather than trying to set an arbitrary standard of what degree of the latter is necessary for the former to deserve to be called racism is far more productive, if not as convenient for providing certain racists cover.

Why do the people trying to rearrange the meaning of the term "racism" always seem to also be racist in terms of the traditional meaning, which is discriminating against others based on race.

If

    racism = prejudice + power

then may still be possible for a local maxima of power to exist.

When a bunch of black power advocates beat a white person because nobody is around to stop them, they clearly show that they are the local maxima of power. Even if "systemic oppression" occurs as a global maxima, there is certainly a lot of variability of who's in power in the rest of the graph.