Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Well, here is how it works (takes off thinking cap):

You don't go all the way in one go, the algorithm just does 1% compression each time so it leaves something for the next rounds. You then keep on doing that until you reach the desired size. It's not super fast but first let's get it to work. I already have a patent!

There is just this one minor technical problem that I still need to overcome, but once that's done the sky is the limit.



"one minor technical problem that I still need to overcome"

Presumably that would be decompression? ;-)


No, for some reason after 30 rounds or so of running it depending on the data that I feed it it will stop compressing. It's lossless after all, and after that the filesize slowly increases. Must be a bug somewhere. It decompresses just fine. How hard can it be? /s.


I had rather hoped you could repeat the process indefinitely = maybe to get down to one byte or even one bit. Only requiring one bit of storage would help a lot in lots of situations - not to mention the potential for increasing network throughput!


I implemented it once as well, all the way down to single but storage.

Turned out to be a huge security nightmare though, because it kept decompressing into other people’s files!

Another problem was telling the compressed bits apart.


Why did you stop at 1 bit?


Otherwise investors might get suspicious that something was up.

On another note, I have this protocol that can transfer any file in just 64 bits. It needs a rather large dictionary on both sides but other than that it 'just works'. Interested?


Nah, I got another guy who can do it in 56 bits.


I know, but first let's fix this one bug. With a potential of approaching infinite dollars any kind of valuation for this algorithm would be fair so how about it?

One of the very first tech DDs I ever did was a totally rigged demo. I don't think I ever wrote about it. Those people must have really hated me for puncturing their balloon. Some other investors took the bait and lost tons of money.


Why even bother with this when you store infinite data in pi[0]?

[0] https://github.com/philipl/pifs


You can really do that, sort of, for example:

https://github.com/libnoon/compress

The problem, of course, is that you have to decompress the file a large number of times. (The length of that number looks suspiciously similar to the length of the original file.)


Hehe, that one is actually really funny and neat.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: