Here’s the specific line that cites AFL++. > “Also, we use the implementation provided by [17] for Radamsa, Redqueen, and LAF-INTEL. Note that autofz can adopt

Here’s the specific line that cites AFL++. > “Also, we use the implementation provided by [17] for Radamsa, Redqueen, and LAF-INTEL. Note that autofz can adopt any fuzzer sup- porting seed synchronization, and no fundamental obstacles prevent their integration. Therefore, autofz can truly enjoy the benefit of increasing diversity of the fuzzers.” From my read of it, they took very specific parts of AFL++, but not the whole thing. The strength of AFL++ comes from having many optimized subcomponents, which is similar to the goal of this paper. Why not compare against the optimized AFL++? This leads me to suspect that autofz compares poorly to AFL++.