i wrote it a few years ago for a site that never happened (mindclash.com); having seen the slashdot story, i thought this might be an appropriate place to post it.
he wouldn't void the entry if it genuinely wasn't hiding data in the filesystem, ie. the decompressor would accept the input file on stdin. that wasn't the case in patrick's attempt.
i guess you just don't get it. some things are highly unlikely, and some things are quite literally impossible. read section 9 of the faq; maybe that'll clarify it for you.
it'd work for numbers that were close to a prime. unfortunately, there are far more numbers that are a long way away from their nearest prime - so storing the prime index and the difference would eat more space than storing the original number. your recursive tactics don't solve that problem.
it's not as ambiguous as you seem to think it is. can you represent any integer quantity between 0 and 10 using only an integer quantity between 0 and 9? no, you can't; there are more things to represent than you have things to represent them with. i'd say that's pretty inviolate.
good point; at the level that craig was discussing, though, something as desperately simple as ensuring that you had no repeated or related sequences (given a suitably large random file) would probably be a first step towards eliminating the most obvious candidates.
probably statistically true in the general case of "compress a truly random file", but not so accurate in this challenge where the rules are "compress this specific file which has been carefully generated to resist compression".
of course, there's always the chance that your algorithm would fortuitously perform wonders on any given file, but i think the odds are stacked against you.
it's more of a thought experiment than a scam - the idea being that you can't win the $5k, however bad you want it. and it's a win-win situation for goldman, since he either gets a bunch of $100 checks from a bunch of loonies who think they can compress his random data, or he ends up paying out the not-impossibly-large sum of $5k in exchange for playing a part in what would be one of the greatest mathematical revelations in recent memory, which is probably more than good enough value as far as he's concerned.
in this case it wasn't entirely random, since, although the file was sourced from random.org, goldman got the chance to review it before delivery to craig, which (presumably) gave him the chance to check that he didn't fortuitously generate three meg of pi or the decss source - or any of the number of less obvious cases which may have inadvertently yielded the very slight compressability required to win the challenge.
there's no scamming involved. goldman required the (suitably small) entry fee to deter time-wasters; craig wasn't after the money in the first place. they were both just out to prove a point (goldman, that truly random data can't be compressed in the general case; craig, that goldman's challenge was amusingly open to abuse), not to make any money.
interesting paranoia, but in today's climate it turns out that being answerable to government is a much lesser evil than being answerable to advertisers. the BBC news department alone is so fantastically ahead of any news reporting in the US, it blows you away. you'll never see "disclaimer: the company we are reporting on is the parent company of this web site" notices on stories, and the government simply aren't able to suppress political stories and reporting, at least not in any legal or visible way. the internal politics of the globalmegamediacorps are impenetrable, but governments must be as open and transparent as possible. admittedly that's a naive assessment of government in general, but governments must have finite covering-stuff-up-without-a-trace resources, and i'd bet that they're expending them on suppressing and controlling things far more worrying than what their funded media are producing. ----
Re:Channel 4 is not government owned!
on
The Pledge
·
· Score: 1
correct. in fact, just last week hague was talking about plans to sell off channel 4 to raise some dough, so that the convervatives can CUT TAXES like mr burns in that episode of the simpsons. ----
Slashdot Mirror
i wrote it a few years ago for a site that never happened (mindclash.com); having seen the slashdot story, i thought this might be an appropriate place to post it.
he wouldn't void the entry if it genuinely wasn't hiding data in the filesystem, ie. the decompressor would accept the input file on stdin. that wasn't the case in patrick's attempt.
it'd work for *some* numbers (14159265 compresses wonderfully!) but not for others, which is true of every other compression scheme ever.
i guess you just don't get it. some things are highly unlikely, and some things are quite literally impossible. read section 9 of the faq; maybe that'll clarify it for you.
look at the date of the article he linked to.
ultimately it's just a bit of fun. nobody got hurt, nobody got scammed, and it entertained slashdot readers for a minute. where's the harm in that?
it'd work for numbers that were close to a prime. unfortunately, there are far more numbers that are a long way away from their nearest prime - so storing the prime index and the difference would eat more space than storing the original number. your recursive tactics don't solve that problem.
it does not work period. storing the factors of a number takes more space than storing the original number.
it's not as ambiguous as you seem to think it is. can you represent any integer quantity between 0 and 10 using only an integer quantity between 0 and 9? no, you can't; there are more things to represent than you have things to represent them with. i'd say that's pretty inviolate.
you've got them over a barrel! go claim that 5k!
this is a troll, right?
good point; at the level that craig was discussing, though, something as desperately simple as ensuring that you had no repeated or related sequences (given a suitably large random file) would probably be a first step towards eliminating the most obvious candidates.
clare-ents was talking about one compressor that would win 2% of the time
- gzip your data
- represent it as a large integer
- find the next prime number that can be represented by appending garbage bytes to your integer
so, just regular prime discovery afaik.is this a troll? you can represent a gzip of any data in this way.
of course, there's always the chance that your algorithm would fortuitously perform wonders on any given file, but i think the odds are stacked against you.
it's more of a thought experiment than a scam - the idea being that you can't win the $5k, however bad you want it. and it's a win-win situation for goldman, since he either gets a bunch of $100 checks from a bunch of loonies who think they can compress his random data, or he ends up paying out the not-impossibly-large sum of $5k in exchange for playing a part in what would be one of the greatest mathematical revelations in recent memory, which is probably more than good enough value as far as he's concerned.
in this case it wasn't entirely random, since, although the file was sourced from random.org, goldman got the chance to review it before delivery to craig, which (presumably) gave him the chance to check that he didn't fortuitously generate three meg of pi or the decss source - or any of the number of less obvious cases which may have inadvertently yielded the very slight compressability required to win the challenge.
there's no scamming involved. goldman required the (suitably small) entry fee to deter time-wasters; craig wasn't after the money in the first place. they were both just out to prove a point (goldman, that truly random data can't be compressed in the general case; craig, that goldman's challenge was amusingly open to abuse), not to make any money.
interesting paranoia, but in today's climate it turns out that being answerable to government is a much lesser evil than being answerable to advertisers. the BBC news department alone is so fantastically ahead of any news reporting in the US, it blows you away. you'll never see "disclaimer: the company we are reporting on is the parent company of this web site" notices on stories, and the government simply aren't able to suppress political stories and reporting, at least not in any legal or visible way. the internal politics of the globalmegamediacorps are impenetrable, but governments must be as open and transparent as possible. admittedly that's a naive assessment of government in general, but governments must have finite covering-stuff-up-without-a-trace resources, and i'd bet that they're expending them on suppressing and controlling things far more worrying than what their funded media are producing.
----
correct. in fact, just last week hague was talking about plans to sell off channel 4 to raise some dough, so that the convervatives can CUT TAXES like mr burns in that episode of the simpsons.
----
I'M FEELING LUCKY!
----
heh. clearly you have no hidden agenda here. what delightfully unbiased commentary!
----
surely nobody is that desperate for a job.
----
man, you're so right... life has been much better for me since i found slashdot 2 too.
----