HN Explorer

Brilliant/retarded idea for startup. Brilliant/retarded developers needed.

5 points by cthulha ↗ HN
I've already submitted a YCombinator app with friends (under the username ltoop, which has a secure password I can't remember) but I had a really stupid/awesome idea after my coffee today.

Pi is infinite and non-repeating, so there must be a sequence in there somewhere which corresponds to any binary sequence you care to name: text, audio, video, application binaries (Source code!)

So, we could start up a RapidShare/BitTorrent competitor (I have registered ThePiBay.org in preparation!) where is a database of offsets and lengths within Pi for a given file. We don't store anything illegal, have really low bandwidth consumption, and can access an infinite amount of material.

Users upload a file to us. We search for it in Pi, and give them a link to a page which has the details. They'll need some kind of application to use these details to generate the file, and since this could be computationally intense, ideally it could distribute the load across a range of peers and share chunks of the resulting binary sequences. That is, a Pi2Pi application.

Heck, why would anyone waste time downloading 0-day warez releases when you can search for software yet to be designed/released? With appropriate data-mining, people will use us for -1-day warez, like MSOffice 2020 or Mac OS X 10.9. How about i-day warez, imaginary programs that people have never have written and will never write, that are waiting in the digits of pi for us to find them! The winner of the NetFlix competition is in there somewhere!

I'm almost out of puns and stupid jokes (finally), so lets talk about business plans/monetisation.

Perhaps we have free and paid users, where free users can only search within the first few billion digits of pi. Or maybe their algorithm approximates pi with 22/7 :)

Maybe we search for versions of the sequence that incorporate ads for our sponsors (these versions are already in there somewhere!) and serve them to non-paying users, or we give them low res versions and spelling mistakes or captions in Polish. Pi is an infinite resource, just waiting for us to mine it!

Seriously though, I wish I thought of this last week so I had time to put up a fake page for April Fool's day. Anyone wanna help design a fake website for it with me :)

13 comments

[ 3.1 ms ] story [ 42.4 ms ] thread
rms
The sad thing is that this is probably still illegal.
cthulha
If RapidShare, TPB and LimeWire can operate, there must be a way to make it work!

The problem is takedown notices -- are we expected to redact parts of a mathematical constant in some jurisdictions :)

rms
>are we expected to redact parts of a mathematical constant in some jurisdictions :)

Do you remember the fiasco on Digg when the Blu Ray decrypt number was released? Numbers can be illegal. In that case, it didn't go anywhere, but with ThePiBay I think Big Media would try and push for the redaction of a mathematical constant.

cthulha
"In other news today, the value of pi was rounded up to 4 in new copyright legislation sponsored by the RIAA. This will also prevent terrorist groups communicating via mathematical functions."
mixmax
Try searching google for "if you give a million monkeys a million typewriters" and you might see the problem.

You are joking right?

cthulha
I've also thought of a name for the client software: if our users are Pi-ers (short for Pi-rates?) then it's Pier to Pier and hence should be called 'Ferry'.

Anyone have a worse pun to add?

delano
This is a very pious debate.
jgrahamc
> Pi is infinite and non-repeating, so there must be a

> sequence in there somewhere which corresponds to any binary

> sequence you care to name

Sort of wrong. The question you need to ask is "Is pi a normal number?". We know that pi is irrational (i.e. can't be expressed as a fraction) and that is is transcendental (isn't the root of a polynomial with rational multipliers) but we don't know if all sequence of digits occur in the expansion of pi. In fact we don't even know if 0 through 9 with the same frequency in the infinite expansion.

That last part is important because it could be that at some point in the expansion one of the digits doesn't occur as often which will make finding your blocks impossible.

Happily, we do have digit extraction algorithms that would allow you to jump to any point within the expansion without having to start from the 3, but what you are proposing would require a very large amount of time to find the appropriate offsets.

Also, if you try the pi searcher (http://www.angio.net/pi/bigpi.cgi) you'll see how hard it is find even short sequences (e.g. I can't find any of my phone numbers in the first 20 million digits)

What might be more fun to try is for any given sequence of numbers (a file) create an appropriate formula that when repeated (like a random number generator) outputs the file you want. Then you just publish the formula and initial values.

yan
This is akin to an idea I had in college that if we take any multimedia or any data that is in the format of header+fixed size of data, we can fourier transform it and the frequency at which headers occur will spike as they repeat with the same frequency, thus we can easily identify data that looks like multimedia and even have an idea about the contents! Man did the stupidity of that idea shine through when I actually read up on how things /really/ worked.
andr
Even if it is possible given the properties of Pi, information theory tells you that the size of the offset will be comparable to the size of the actual data. And it will be slow as molasses.
mechanical_fish
Your first hire will have to be a veteran Microsoft sales rep.

"In order to further enhance the exciting new features of PIrated MSOffice 2020, we've had to push the shipment date back by 2e15 age-of-the-universe units. But, rest assured, it will be worth the wait."