I have now uploaded MakePlaylist version 1.100 to SourceForge:
https://sourceforge.net/projects/makeplaylist/
This is the version I have used to demonstrate that MakePlaylist is what I call a Lossless Information Procedure. Even the smallest increase in data or data quality is reflected in the output data quality. To do so it needs to generate millions of virtual playlists, which required considerable internal streamlining (what I've called Cyclotron mode) to be practical. It also has an option to test what I call Paradoxical Information Procedures, for which some increases in data or data quality can cause less output data quality, ie, going backwards. I discuss these types of information processing in an earlier post (which now needs a bit of updating to include the latest results obtained with 1.100 before it was finally uploaded, not to mention general editing, but the post is still basically correct):
https://verydeepleft.blogspot.com/2026/06/paradoxical-and-other-information.html
I need to get back to making other improvements to MakePlaylist for it's intended purpose of actually making playlists. But eventually I also plan to add a mode to demonstrate Truncating Information Procedures which curiously requires adding more random numbers. The Paradoxical type is achieved with simple data loss, which I think has far reaching implications. And improve on the results and show they hold up for other conditions (which to some degree I have already done but not written up yet).
No comments:
Post a Comment