A man claims to store multiple movies on 64kb

[unmei]

even in the strange case of a (dead) forest in vacuum, the tree falling probably produces sound in the trees it touches during that process. The sound will just not propagate to onjects not connected by some medium.

Another question, why do you even load that forum page to read ..how come you assume you can read it by entering the url and pressing return before you are actually "observing" it. Computer users assume a lot of physics to just work all the time by pure experince. And i for myself don't even know how all the processes are supposed to work, i just assume they do.

[LordRPI]

Originally posted by unmei
And i for myself don't even know how all the processes are supposed to work, i just assume they do.

They don't work. It's a conspiracy. They found a way to read all the minds of people and put what they think they should see on the computer monitor. Click here for more details. (http://sprott.physics.wisc.edu/pickover/esp.html)

[Soulhunter]

Originally posted by LordRPI

They don't work. It's a conspiracy. They found a way to read all the minds of people and put what they think they should see on the computer monitor. Click here for more details. (http://sprott.physics.wisc.edu/pickover/esp.html)

Uhm... :eek:

EDIT: Hehe, after some sleep I got it... ;)


Bye

[stephanV]

Originally posted by theReal
I think this philosophical question was asked in a time where there wasn't physical certainty about moving objects causing sound waves.

You are forgetting that we hardly understand this sort of physics at all. From my study i can see we already have trouble enough calculating the heat flow through a perfectly spherical object whilst air is flowing around it. All sorts of arbitrary constants come into play then (e.g. Reynolds). We certainly don't have the knowledge to predict what kind of sound an immensely complex object like a tree would produce when falling down.

Remember that from experience you could also conclude that the earth is flat.

I'm not saying that trees dont make sound if they fall when we aren't around; there's just no way to be sure. Science is at its best highly inaccurate.

[Soulhunter]

Crash-tests...

Feeding all info into a computer and you will get a perfect pre-construction of the incident !!!

Real crash-tests are only made afterwards to be 100% sure that the pc simulation was right...

Sould be possible to do the same with the tree as well, no ???


Bye

[stephanV]

Originally posted by Soulhunter
Crash-tests...
Feeding all info into a computer and you will get a perfect pre-construction of the incident !!!

You perhaps get an approximation of the incident. But can you be sure you are gathering all relevant info? No computer simulation has ever perfectly predicted the outcome of such complex events. Try to make a perfect re-creation of any currently existing tree in a computer: you will fail. Modeling can only go so far, all models have a limited range in which they can be used.

A computer is nothing more than a overpowered calculator. But what if there is relevant data to take into account that we have forgotten to calculate? Or even worse, what if there is relevant data that cannot be calculated with?

Another thing is: most models are not build by understanding the physical nature of things (those models that are, are usually highly inaccurate), but from experience. Of course, most of what we see as our understanding of the physical nature of things is just experience too.

If you create a model from experience you are only confirming what you already thought. If your model goes against your test results, you say the model is wrong, but if your model confirms them you say it is right. But almost no one ever considers that the test results themselves might be wrong.

[Soulhunter]

There are even more complex simulations (http://www.es.jamstec.go.jp/esc/eng/ESC/mission.html) than a "ordinary" tree-fall !!!

But sure, you would still need dozens of parameters to do the tree-fall simulation...


Bye

[stephanV]

:)

computers have difficulty prediciting the weather... what they are trying to do is hopeless ;)

[unmei]

Wow that test is pretty cool. Smart guy. But the comments on the explanation page are even better.. i wonder whether he picked only the dumbest or produced them himself :) ..and the second or so is plain out lying by claiming it were wrong once out of two times (as it is bound to work everytime until the user cheats).

[Zap_Bee]

i cannot believe that a hundred or so people tried this test and not even one found out the trick. i don't post it because i don't want to spoil this for other people, but i know there are enough guys who couldn't expect to show off their greatness. Strange.

Zap

[stephanV]

Originally posted by unmei
(as it is bound to work everytime until the user cheats).

How can you cheat this test? It's impossible... :)

[Didée]

Yeah, what a cheek. If they would sell that stone old card trick just as the lame trick it is, one could laugh about it. But selling it as an ESP experiment, clothed in a science-like appearance ... that's already impertinent. I could get angry about such things. Less because of the bumbledom of the site's creators, but more because of the possible damage done to the not-so-bright people that stumble about it and believe that crap. That's about comparable of telling all sorts of lies to a small child, just because it will believe them. Which is no good way of getting acknowledgment, and for sure its not funny.
There are really poor minded people out there.

[Didée]

>> [Soulhunter]
>> But sure, you would still need dozens of parameters to do the tree-fall simulation

You forgot typing the word "centillion" - some dozen centillions of parameters.

Any assumptions that the flow of things could be predicted, at least theoretically, if only the start conditions were known close enough, are naive. In one word: Mandelbrot. Or more verbose: Not only that many physical systems show natural chaotic behaviour in general and alone therefore can't be predicted. You can't even know the start conditions close enough, since this is simply impossible: Firstly, you would need to know the exact state of your object down to all of its smallest possible elements. But remember Heisenberg - you cannot evaluate the exact status of small particles. So you'd already fail here. Secondly, in order to make an exact prediction, you would need to know the exact state of EVERYTHING in the universe, since you cannot assume that interaction of things doesn't exist beyond a certain distance.
Of course one can do rough approximations with a much simpler set of data. But then you're only doing a rough approximation, not an exact prediction. But for the ones claiming that a physically based prediction would be theoretically possible, the above applies. (Apart from the fact that all discussion about prediction and/or predetermination is superfluid, when looking at it from the point of a wider spanning framework - but that's another story.)

No. That thinking of "gimme a big enough computer, and gimme enough data, then I'll compute you everything that will happen in future" belongs to childs in sandpits - read: scientific childs in scientific sandpits.


>> [stephanV]
>> If you create a model from experience you are only confirming what you already thought.
>> If your model goes against your test results, you say the model is wrong, but if your model
>> confirms them you say it is right.
>> But almost no one ever considers that the test results themselves might be wrong.

Ah, we start getting closer. Although the wording of "wrong test results" is not too lucky. If you made a test, and the test finished with a result, then there *is* a result, being the necessary product of the test setup, and can't be lied away. However, what might be wrong (and ever is) is the test setup. To create a correct test setup, everything belonging to the test system has to be considered. This includes man itself, and that's the point that always gets dismissed. (I'm not aiming at humans making errors during the experiment. I'm aiming at humans being part of the system.)
It's been so long now since the equivalence of energy and matter has been revealed (to stay on and view from the scientific angle), I wonder why it is so hard to draw the conclusion. If nothing else, at least this one simple conclusion: There is no matter. Not really. Not how we usually understand it. Matter is a special appearance of energy - thus there lastly is no point in differentiating them. There is energy, only energy, and nothing else but energy. There is no such (discrete) thing as matter. That's the point that really must be understood. Understood not only by cognitive thinking, but by heart, mind and soul. Only if one has really internalized this, a way leading out of the dead-end-road of our current sciences appears, and the possibility of getting a faint idea about the spanning interrelations arises. Which, in fact, make the traditional sciences appear a little ... obsolete.

Hell, the scientists can't even give a senseful explanation what "energy" is, after all. All our technics are up and running, but in last instance, no one can explain why it is so. Even our "hard sciences" are grounded on way too much things that "are just like they are, don't ask" - which demands for nothing more than "believing", making the sciences sort of a religion themselves. This isn't realized by most, but it is like that indeed.

[duartix]

If there is still anyone interested in the topic of the original post, as Phoenyx pointed out:
Could I point out one thing? Think of a program that can create an infinitely complex picture using a single equation. Fractal compressors are even now being worked on. There was a mathemetician (sp?) (and I don't remember his name) that said that an image of photographic quality could be encoded using about 10k of equations and 500k of base data. As I remember (could be wrong) that was supposed to be a 1200dpi equivalent 16 million color picture.

Can you extrapolate this to 64K of equations (not keys, there's a big diference!) and 370MB of base data? I believe this has some grounds even though the encoding/decoding process could take ages...

But if you allow me to follow the key path, I could do it myself if I wasn't a honest person. For 10 movies, I could have used 37MB of data per movie and just 10 bytes of keys. Give me 37MB of Flash and tell me how long do you want the movie to be...

Without knowing exactly what kind of content Jan Sloot presented, there is no way to guess whcih technical approach was taken.

We don't even know if the box, which was opened later without is consent, was feeding his laptop with just 64K of data or 20GB of video that only his portable could understand, and Philips decided to hide the fact that they were taken...

I guess we'll never know. :(

[Phoenyx]

Originally posted by duartix
If there is still anyone interested in the topic of the original post, as Phoenyx pointed out:


Can you extrapolate this to 64K of equations (not keys, there's a big diference!) and 370MB of base data? I believe this has some grounds even though the encoding/decoding process could take ages...

But if you allow me to follow the key path, I could do it myself if I wasn't a honest person. For 10 movies, I could have used 37MB of data per movie and just 10 bytes of keys. Give me 37MB of Flash and tell me how long do you want the movie to be...

Without knowing exactly what kind of content Jan Sloot presented, there is no way to guess whcih technical approach was taken.

We don't even know if the box, which was opened later without is consent, was feeding his laptop with just 64K of data or 20GB of video that only his portable could understand, and Philips decided to hide the fact that they were taken...

I guess we'll never know. :(
^^^^^^^^^^^^^^^^^^^^^^^^^

And that is the key to this thread. :D

(At least until someone invents a 3D wide area past time viewer, anyway:p )

[unmei]

How can you cheat this test? It's impossible...
By remembering a card that is not in the presented set ..that is actually how i realised the trick as this card then showed up in the "set with your card removed" :rolleyes:

[Soulhunter]

Originally posted by Didée

>> [Soulhunter]
>> But sure, you would still need dozens of parameters to do the tree-fall simulation

You forgot typing the word "centillion" - some dozen centillions of parameters.

Any assumptions....


Common, this was only about -> "Does the tree-fall produce a sound..." !!!

Maybe with a "simple" simulation you can't tell if it produces a sound of 74.964565 dB...

But Im sure you could calculate that it produces a sound somewhere between 70-80 dB !!!


Yes, its a rough approximation as you said...

But as more parameters you know the more accurate the simulation is !!!

For a 100% accurate simulation you would need to know... uhm, everything... :rolleyes:

But even with some dozen parameters you would know that the tree-fall produces a sound !!!


EDIT:

Some "tree-fall-question" links... 1 (http://www.physlink.com/Community/Forums/viewmessages.cfm?Forum=7&Topic=1326), 2 (http://www.eadon.com/phil/fallingtreetb.php), 3 (http://www.spectacle.org/396/scifi/tree.html), 4 (http://www.iidb.org/vbb/archive/index.php/t-84854) !!!

Interesting...

Visualizing (measuring) the sound from far away without actually hearing it would solve the problem !!!

But to visualize the sound, there must be some sort of microphone that "hears" it... :rolleyes:

Wait, the question was "someone" and not "something" !!!


Bye

[SeeMoreDigital]

There's nothing like a good conspiracy theory!

The bloke probably never even existed and was created by some nerd somewhere to see how far the story would run ;)


Cheers

[Klendathu]

the length of this discussion undermines the reputiation of this site.

please stay tuned for our upcoming topic:
my hamster was an agent for the federal governmnent and spied on me.

[Mug Funky]

uh? it was just starting to get interesting... i love arguments about determinism and stuff. it's a bridge between philosophy and science.

[scharfis_brain]

multiple movies into 64k?

you are not even able to store multiple soryboards/scripts (Drehbuecher) in plain ASCII format into 64kb, even if they are heavily compressed (RAR, ZIP or whatever)!

how should fit video & audio in 64k then?

[DK64_MASTER]

I love reading the comments about people who think the card trick is real ESP. It's a well known illusion, and I bet all those testemonials were fake.

[Soulhunter]

Originally posted by Klendathu

please stay tuned for our upcoming topic:
my hamster was an agent for the federal government and spied on me.

I have a better one...

All TV's are working backwards -> The CRT is giant camera-lens !!!

This way the broadcasting stations are able to see what you do while watching TV... :D


Bye

[Mug Funky]

Soulhunter: that's already been done

http://www.online-literature.com/orwell/1984/

Winston Smith! excercise harder! (says the TV)

[Soulhunter]

Originally posted by Mug Funky
Soulhunter: that's already been done

http://www.online-literature.com/orwell/1984/

Winston Smith! excercise harder! (says the TV)


Ok, whats with this one...

All politicians are actors !!!

A dictator controls the whole world... :eek:


Bye

[SeeMoreDigital]

Originally posted by Soulhunter
A dictator controls the whole world... :eek: Don't you know... a modern democracy is a dictatorship ;)


Cheers

[LordRPI]

Originally posted by scharfis_brain
multiple movies into 64k?

you are not even able to store multiple soryboards/scripts (Drehbuecher) in plain ASCII format into 64kb, even if they are heavily compressed (RAR, ZIP or whatever)!

how should fit video & audio in 64k then?

Ok, new idea here. You don't store anything in your movie file. That file just unlocks part of your decoder which already has your movie information in it :)

[Joe Fenton]

I think some people here realize the "trick" of fitting 10 movies into 64K. You work the problem backwards.

Instead of a 64K video player in the computer and ten 37M movies on the media, you just turn it around. Ten 37M movies in a database in the computer and the 64K player program on the media. 64K - 1 byte representing the movie you want to play for the player... I could do that easy. Take a single codec and strip all non-essentials out of it. No input other than selection of 1 of x movies, simple output to display device. It's hardcoded for everything, including looking for the movie data in the 370M database in the PC. Depending on the complexity of the codec I choose and the size and length of the movie, I can determine how good the movies will look in 37M. Easy, easy, easy. Philips probably whacked the guy when some sniggering engineer told the PHBs how they were taken by a con-artist.

[callmeace]

:cool:

Think of an MD5 sum - it's essentially a 'hash' number that is generated by the contents of the file.

Think logically about that.

If you could have a system where a unique number is generated by the contents of a file (for arguments sake we could put a limit of 4 TB on a file size), then any file could be represented uniquely in maybe less than 4KB - size of a text file to hold the hash number and maybe small amount of other data like original filename and maybe some prompts too (sort of like cue file that you get with BIN files)
Now all you would need is to put that together with a well optimised code generator of some kind that would reassemble the original file from the unique hash. Code Generator could maybe be less than 60KB

The concept is there :cool: I believe it could be perfectly feasible as I've explained above.

[Mug Funky]

hashes don't work like that. otherwise we'd be seeing ZIP and RAR using these techniques, and nobody would ever need more than a 56k connection.

[Cyberman]

Originally posted by scharfis_brain
multiple movies into 64k?

you are not even able to store multiple soryboards/scripts (Drehbuecher) in plain ASCII format into 64kb, even if they are heavily compressed (RAR, ZIP or whatever)!

how should fit video & audio in 64k then?

The text isn´t stored at maximum efficiency(sp?). The compression had to check for whole words, storing THEM instead of the characters.

For example:

1 = codec
2 = computer
3 = video
4 = encoding
5 = using
6 = my

"I am 5 a 1 for 4 of 6 3 on 6 2."

Quite short, ain´t it? Reversed, it´ll become:

"I am using a codec for encoding of my video on my computer."

The downside is that everything has to be known in advance.

[theReal]

The text isn´t stored at maximum efficiency(sp?). The compression had to check for whole words, storing THEM instead of the characters.

For example:

1 = codec
2 = computer
3 = video
4 = encoding
5 = using
6 = my

"I am 5 a 1 for 4 of 6 3 on 6 2."
That's about how Zip and Rar work (only they'd probably first use a number for words like and "am" or "my" because they are the most frequent in a text and so profit most from abbreviation). Every repetitve pattern can be stored much shorter than its actual length.

Now check what scharfis_brain said:
(not) even if they are heavily compressed (RAR, ZIP or whatever)!He wasn't talking about plain ASCII...

[DigitAl56K]

Originally posted by Cyberman
The text isn´t stored at maximum efficiency(sp?). The compression had to check for whole words, storing THEM instead of the characters.

For example:

1 = codec
2 = computer
3 = video
4 = encoding
5 = using
6 = my

"I am 5 a 1 for 4 of 6 3 on 6 2."

Quite short, ain´t it? Reversed, it´ll become:

"I am using a codec for encoding of my video on my computer."

The downside is that everything has to be known in advance.

Good explanation :) If you go back and look at the math I did originally, remembering it was equivelent to looking up just one word per frame, using the minimum possible address size. I.e. in your example, you used 7 indexed lookups to compress part of a sentence "5146362" - modify that to one indexed lookup to represent an entire frame, and you'll get what I got, and still be nowhere near 64k, even for a single movie!

[callmeace]

Originally posted by Mug Funky
hashes don't work like that. otherwise we'd be seeing ZIP and RAR using these techniques, and nobody would ever need more than a 56k connection.

Well, in fact they could work like that if someone took the time (20 years ;) )

It is possible:- given that you put a maximum size on an input file, to have a string of characters together with a sum 'formula' which can recreate any file. It works by reversing, like so:

You have your input file. You have your formula which generates a unique string of characters derived by the contents of the file. Assume this string of characters can be held in 4kb. The formula (which is under 60KB) can recreate the file because the character string is unique and was generated by the contents of the file - there is only one true answer for the 'sum' to add up to; just like any mathematical sum if you like.

Now interestingly, like with any compressor or decompressor it works more efficiently if there are hints given as to what the data will be in advance. So your <64KB of data in addition to the character string and the formula probably would contain a few hints about the file that needs to be recreated, which would surely speed this process up massively.

It's amazing what can be done by the dedicated :)

[Manao]

callmeace : you claim it could be possible to store any file whose size is under M Bytes on N bytes, with N < M. That's impossible. You have 2^M possible different files as input, and only 2^N possible key strings. It's bound to have key strings which reference at least two different input file.

[Mug Funky]

mm. listen to manao...

hashes are useless without the file they are hashed from. all they can do is check whether they were derived from a given set of data. they cannot reproduce that data, any more than you can recover a dead hard disk using only the parity bits.

we can't recover CD-R data if all we have is the ECC codes in the sector and no actual data, but we can often recover the sector if there's enough intact data to make the error-correction work.

[Phanton_13]

It's posible to do this tipe of good compresion but unlike the actual compresors than uses redundancy in files, they need not only to obtains a data string, they need to optain a complex matematical expresion an then store both. But this is not a trivial problem, this is very computational expensive, complex and had some limitation but it can deliver greater performance than actual tecniques, for a 10 megabyte files you need days in the nowdays most power computer on the heart to obtain a good compresion ratio, but you can decompres it it seconds.


Sorry for may bad english.

[Manao]

No, it's simply not possible. Consider compression as a function which associates to a number x another number y=f(x).

If x can range from 0 to M, and y from 0 to N, with N < M, you are bound to have x1 != x2 with f(x1) = f(x2), which means f is non invertible, which means you can't use f for compression.

[virus]

@Manao

maybe it's worth reminding that your "proof" only stands for f(x) defined in a discrete domain and with discrete values for every x, taken from a finite set of values (example: the numbers you can represent with N bits).

(obviously it won't work for continuous functions... take y=0.1x as an example ;))

cheers :)
virus

[Didée]

Originally posted by magicclue

quote:
----------------------------------------------------------------------
Originally posted by Didée
One just cannot port those quantum behaviours into the "real world". That would be the same as saying "an electron can appear either as a wave or as a particle. Therefore, my pencil possibly can appear as a wave as well." - It makes no sense to look at it this way.
----------------------------------------------------------------------

Remember school? Remember physics?
Remember LIGHT?
Light is being transmitted as wave as well as particle.

Now what?


Yes? What "what"?

Draw a conclusion, state a hypothesis, ask a question. But don't sound out things wildly.

(I'm not going to develop a new sight of the world in this thread, never fear. But I'm sure enough of mine.)


edit: added quote, since magicclue deleted his post.

[Atamido]

Originally posted by Manao
callmeace : you claim it could be possible to store any file whose size is under M Bytes on N bytes, with N < M. That's impossible. This has been brought up before, but there is a standing reward of several thousand dollars to anyone that can break this. Basically, you ask for a piece of random data that is X size. Then you must make a program and a "compressed" file that can be used to reconstruct the original file and whose total size together is less than the size of the original file.

Of course people want this to be wrong, but there is no reason to believe that it is. If you can prove it, go collect your money.

[Manao]

Pamel : it's not the same. callmeace was speaking of on program that could compress anything to a lesser size.

Anyway, in your case, it's still not possible. You're bound to find a file that won't be compressed under it's own size. If it was not so, you'd have found an invertible function defined on a finite space, which goes into another finite space whose size ( cardinal ? ) is stricly inferior to the first one. And that doesn't exist.

[callmeace]

:)
Manao & Mug Funky

I was not talking about traditional means of compression - as in 'pointers' and repetitive sequences of data etc, nor when I used the term 'Hash' was I meaning it in as an MD5 type hash (or at least not to be used for a purpose like that) which is only useful to check existing data that you already have.

No! It is a lot more advanced than that, it would need a lot of work and dedication to create such a formula as I suggest.

Let me try and explain it again with a bit more detail :)

The formula would have to be very complex.
The formula would probably have to work at the binary code level (as in analysing the data).
There would have to be given limits on the size of the input file so that logically there would be a limit on the variations of data one could ever have - and hence it is actually possible to make the formula we wnat in the first place ;)
The 'hash' or code which will be generated will so be unique for every variation of code there could be.
The 'hash' will be a unique sequence of letters and numbers generated from the data of the file, the formula will recreate the original file as there can only be one correct answer to the 'result' - that's how the formula works as in 'input data' - unique code generated - reverse using code to recreate what must be the original file as the only answer to the 'sum' (the code) using the formla.

[virus]

Come on guys, 50+ years of development in the realm of information theory are still not enough to prove that you cannot compress arbitrary sequences of random data? There's no algorithm that can do that, and this has been proven a long time ago.

Anyway, I want to take the chance of launching a contest myself. I'm ready to award 1.000.000.000 euro to the one who will build a spacecraft that can travel faster than light ;)

Who knows, maybe with a very complex formula... :)

[Soulhunter]

Originally posted by virus

I'm ready to award 1.000.000.000 euro to the one who will build a spacecraft that can travel faster than light ;)

Is bending the space allowed ???

Or is this cheating... :D


Bye

[callmeace]

Originally posted by virus
Come on guys, 50+ years of development in the realm of information theory are still not enough to prove that you cannot compress arbitrary sequences of random data? There's no algorithm that can do that, and this has been proven a long time ago......

But this is a different approach to 'compression' :)

a unique string is generated from the data contents (to be used to gether with a formula)

The formula can recreate the original data because mathematically with the formula there 'is only one result that it must add up to'

Complex formula, yes :) but skill & dedication would pay off. But it seems the one who did something like this got killed

[fccHandler]

I agree with Manao and virus. It's just not possible to losslessly compress everything that exists. For example, already compressed data isn't able to be further compressed. (And if it is, it only means the original compressor wasn't 100% efficient.) I believe there must exist data that cannot be compressed by ANY method, thus all of the methods we have are compromises in a way. Note how they all seem to be optimized only for specific types of data streams.

ZIP is actually a suite of compression methods, but I know I've seen MPEG videos which can't be zipped to a smaller size. I've also seen Huffyuv produce larger files than uncompressed video (try capturing the static from an unused TV channel). ;)

[r6d2]

Interesting story. Reminded me of this other one:
Once the great mathematician David Hilbert was invited to give a talk on any subject he liked during the early days of air travel. His subject: The Proof of Fermat's Last Theorem.

Needless to say, his talk was eagerly anticipated. The day arrived, the talk was given, and it was brilliant -- but it had nothing at all to do with Fermat's Last Theorem.

After the talk, someone asked Hilbert why he had picked a title that had nothing to do with the talk. His answer: "Oh, that title was just in case the plane crashed." :)

[stegre]

The 'hash' or code which will be generated will so be unique for every variation of code there could be.
If you have a unique "hash" for every possible variation of code, well, that's not really a "hash", but more to the point, there would be no associated compression. It would take as many bits to represent your hash as it would to represent your original item. If there are exactly a million variations of your item, and you uniquely assign each to one of a million codes, what have you gained? You've just "numbered" them.

With the standard meaning of "hash", you have a "many to one" relationship of items to codes. Once you have fewer codes than items, then sure, that's compression, and using the hash "backwards" (to go from a code to its item) is the decompression. But since the codes can map to more than one item, the scheme is either lossy or requires you to put limits on the data.

Say my hash for the 256 possible 8-bit numbers was simply "use the 7 most significant bits". Nice 12.5% compression - but when I decompress, each maps to 2 numbers. If "always choose the even number" - that's lossy compression (result is always within "1" but may not be exact). If my needs were such that odd numbers would never come up (socks in my drawer [in a perfect world!]) then the compression could be considered lossless - but only because of the limitations placed on the original data.

[callmeace]

aargh! :) Okay, I realise I still did not really explain the 'concept' or how it works very well, and my use of the term 'hash' was poor - sorry.

Here is a further attempt to explain how it works (would work)

(1)This way of compression would take very long, and a very dedicated person for sure! It does not use your 'normal' common ways of compression! It does not work by finding repetition in the data or anything like that! This is a different concept of approach to compression, probably because it would take so long to develop the formula. This is the idea I would like you to realise

(2)It is not a way of compression where the method can be compared as being like zip, ppmd, or anything similar. The hash I mention is not like some of you thought I meant (my misuse of the term I guess)

(3)Okay, here is how it can be done. Please bear with the concept.

You have a formula which calculates a result you might say, from an input.(a)If input is some data like a big file, then the result will be a character string of say letters and numbers (what I called 'hash' before :D) (b)or reversing, if the input is character string then the result will be original data file.

Just like maths, the result is computed and there is only one answer to the sum (that is the important concept). Essentially the complex formula works like so 'if the input = 'this' then the result must = 'that'.

Yes, the formula is complicated, but feasible. It does not work like common used methods which generate an compressed archive. It generates a character string derived from the contents of the file and unique to it, and by reversing can recreate the file from the string by determining what it must have been derived from.

(4)The maths could be worked out and indeed I believe has - like I said, dedication and commitment of time and energy, but in the long term a far superior way to what we have currently!

Please I hope you get the concept of how it works. I guess a starting point is consider how an MD5 is generated but realise it is beyond that.

:edited MD5 (had typed MDG )