At 23:22 08/05/2007, you wrote:

>>> see:
>>> about the " Sloot Digital Coding System". great stuff.
>>        Danke.
>that seems to be German, in Dutch "Bedankt" would be appropiate. i
>appreciate the effort though.
>>I found the english wkik entry so I could understand the
>>        piece.  Since Sloot is dead, no way of knowing.
>actually there is. the Dutch wiki-article has much more detail about the case.
>this Sloot guy claimed he could compress any movie into 1kb ( 1024
>bytes ). he stored this 1kb movies on smartcards, which he would feed
>into his magic machine to show the movies.
>in The Netherlands there was press-coverage about this, and he was
>able to attract investors and even a ( up to then ) reputable IT-guru
>assiocated with Philips to back him. they found silicon-valley
>investors who were interested.
>he died / killed himself / was murdered before his hoax could be uncovered.

It's easily covered. Check usenet comp.compression faq. It's the "Counting 
Theorem". For a sequence of n-bits there are 2^n possible files (f.e. there are 
256 files of 8 bits). Unfortunately there are only (2^n)-1 possible files 
lesser than the original (f.e. there are 128 files of 7 bits, 64 of 6, 32 of 5, 
16 of 4, 8 of 3, 4 of 2 and 2 of 1 bits, total, 255 files) so using an 
algorithm you can't compress all n-bits sequences.

For this case, you can easily check that using this guy algorithm, you can have 
only 2^8192 movies. Perhaps you think they are a lot, but nearly all are white 
noise movies.

There are 1-2 monthly of this claims on comp.compression. Normally they are 
beginners to Information Theory / Compression but others are simply cheaters.

>sounds a lot like the perpetuum mobile stories.

In fact, it is. Information Theory uses entropy concept too and claims like 
this breaks the 2nd and 3rd principle.

>>But this is the
>>        kind of leap forward that would save, oh, a few measley
>>        $Billions.  And give millions of us faster and broader access.


_______________________________________________ mailing list
To unsubscribe, send any mail to "[EMAIL PROTECTED]"

Reply via email to