Hackers look for patterns in encrypted data in order to find clues on how to break the encryption method, or at least to find ways to trim the number of paths they must follow. Encryption effectiveness increases to the extent that a file's byte distribution becomes more homogeneous. There is no generally accepted definition or metric for high entropy or randomness in a computer file, but there are indicators that a technical person would look for that spell headaches for the hacker.
Tech note: Indicators of high entropy include convergence of the frequencies of the 256 byte values, reduction in the standard deviation of their frequencies, and multi-byte groupings that are shorter and less frequent. We wrote a ByteSeq utility to look for every sequence of four bytes that occurs at least three times in any input file of any type. This utility nicely shows up patterns among files that have been encrypted by any method whatsoever. Our Bytes utility displays byte distributions. Source code is available on request for both utilities.
From a non-technical viewpoint, what matters is that there is little or nothing in the way of clues for hackers in files and messages encrypted with our technology.
One method of pattern reduction is compression. Step one is to sample byte patterns within the content to be encrypted. Natural language text, repetitive files (the tags in RTF Rich Text Format) and block-based files (such as Microsoft Office 2003, loaded with nulls) all reduce size and pattern content considerably using publicly available compression software. There is a slight time cost to compression, but it offers side benefits such as reduction in transmission time and lower storage costs. Hence, after the test for compressibility, the process compresses content when the reduction of patterns and size will be significant.
Marpex Inc. is a research house. We watch and listen for problems that trouble our society, problems that might lend themselves to computer-based approach. Example: The frustration for graduate students of conducting serious research using a conventional search engine. Another example: The impending threat to our encryption methods posed by quantum computing (a three hour video of an October 24, 2017 Congressional hearing on quantum technology).
Our solutions are creative, multi-disciplinary, unusually efficient, and (we would like to think) effective. You be the judge!
Our patent application US 2019/0158271 A1 uses a very specific sequence of steps. The magic is in that sequence. After two steps, patterns are indistinguishable. In our current products, we keep going. After seven steps, there is nothing left for the hacker.