Re: Is There a Privacy Risk in Google Flu Trends? - Bits Blog - NYTimes.com

[David Farber <dave () farber net>]

Begin forwarded message:

From: "David P. Reed" <dpreed () reed com>
Date: November 18, 2008 10:34:22 AM EST
To: Tony Finch <dot () dotat at>
Cc: dave () farber net
Subject: Re: [IP] Re:   Is There a Privacy Risk in Google Flu Trends?  
- Bits Blog - NYTimes.com

This is easy.  When you observe a stable correlation between a  
bacterium and a disease state in *scientific* investigations, that is  
suggestive of a "causal" connection.   But it is not proof at *all* of  
a causal connection.   That is why Koch invented the known process  
that is now the standard for proof that a particular disease is caused  
and not correlated.

"demonstrated predictive ability" is doubtful in this case.  No double- 
blind studies, nothing.  What if people got nervous about something  
that turned out to be SARS, but CDC thought from this "science" that  
it was "flu".
There's no medical science here.  Not in any usual sense.  Just the  
same kind of stuff that justifies homeopathy - I tried it and the  
symptoms went away.

Tony Finch wrote:
> "David P. Reed" <dpreed () reed com> wrote:
>
> > For example, there is no scientific validation of the core claim that
> > queries about "flu symptoms" are correlated with epidemiology.   All
> > we are presented with is a suggestive correlation, coupled with wild
> > speculation that the CDC ought to adopt it because it is supposedly
> > "better" than the currently available systems.
>
> What do you mean my "scientific validation"? Is over four years of
> correlation and a demonstrated predictive ability not enough? What's
> the difference between a suggestive correlation and a validated one?
>
> Tony.




-------------------------------------------
Archives: https://www.listbox.com/member/archive/247/=now
RSS Feed: https://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com