I'm sorry, but there is no better method to guess a random number. The ripemd160 function doesn’t somehow “understand” that it should generate less likely similar prefixes if I input the hash of public keys derived from nearby private keys. That would require consciousness, and it would need to be able to reverse sha256 and secp256k1—which I very much doubt it can do.
I suppose finding two "aa" in 256 is just as likely as finding one "aa" in 256. After all, hashes don’t "understand" that... lol.
The mistake lies in comparing the two methods while keeping the same block search order. That is, you're forcing both methods to iterate through the blocks in exactly the same order.
I'm sorry, in the Olympic competitions, two archers shoot at two different targets, and the runners race on different tracks.