Predictive compression algorithm

1. a File Z(converted to floating point number) length N bits is represent as Log2(Z) stored at N bits precision
2.the first bits of log2(Z) are equal to 2^N bits length since range is converged to:.
 2^N > number is here <2^(N+1)
3. since we can remove the first bits(e.g. a single byte) if N is known(and reduction is possible), the stored precision is lower than N.
4.if the process is applied recursively, it would reduce the file substantially(until N is too small to be useful)

>>2
Thats not like i'm developing a single one, i'm hacking algorithms every day. it doesn't really fit in infinite compression thread because its
fairly limited in range(N has to be large enough) and uses no range search or any "technique".
Its a straightforward conversion log2(Z)-remove first bits->store ->load->add first bits from (filesize+expandFactor).