This is a good point, and yet -- as high as position 24 (out of 208) in the list, we have "bsc-m03 0.4.0" at only ~160ns per byte compressed (compared with most above it being two or three orders of magnitude slower). Below position 24, there's one in the 20s of ns per byte. Looks like the cost was in the size of the compression program. Not sure if there was any "cheating" going on, hunting for benchmark scores with these two famous test bodies of text.

By comparison, the top contender clocks in at ~240,000 ns per byte (making your point).