That page mentioned above is a really great read. Everyone who hasn''t should go check it out.

Here''s a quote I had to grab:
"... the compression algorithm used by DataFiles/16 is not subject to the laws of information theory."

That would be like NASA saying:
"Our space shuttles are not subject to the laws of physics."

Funny stuff.

Jesse Chounard
stinkygoop@crosswinds.net

It should be possible to test the software without keiren sending it to anyone by implementing the compression and decompression programs as cgi apps, then making a webpage to allow the upload of a file, the compression or decompression, and the download. That way the code executes on keiren''s server only, so we couldn''t get at it to disassemble. I''m not sure how to get cgi to upload/download, but I know that many sites that offer free webspace do it all the time.

Anyone know how to set this up?

aig

I think I remember a few years back some kid for melbourne(Australia) came up with some spanking compression for sreaming video that was meant to make videophones viable over copper.

But I''d still like to see your code released

As a matter of interest I was wondering how well the algorithm works on miniscule amounts of data, like sub 1K strings of numbers...

gimp-o

i didn''t want to say any more to this ridiculous post, but now that so many people actually BELIEVE what kieren_j claims, i just feel obliged to comment.

i won''t try to explain WHY it is impossible to compress true random data up to a certain amount, because it has been done so very good already. look what deathlok
said, even my little brother will understand why CAR cannot work.

i just makes me somehow sad that there are ppl on this board who even consider the possibility that he found something revolutionary.
i MIGHT consider that he found a good new algorithm that is slightly better than the common ones today, (mind you, i said *slightly*) but even that is very unlikely.
just imagine how much money is put in research on this subject, and did someone find out something like that? no. of course not.

kieren_j, please stop trying to explain yourself, either give us a demo (i would be satisfied if you only put 1000 chars uncompressed and then the compressed ones on the board) or shut the hell up.

First AP: The reason why you don''t need to describe much is because you just repeat the pattern throughout the file. You could also use bit masks.

And I read the compression FAQ. I still think there''s a definite possibility that he''s right.


Lack

Christianity, Creation, metric, Dvorak, and BeOS for all!

hey keiren.... do this:

1) take 10-5 files of very different sizes and types
2) compress them
3) decompress them
4) test if they are exactly the same as they were

if they are , your CAR compression should be working right
(i hope they are)


"Your thoughts are the reality, the phisical side is the illusion"

Indeed Ridcully,
I had made it a point not to post here for a while, because everyone here seems to ignore all those minor details like math and logic. Thank you for posting though, I was bordering on doing so again. Of course, the fact that this many people believe him just illustrates a point I believe: People without formal education in computer science/math/etc... are so very quick to dismiss the need of such education because they have mad l33t skillz. Then, something like this comes along, and the true value of such an education demonstrates itself.

I realize everyone here seems to think that I''m somehow jealous or bitter... but I''m not. I have no reason to be, because I KNOW that kieren has made a mistake somewhere. Probably the same kind of mistake I made a long time ago when I thought I had revolutionized data compression with an amazing data-reordering algorithm. No, I''m not making this up.

I don''t really have the time to keep up with this nonsense and continue to illustrate the intuitiviely obvious reasons that kieren MUST be mistaken, so I suggest that you all just head on over to the nearest compression theory newsgroup on Usenet. Oh, and when you do, save a log of it all for me, there ought to be enough flame to heat my apartment all next winter.

-Brian

i didn''t see it mentioned anywhere that kieren_j can compress ANY 100mb into 10mb. maybe that was just an exceptional case. still, if it only works for some file types, or only produces a file 20% smaller in most cases, that''s still improvement.
i hope this doesn''t go onto a 6th page.

crazy166
some people think i'm crazy, some people know it

It''s amazing how one post can span 5 pages of CRAP. So I''m going to put my crap with the rest of the crap, and soon there will be 6 pages of smelly cr... I''ll just get on with it.

If it really does exist, I think Kieren should patent the algorithm, and then send a copy of the source code to a company like IBM or something, and ask them to check it over. This way he can be sure that there are no bugs in the code, and that it works efficiently. IBM wouldn''t be able to rip the source code because of the patent, and you could prove that it really does work. The only reason I suggested IBM is because some chaps there came up with the now called DES algorithm.

Personally I think Kieren is the best wind-up artist ever to grace this message board, as I think everyone needed a good laugh (I sure as hell had a very large one).

One last thing. Is is possible that the people who worked out that information data thing (I cant remember the full name) could be wrong? Bah, doubt it.

MENTAL

PS: If Kieren really does have this algorithm, then RELEASE A DEMO. You cannot expect people to believe you when all your doing is posting what could be considered fictitous information. Making the type smaller does not make is proof

Why do so many people believe kieren since his last post?? Think about it on your own! You can''t compress random data - and compressed files are nearly random - because with n bits there are 2^n possible patterns and if you want to give each pattern a number you''d need a data type which can hold 2^n values - and such a data type would need n bits so there''s no compression. Compression algorithms work only if there''s a regularity in the pattern and the number of patterns with regularities is much smaller than 2^n and you can''t find them in compressed files.

Visit our homepage: www.rarebyte.de.st

GA