Can anyone see any sort of relationship between these numbers?

[peterb]

So the first set of numbers is a seed for the code that generates the second set?

That would imply that there is no casual relationship between the two, especially if it is being used as a security application..

It could be some sort of hash function or a seudo-random number generator but I don't think you will find any obvious pattern.

You could generate some form of rainbow table by writing code that generates the input sequence of numbers starting at 0000-0001 and ending at 9999-9999 (assuming that is a valid range - if they are time related that may not be the case) and analysing the output generated by the code (it is the source code that has been lost?) but I don't think that would help much in deducing the algorithm.

[kalniel]

So you have an encoding function translating the first set into the second.. you're going to need a much bigger sample and some number crunching to figure out the function imho.

[jim]

I agree with Kalniel.

The program will surely have been designed to protect against this kind of thing? Or at least it won't make the numbers easy to guess. It could be doing any combination of a fairly wide array of functions, since a chain of 50 operations would only take milliseconds on a modern PC to run.

If it's dividing, throwing away remainders, picking a digit within the result, multiplying that by another digit, and then dividing by the square root of a different digit... no-one's got any hope.

Unless you know that it was a fairly simple function in the original source code, I'd give up

[TheAnimus]

You could generate some form of rainbow table by writing code that generates the input sequence of numbers starting at 0000-0001 and ending at 9999-9999 (assuming that is a valid range - if they are time related that may not be the case) and analysing the output generated by the code (it is the source code that has been lost?) but I don't think that would help much in deducing the algorithm.

And the winner is!

Look for the convergence where you have the same output from different inputs. Normally these types of hashing functions do collide. But I don't think you have that code to be able to run it any more, so your a bit stuffed.

Or you could send me the compiled binary that validates the code, I've a long train journey on Friday assuming I don't fly down, and its been a while....

[mikerr]

Hexus = cracking forum now?

You've effectively said "hi, write me a keygen pls, kthxbai"

How about a different angle - is the database itself unencrypted?

[TheAnimus]

err he claims its an in house bit of VB, some places to curtail external use of their binaries do this sort of thing.

I've seen a lot of code that simply throws if the PC isn't a member of a certain domain (obviously compiled as a string const!)

[mikerr]

Yes, but the point needed to be made!

Hacking the binary may be a better idea, if contacting the author ( if even known) is out of the question.

[Steve]

Train a multi-layer artificial neural network on the keys you have, then (hopefully) it'll "Just work" for any input you give it

[kalniel]

Train a multi-layer artificial neural network on the keys you have, then (hopefully) it'll "Just work" for any input you give it

And you won't be any closer to the underlying algorithm They've got a black box already

[Terbinator]

http://www.youtube.com/watch?v=fldj5...eature=related