99% compression kid

Does anyone remember the kid from around 5 years ago that won some contest and claimed he could compress any file to 99% of its original size? What ever came of that? He was supposed to be working on a startup company.

Other urls found in this thread:

cyborg.co/
web.archive.org/web/20131125185235/http://www.cyborg.co/
twitter.com/CyborgUSA
twitter.com/SFWRedditGifs

Never heard of that one.

Are you sure you're not confusing real life with the TV series called "Silicon Valley", OP? I know, sometimes I can't tell what's real life either.

Obviously not completely true. If you can compress any file by even just 1%, you can compress an output file from your program too. Repeat enough times and you can get any file down to just 1 byte. A byte has 256 possible values, meaning there are only 256 possible unique files fully compressed by this method.

Cyborg Industries?

cyborg.co/

That's the one!

>If you can compress any file by even just 1%, you can compress an output file from your program too. Repeat enough times and you can get any file down to just 1 byte. A byte has 256 possible values, meaning there are only 256 possible unique files fully compressed by this method.
But then you can decompress any of those 256 unique files to a large amount of files, and then you can decompress any of those files to an even larger amount of files and so on.

I have a slightly lossy compression algorithm that can make any file just one byte smaller. It just removes the last byte. Many files will still work correctly after being compressed. If you perform my compression repeatedly, you can achieve any compression ratio you want.

Can I have a prize?

If the same compressed file can decompress to more than one possible thing, or if the decompressed file isn't byte identical to the original, the compression is lossy.

Just say pidgeonhole principle next time.

Yeah I remember him. What he did is use a hash to reference something in a table, if I understood it correctly.

Of course it's bogus.

The guy's name is Nicolas Dupont and it doesn't seem like he has done anything. He has a "company" but I can't tell if it is real.

Sounds like a bunch of bullshit you just made up in hopes that it actually existed.

You're a fucking imbecile that should refrain from posting.

already found what I was talking about. I only wanted to find it again because of how fucking stupid it was.

99% compression is impossible unless the file is just the same thing repeated 100 times

Hell, someone else just posted this in another thread lol

Most of the time that's true though. For example, most anime has the same characters on screen for much of an episode.

his site hasn't changed at all since it was first posted here in 2013
maybe he's realized his grand ideas have already been considered or just plain silly
like his spectrocable idea, he even put together a patent for it, which details using every colour in RGB24 for fiberoptic multiplexing. it's the kind of idea any kid could come up with ("why not just use more colours?!")

Closest thing to 99% compression would be the procedural content generation done by demosceners. Their 64k competitions manage to pull some hefty amounts of content out such a small executable but it's no way a form of file compression.

Holy shit that hairline
Isn't he in his 20's or something by now?

>like his spectrocable idea, he even put together a patent for it, which details using every colour in RGB24 for fiberoptic multiplexing. it's the kind of idea any kid could come up with ("why not just use more colours?!")

already exists, it's called wavelength-division multiplexing

it's not compression in the typical sense (finding patterns in arbitrary data), but rather the opposite (writing patterns that generate data)
it's like they're directly writing compressed data, if that makes sense
what they do can't be applied to arbitrary datasets

i know, all he's doing is saying "use all the colours", without addressing how exactly he plans to do that. it's like claiming you'll make a 20GHz processor and it'll be awesome because 20GHz is fast and shit.

ps. how do you feel about him getting a patent for something that has been in practical use for ages already?
wavelength division multiplexing is nothing new, and i don't see how "using more colours" is good enough to warrant another patent

>his site hasn't changed at all since it was first posted here in 2013
uh ahcktually
web.archive.org/web/20131125185235/http://www.cyborg.co/

rip hairline

I always tar.gz.zip.zip.zip.zip.gz.zip.xz.zip.zip.zip.gz.7z.zip my files.

>he actually has booths at CES and shit
The fuck?
twitter.com/CyborgUSA