People have messed up crypto even by almost cloning good implementations on accident leading to compromises.  If they do not understand why publishing the hashes are important then they may not consider it bad to stick it out of the way to avoid the clutter of information many average people just  wont understand.

real life example: CEX messed up crypto for their API, it is a HMAC done by someone who does not understand what all HMAC is supposed to do.  They use it just to authenticate a user not the user+message.  The message itself is never HMACed and thus could be altered in transit (sha1 ssl cert not withstanding and while sha1 is broken and has been for  years no "in the wild" attack has been discovered and broken means different things to cryptographers than average people although sha1 is pretty bad).

Sony and some Bitcoin  wallets with weak random in ECDSA is another good example.  Although Sony was worse because it was not just a  group of related transactions it  was many transactions separated by a large time window.  There are something like 300 weak r wallets out there.  Most addresses havent been used in years though.  A few are still in use with instant withdraws on  the few thousand satoshi transmitted through at infrequent intervals.  No idea if someone is sweeping coins or if the legit wallet owner is transferring but the transfers out come in within 1 second of the deposit.  Last transaction I saw on a weak r wallet was Jan 31, 2015 for 6000+ satoshi.  Oh I stand corrected Feb 8, 2015 for 6717 satoshi.

End result by effectively cloning a REST API they have done it in a way that its horribly broken.  Other examples include some other crypto libraries where it tries to short circuit the encryption and returns early upon a mismatch.  Timing attacks then ensue and you just brute force the password one character at a time until you have it.  This is akin to an oracle attack on crypto which is another example of how good crypto can be improperly implemented merely by different error messages or return values.

I can see it being innocent that 999dice just thought they would move the cruft to the side without thinking about the effects - effectively removing "provably" from "provably fair".  Either they are clever cheaters or of a more innocent mind that they do not see the potential for evil in the implementation.   Without knowing more about the specific individual (not the speculated one and then more than "well he is a known scammer" not all con artists are the same) it is hard to tell.