Pure speculation, but perhaps they thought they were making a compromise between obfuscation and performance. I've heard this argument made in similar contexts

Pure speculation, but perhaps they thought they were making a compromise between obfuscation and performance. I've heard this argument made in similar contexts on the theory that software implementations of 1DES should be faster than software AES-128 or other alternative algorithms. In practice, however, the performance between 1DES and AES-128 is not that large, and probably not large enough to matter.