Looool this is the ultimate way to prove new records... seriously

POSQ7GZJBGIVJB87

This is what I just did first try, with a broken controller, tired, and absolutely rusty xD

Very bad time Chris

But seriously.. that's awesome

So is this gonna be the new way to submit times then?  

I would advise caution. For instance, if I change the P at the start of Chris's code into an A, it gives me 1'06"868. Obviously that is unrealistic (it is in fact impossible with a fastest lap of 26"942), but with some more trial and error one might be able to find something that is not beyond the realm of possibility. Perhaps one could even drive a certain poor time on purpose if one knows how to subtract a certain amount of time, in order to get something that is just below the WR.
Also, the fact that changing this one character only changed one of the data makes me suspect that the code is not that sophisticated and could be broken.

I've got some mkdd times to submit! Although mk64 keeps calling me back so I haven't had time yet.

This is quite awesome, and this cool code thing might give me the motivation to finally join here.

Edit: our sites are on a trust system anyway for the most part... so I really don't think a decoder would matter.

Double edit: I would only count it as a form of proof up to the top 100

Lol, that's genious and true.. I still don't see the reason why the hell people would cheat in mariokartgames

Proof standards here are pretty lax compared to some other sites.  Tricks like this just give people excuses to "test" the trust system.  

I am the one who wrote the flash version of the decoder. I wish that I could claim credit for figuring out the code format, but that was a random guy I met on IRC.

All I did, aside from porting his C code, was compiling the character and map name lookup tables as well as identifying two fields he didn't.

Here is some info on the security of the codes: nonstandard character numbering, 5 bit checksum and a rng scrambler for all the data with a 16 bit seed.

It lacks redundancy and bit juggling, making it easy to figure out which characters in the code carry which data by getting lucky with the too small checksum.

In short, it's nothing like the animal crossing codes, those used multiple layers of real crypto.

Still, it is better than a pure honor system, since it is at least a little work to bruteforce the checksum and the scrambler makes it impossible to pick arbitrary times.

But it's rather useless when it's so easy to just share a ghost savefile.

There is however two mystery values I never figured out. One 2 bits large and one 1 bit large. I checked, it doesn't appear to be the 50/60 Hz status.