I may have possibly discovered the ultimate compression technique

364 views
Skip to first unread message

Immortal Discoveries

unread,
Aug 7, 2019, 5:47:51 PM8/7/19
to Hutter Prize
Hi friends! I wanted to present myself today, I may have possibly discovered the ultimate compression technique. I went through it fully just now and it looks great. I'm expecting I can make the 100MB wiki8 into maybe 1-3MB. I'm going to hire a programmer soon now, because I don't code. So I wanted to get it out there first, in case there is a person claiming to have something very similar to my algorithm. Is it ok if I have Gmails, a YouTube video, and freelance trail history to prove on a prior date that I had the full description of the algorithm first? All 3 have dates of upload, which are set by the website, not me. So I hope those can stand as a proof of dispute if someone happens to claim a very similar algorithm in the next week(s).

James Bowery

unread,
Aug 7, 2019, 6:05:56 PM8/7/19
to Hutter Prize
There is no provision for such a judging criterion, although in the old days of patent law, I believe it was encouraged for people to disclose their inventions even if prior to filing.  Nowadays, however, public disclosure may create problems for the inventor IIRC.  Please consult with an attorney if you want to retain intellectual property rights via disclosure.

The Hutter Prize rules now require disclosure of the compression method as necessary, but insufficient, for a Hutter Prize to be awarded.  Indeed, if someone else publishes your method before you disclose it, but you meet the prize requirements before they do, you win and they lose unless they substantially improve on your compression ratio.

On Wed, Aug 7, 2019 at 4:47 PM Immortal Discoveries <immortal.d...@gmail.com> wrote:
Hi friends! I wanted to present myself today, I may have possibly discovered the ultimate compression technique. I went through it fully just now and it looks great. I'm expecting I can make the 100MB wiki8 into maybe 1-3MB. I'm going to hire a programmer soon now, because I don't code. So I wanted to get it out there first, in case there is a person claiming to have something very similar to my algorithm. Is it ok if I have Gmails, a YouTube video, and freelance trail history to prove on a prior date that I had the full description of the algorithm first? All 3 have dates of upload, which are set by the website, not me. So I hope those can stand as a proof of dispute if someone happens to claim a very similar algorithm in the next week(s).

--
You received this message because you are subscribed to the Google Groups "Hutter Prize" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hutter-prize...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hutter-prize/12cfe96b-e33f-432b-98c9-63b64e063e71%40googlegroups.com.

Immortal Discoveries

unread,
Aug 7, 2019, 7:03:11 PM8/7/19
to Hutter Prize
Yes, there is publishing it first, and patenting, but should I?

Usually, you guys just submit your own code to the prize, with no one coding it for-you. But do yous paten or publish your algorithm compressor?

So, I'm asking, ignoring patenting/publishing at least, that, if I have youtube/etc dates of proof of the exact algorithm, then will Matt himself verify I deserve the prize money?

James Bowery

unread,
Aug 7, 2019, 7:10:15 PM8/7/19
to Hutter Prize
The date of submission of the compressed executable archive (or compressed data and executable decompressor files) is the date of priority, regardless of prior disclosures.

Matt Mahoney has a separate competition called the "Large Text Compression Benchmark" of which he is the sole judge.  After you have read his rules, questions regarding that may be put to him.

--
You received this message because you are subscribed to the Google Groups "Hutter Prize" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hutter-prize...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hutter-prize/9268d141-1a8e-4810-aa56-73ca8d1a3b9f%40googlegroups.com.

Immortal Discoveries

unread,
Aug 7, 2019, 10:18:14 PM8/7/19
to Hutter Prize
Ok thanks. I will have to think out how to avoid my coder submitting it first.

Do yous usually use C, or C++? Which coder should I hire :-) Which is smaller in size...

Immortal Discoveries

unread,
Aug 7, 2019, 10:18:14 PM8/7/19
to Hutter Prize
"Create a C Windows executable of size."
"Create a compressed version (self-extracting archive) of the input file."

Question, how do I say this to my programmer, I mean, what does this mean? A C.exe file inside a zip file unzips itself !, and decodes into 100MB? :-) Just have to make sure, go into detail...

Matt Mahoney

unread,
Aug 8, 2019, 4:41:21 PM8/8/19
to hutter...@googlegroups.com
Most data compression developers use C++, sometimes C and possibly assembler. But you can use any language you want because you will be compiling to .exe. Python will probably be too slow.

You can use a dictionary or other files but they are included in the size of your decompressor. 

If you hire a programmer, then you have to negotiate the terms. We can pay teams if you tell us how the prize is to be divided.

Some personal advice. I seriously doubt you can achieve 1 to 3 MB. The current result of 15 MB is the result of years of research and development. A competent programmer will probably spot the flaw in your algorithm before they start coding. If it was me, I wouldn't take the job and be blamed for not producing working code.

You may find my online book useful. 




On Wed, Aug 7, 2019, 10:18 PM Immortal Discoveries <immortal.d...@gmail.com> wrote:
Ok thanks. I will have to think out how to avoid my coder submitting it first.

Do yous usually use C, or C++? Which coder should I hire :-) Which is smaller in size...

--
You received this message because you are subscribed to the Google Groups "Hutter Prize" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hutter-prize...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hutter-prize/ae20557f-6259-49b4-8015-c1d4da6baf70%40googlegroups.com.

Immortal Discoveries

unread,
Aug 8, 2019, 4:54:10 PM8/8/19
to Hutter Prize
Maybe you mean to compress the wiki8, and then put it in a zip to compress it yet more? Of course I will...

Immortal Discoveries

unread,
Aug 8, 2019, 4:54:10 PM8/8/19
to Hutter Prize
Ok thanks so much.

Yes I know, my 1MB or less claim is crazy, but it looks fleshed. Implementation will tell the tale. (My idea works on any txt file, even jumble without patterns, and will on image/audio.) Yet more impossible! :-)

James Bowery

unread,
Aug 8, 2019, 4:59:41 PM8/8/19
to Hutter Prize
Don't bother hiring anyone.  The only way you can compress a file that is reasonably compressed is to, in essence, first decompress it and then compress it with another compression algorithm.  You really need to read up on algorithmic information theory as Matt suggested.

On Thu, Aug 8, 2019 at 3:54 PM Immortal Discoveries <immortal.d...@gmail.com> wrote:
Maybe you mean to compress the wiki8, and then put it in a zip to compress it yet more? Of course I will...

--
You received this message because you are subscribed to the Google Groups "Hutter Prize" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hutter-prize...@googlegroups.com.

James Bowery

unread,
Aug 8, 2019, 5:03:06 PM8/8/19
to Hutter Prize
Here's a great answer by Matt at Quora:

https://www.quora.com/Can-compressed-files-be-compressed-further/answer/Matt-Mahoney-2 

The money paragraph:

There is no general method to recompress files or to compress every possible file larger than some threshold. If there were, then it would be possible to compress every file down to this threshold in multiple passes. That would imply a 1 to 1 mapping between the infinite set of possible uncompressed files and a finite set of possible compressed files.  
 

Immortal Discoveries

unread,
Aug 10, 2019, 8:53:13 PM8/10/19
to Hutter Prize
I know...but the idea is really simple, unique, and exciting. We'll find out soon.

James Bowery

unread,
Aug 10, 2019, 8:56:35 PM8/10/19
to Hutter Prize
I'm approving this message for distribution in the interests of fairness, and to reduce the noise level I'm terminating this discussion until such time as there is a prize entry.

On Sat, Aug 10, 2019 at 7:53 PM Immortal Discoveries <immortal.d...@gmail.com> wrote:
I know...but the idea is really simple, unique, and exciting. We'll find out soon.

--
You received this message because you are subscribed to the Google Groups "Hutter Prize" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hutter-prize...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hutter-prize/c34f47af-e548-4d2a-b35e-41706ab513d3%40googlegroups.com.
Reply all
Reply to author
Forward
0 new messages