[Selene]

Hello, all.

This is my first “official” post here. And – yes – I’m the same Selene who posts on EaglesDisobey.net (Dan Burisch’s forum) and GrahamHancock.com message boards (Ancient mysteries, Egyptology etc.).

Speaking as someone with some experience in testing and developing “expert systems” – which is what the Half Past Human system is – it looks to me to be “unproven”.

That is: nothin’ up and running here yet, folks.

Here’s why:

• The system relies on subjectively weighted input for its parameters. That is, its numbers are based on whatever its programmers wanted them to be. That’s not necessarily a problem but you’ve got to have some solid data or statistics available to support those weightings. They determine your output.

• As a proprietary (secret) system, there's no way for an observer to test or evaluate those parameters. In other words, they could depend on casting bones or voodoo or whatever the programmers had for breakfast; we’ll never know.

• Any system based on conforming to or supporting past events/data as a developmental tool runs the risk of “curve fitting” its parameters. Simply put, twisting your system to make it “do” what the (selectively chosen in hindsight) data showed you. Curve fitting is a subtle but pervasive problem in expert systems development that they haven’t shown they’ve addressed. It usually defeats most expert system attempts.

• One “hit” – Musharaff’s resignation – doesn’t make a robust, reliable system. Any database needs at least 30 data point (hits) to even begin to provide any kind of statistical validity. 300 points would be better.(And c’mon, his resignation was buzzed about for weeks before it happened. And in politics, “nothing is official until it’s officially denied…” so it wasn’t exactly out-of-the-blue. And if the developers are claiming that “picking up on the linguistic buzz” is exactly what the system is designed to do, well, really, not this time…. The Musharaff thing wasn’t exactly a sterling example of a neat system.) Come back in a few years with some good published in advance predictions of unanticipated events, and we’ll talk, guys.

• Meanwhile, they’re charging big bucks for their newsletter. Red flag here, bigtime. This isn’t being done out of a spirit of service to humanity.

I wish the developers well here, but they’ve got a lot more work to do before the rest of us should take their prediction power too seriously. As they say themselves on their website, it’s “for entertainment only”.

So let’s lighten up here, guys.

Best regards,
Selene