Bunch O Math

As part of Nemo's work on NIPL, Nemo designed NCDI, a fantastic system for generating a globally-unique identifier for a particular CD. Like all good researchers, he started by looking at other approaches people had made to the problem in the past - specifically, the efforts of the fine folks who designed CDDBp.

Now, the CDDBp DiscID has some problems, the main one being that the range of CDs in the marketplace does not generate evenly distributed keys. Trying to figure out exactly how badly the disc ID was broken was a simple task of statistics, accomplished through some seriously deranged bash scripts. Muchos thanks to GuineaPig for help on that (see FreeDB for some results on this)

The next stage was designing a replacement discid - what is, or will become, NCDI. The theory of this was fairly straightforward (simply upping the level of detail to the maximum available, while still using the basic idea that the CDDBp method used), but a proof of concept was needed. To do this some serious looking at the existing freedb database would be needed -- so serious that deranged bash scripts wouldn't cut it: this would be a job for a Perl Hack. Enter japh, an #afda regular. One subcontract later, and the official bunch-of-math script is finished, though sitting idle till Nemo runs it over the database and generates more numbers for more deranged bash scripts, just so as to determine some proof of the concept! (running the script over an almost-400meg datafile ran into out-of-memory errors. oh dear. jaaaaaappphhhhh!!!!! )

Besides all that, nothing more has progressed, and "Bunch-O'-Math!" has become a catch-cry in #afda, used at such times when the conversation lulls and nothing better can be found to say.