I have reproduced both MD4 collisions from the recent paper. The example given had endian problems similar to those noted by Eric Rescorla for the sorta-MD5 collision. Also similar to Eric's results, the hash value (while a collision) does not match what the authors give in the paper.
Example one: $ od -tx1 file1.bin 0000000 83 9c 7a 4d 7a 92 cb 56 78 a5 d5 b9 ee a5 a7 57 0000020 3c 8a 74 de b3 66 c3 dc 20 a0 83 b6 9f 5d 2a 3b 0000040 b3 71 9d c6 98 91 e9 f9 5e 80 9f d7 e8 b2 3b a6 0000060 31 8e dd 45 e5 1f e3 97 08 bf 94 27 e9 c3 e8 b9 0000100 $ od -tx1 file2.bin 0000000 83 9c 7a 4d 7a 92 cb d6 78 a5 d5 29 ee a5 a7 57 0000020 3c 8a 74 de b3 66 c3 dc 20 a0 83 b6 9f 5d 2a 3b 0000040 b3 71 9d c6 98 91 e9 f9 5e 80 9f d7 e8 b2 3b a6 0000060 31 8e dc 45 e5 1f e3 97 08 bf 94 27 e9 c3 e8 b9 0000100 $ cmp file1.bin file2.bin file1.bin file2.bin differ: char 8, line 1 $ openssl md4 file1.bin file2.bin MD4(file1.bin)= 4d7e6a1defa93d2dde05b45d864c429b MD4(file2.bin)= 4d7e6a1defa93d2dde05b45d864c429b Example two: $ od -tx1 file1.bin 0000000 83 9c 7a 4d 7a 92 cb 56 78 a5 d5 b9 ee a5 a7 57 0000020 3c 8a 74 de b3 66 c3 dc 20 a0 83 b6 9f 5d 2a 3b 0000040 b3 71 9d c6 98 91 e9 f9 5e 80 9f d7 e8 b2 3b a6 0000060 31 8e dd 45 e5 1f e3 97 40 c2 13 f7 69 cf b8 a7 0000100 $ od -tx1 file2.bin 0000000 83 9c 7a 4d 7a 92 cb d6 78 a5 d5 29 ee a5 a7 57 0000020 3c 8a 74 de b3 66 c3 dc 20 a0 83 b6 9f 5d 2a 3b 0000040 b3 71 9d c6 98 91 e9 f9 5e 80 9f d7 e8 b2 3b a6 0000060 31 8e dc 45 e5 1f e3 97 40 c2 13 f7 69 cf b8 a7 0000100 $ cmp file1.bin file2.bin file1.bin file2.bin differ: char 8, line 1 $ openssl md4 file1.bin file2.bin MD4(file1.bin)= c6f3b3fe1f4833e0697340fb214fb9ea MD4(file2.bin)= c6f3b3fe1f4833e0697340fb214fb9ea David --------------------------------------------------------------------- The Cryptography Mailing List Unsubscribe by sending "unsubscribe cryptography" to [EMAIL PROTECTED]