Re:Long-term versus Short-term incentives
on
Making It Personal
·
· Score: 1
The problem is often that there is no real incentive to companies to protect their consumers, most consumers don't know what is done with that data, and have little way of knowing.
Furthermore, some companies exist in order to exploit such data, they don't have any credibility to destroy to start with.
My guess is that they are tring to improve huffman compression and the likes by optimizing them.
These algorithms always try to be close to the theoretical Shannon limit but that includes things like 1.25 bits allocation for a specific bit sequence - hence room for optimization.
The 100:1 ratio can be acheived by using extremely lossy comression (they mentioned DCT,FFT - which are useful in lossy comression schemes
Given the marketing hype, this may actually be what lies beneath
If we look at what they are actually saying (not much actually) it seem there might be a misunderstanding. I don't think they mean to actually compress random files.
The multitude of the word analogue and some other wording, make we want to wait for more details before saying 'no way'.
It seems that the articles passed so many hyping filters that nothing meaningful can be discerned.
An even better understandable proof (altough true only for this particular claim) - take a 1000000 bit file (~125KB) and compress it thrice - et voila - you instantly achieved a 1 bit file.
Obviously they don't seriously claim that.
Slashdot Mirror
Furthermore, some companies exist in order to exploit such data, they don't have any credibility to destroy to start with.
These algorithms always try to be close to the theoretical Shannon limit but that includes things like 1.25 bits allocation for a specific bit sequence - hence room for optimization.
The 100:1 ratio can be acheived by using extremely lossy comression (they mentioned DCT,FFT - which are useful in lossy comression schemes
Given the marketing hype, this may actually be what lies beneath
If we look at what they are actually saying (not much actually) it seem there might be a misunderstanding. I don't think they mean to actually compress random files. The multitude of the word analogue and some other wording, make we want to wait for more details before saying 'no way'.
It seems that the articles passed so many hyping filters that nothing meaningful can be discerned.
An even better understandable proof (altough true only for this particular claim) - take a 1000000 bit file (~125KB) and compress it thrice - et voila - you instantly achieved a 1 bit file. Obviously they don't seriously claim that.