How much compress text worth ?

[Obama]

I develop a algorithm can compress text from 10000 letters to 9 letters ,it can decompress so.But useless for me , i want to sell it .

[CompressMaster]

What kind of text? Are there many repetitions? If so, it´s easily compressible.
If your file is random, then you are out of luck. But random content is compressible at least to few percent, but not that big as you have mentioned. But never say never - I am working on my custom data compression software that will be able to handle ANY filetype and compress it at least to 90% losslessly, but it will be terribly slow.

Could you post some screenshots of your algorithm or at least compressed sample? Maybe then we can tell you more about it and also we can help you to compress it much better.

Thanks.

[Obama]

Sorry , i can't show you. mine just algorithm not software so got some lack. It can compress random letters and 10k letters is random.

[Shelwien]

You don't need to sell it directly.
Just apply to http://prize.hutter1.net/ or http://mailcom.com/challenge/ or https://marknelson.us/posts/2012/10/...turns-ten.html .
There're also plenty of other contests where you can advertise your work.

[Obama]

thanks.But my algorithm not so advance , can't reach at that level.

[Shelwien]

Just split it to blocks. If you can compress 10000 letters to 9 bytes, it means you can split enwik8 to 10k blocks and compress them to 10k*9=90k total.
It means you can claim the whole 50k euro prize.

[Obama]

but when decompress it need huge database or super computer and it cant compress chinese word .That's why i want let it go.

[CompressMaster]

Try to compress these pure random files and post compressed results in attachment.

Database size is not a problem for me and chinese strings can be completely filtered. So that´s not problem. I don´t need to decompress your results back to original files, I want only compressed archive of original files. Thanks.

[Obama]

for 1 million random alphabet letters.txt -1028 bytes result

[Sportman]

1 million random alphabet letters.txt -1028 bytes

You mean 1,000,000 bytes input, 1,028 bytes output is 973 times smaller?

Did decompress also work and is file compare output equal to input?
What file size has your compress and decompress software and do it use a database, if yes what size has the database?
How long took it to compress and decompress?
What program language did you use?
Any idea for what price you want to sell your algorithm?

[Obama]

You mean 1,000,000 bytes input, 1,028 bytes output is 973 times smaller?

Did decompress also work and is file compare output equal to input?
What file size has your compress and decompress software and do it use a database, if yes what size has the database?
How long took it to compress and decompress?
What program language did you use?
Any idea for what price you want to sell your algorithm?

Yes, 1,000,000 bytes to 1,028 bytes.
Yes,equal if try with short compress.
few GB.
Few day
Python,but i can use any language cause i got algorithm.
No idea,just offer me.

[compgt]

Yes, 1,000,000 bytes to 1,028 bytes.
Yes,equal if try with short compress.
few GB.
Few day
Python,but i can use any language cause i got algorithm.
No idea,just offer me.

Man, 1028 bytes? Seems to me like my output 1K frequency table plus the famed 32-bit filesize.

You might have an algorithm for some tailored inputs, but sure you can't compress *all* the 1,000,000-byte random files into just 1028 bytes.

[Sportman]

few GB.
Few day

Is "few GB" a static fixed dictionary or altered and/or increased with every input?

How can it be a "Few day" while there was only 11:22 hour between the random text test file post and your answer or did you decompress it later?

[Shelwien]

Its not a problem even if it can only compress valid english... Just type out the data as text, ie 0xFF = 255 = "two five five".
Even if enwik8 becomes 1G, it should be still compressible to 900k, so you'd still get the full prize.

Btw, hashing is not a solution for compression, not because it needs "huge database or super computer"
to restore input data from hash value, but because of collisions.
Even with assumed charset [\x20a-z] of 27 letters, you'd still start having collisions with 16 input symbols
and 9 bytes of output:

16 letters = 27^16 = 79766443076872509863361
9 bytes = 256^9 =     4722366482869645213696

Also, you can't sell software rights that easily - the trade has to be officially registered in some way,
usually you'd get a patent for your algorithm, then sell it.
Otherwise you can always claim that your software was stolen, once the buyer starts making money from it.

[Obama]

You so nice , lead the stranger to the point.So good you are here.

[Sportman]

Input:
1,000,000 bytes - 1 million random alphabet letters.txt

Output:
588,286 bytes - paq8px v178
588,001 bytes - cmix v17

-------------------------------------------------------

Input:
1,000,000 bytes - 1 million pure random data.txt

Output:
749,400 bytes - paq8px v178
748,956 bytes - cmix v17

[Shelwien]

1 million random alphabet letters.txt
charset = [a-z], size 26
1000000*Log[256.,26] = 587555

1 million pure random data.txt
charset = [\x0C\x1E07-9;=?ABD-FHIKMO-QTY\x5D\x5Ea-z\x7F\x83\x8D\x9E\xAF\xB0\xC6\xC7\xCE\xD3\xD5\xD8\ xDF\xE0\xE5\xE7\xEC-\xF0\xF3\xF6\xF8\xFA\xFC], size 79
1000000*Log[256.,79] = 787973

[Obama]

how do you know the charset ?

[Shelwien]

This script prints it. Its in perl.

[Obama]

To apply patent it need around RM15k .My algorithm can make unlimited compress data (I think , just tried 100k compress to 11 letters) , if apply patent my algorithm worth it or not ?

[xinix]

To apply patent it need around RM15k .My algorithm can make unlimited compress data (I think , just tried 100k compress to 11 letters) , if apply patent my algorithm worth it or not ?

We do not know whether to patent it.
Since you need more information about the algorithm.
___
By the way, how much can you compress book1?
___
Have you tried to decode the archive and check the MD5 files?

[Obama]

We do not know whether to patent it.
Since you need more information about the algorithm.
___
By the way, how much can you compress book1?
___
Have you tried to decode the archive and check the MD5 files?

compress for me no problem ,but decompress i need huge database or super computer.Thats the reason i want sell it.i cant afford it .

[xinix]

compress for me no problem ,but decompress i need huge database or super computer.Thats the reason i want sell it.i cant afford it .

How long will it take to unpack a 1MB file?

[Obama]

We do not know whether to patent it.
Since you need more information about the algorithm.
___
By the way, how much can you compress book1?
___
Have you tried to decode the archive and check the MD5 files?

Done,just few hours,the result is 847 bytes.

[xinix]

Done,just few hours,the result is 847 bytes.

Thank.
Interesting.
Do you have the opportunity to put this 847 byte file here?
Only compressed file.
I think we can squeeze it again.

[zyzzle]

It is obvious that this fellow is pulling our chains and pressing our buttons. Let's make a graceful exit from his nonsense.

1,000,000 random digits to 1028 bytes? Absolute rubbish, and no way even given 10^9 years of time...

[Obama]

It is obvious that this fellow is pulling our chains and pressing our buttons. Let's make a graceful exit from his nonsense.

1,000,000 random digits to 1028 bytes? Absolute rubbish, and no way even given 10^9 years of time...

Why rubbish please explain to me ?

[Shelwien]

It depends on how you define "random" really: https://encode.su/threads/3099-Compr...ll=1#post59940

[compgt]

> How much compress text worth ?

I had heard of your algorithm in the high places during the Cold War. But there are only 2 ^ (8,224) files addressed or compressed by your algorithm, not enough to cover all the files in your 2 ^ (8,000,000) source file space, of course.

But how much compression algorithm worth? I guess it must be larger than $120 million paid by Microsoft to Stac for the Doublespace program infringement. Maybe compression algorithm must be worth more than $210 million or $220 million, considering that DeepMind was bought by Google for only $400 million.

[svpv]

I skim through this thread for the third time and it strikes me again that this is a "Nigerian Prince" kind of thing. Bogus claims for gullible audience. But what if this could be true? Well, it reminds me that patent offices in some countries are prohibited by law to grant patents related to any kind of perpetual motion machines. The laws must be amended to prohibit the possibility of compression below the entropy.